Weeknote 2026/45

Quad bikes drive in a line up a sand dune under a heavy grey sky. Camels and riders are silhouetted along the ridge. In the foreground, more camels in colourful woven saddle blankets rest on the sand.
Timlalin Dunes, Morocco

There are few things two digital workplace consultants enjoy more than a lengthy discussion about data visualisation. Give us eight attributes to measure, a set of scales to argue over and the opportunity to debate the merits of a box-and-whisker diagram, and we’re happy as pigs in the proverbial.

Unfortunately, on Monday we had an audience.

Somewhere in the Westfjords of Iceland, Íris Ösp Heiðrúnardóttir was listening to Jonathan and me develop our new channel fingerprinting model. I was beside a swimming pool in Morocco; Jon was at his desk in Bristol. Íris had been dispatched to take notes.

In wool.

She’s a Senior Knittaker, part of a gloriously inventive campaign from Visit Iceland in which real Icelandic knitters join your meetings and knit a summary while you get on with something more valuable. AI, in this case, stands for Authentic Icelandic.

I imagine Íris has attended some fairly tedious meetings in her time. But an hour of two consultants enthusiastically discussing the relative merits of different chart types must have tested even Icelandic reserves of patience.

We, meanwhile, were having a lovely time.

At the end of the meeting, Íris held up her finished work.

HUH?

For the first time in my professional life, I feel a meeting summary has accurately captured the experience of being in the meeting.

Bravo, Visit Iceland. Possibly my favourite marketing campaign of the year.

This week at work

HUH? wasn’t her first choice. Íris told us she’d wanted to knit ALIGN, because we’d said it so often she assumed it was the point of the meeting. She ran out of stitches.

Which is also, in my experience, what happens to most organisational alignment efforts. Everyone starts with the best intentions, then somebody runs out of patience, budget or wool.

To be fair to us, it was an alignment sort of week.

On Tuesday Jon and I presented the findings of a discovery to one client. Around the virtual table sat comms, IT, operations and senior leadership, each with a different worry.

IT wanted to know what would happen to the technical documentation. Someone who’d lived through a big reorganisation wanted a rollout that wouldn’t overwhelm teams spread across the world. Another had watched the organisation adopt hard-to-use systems before and wasn’t keen on commissioning the sequel. And the person holding the budget wanted to know how any of it would be resourced.

By the end, everyone agreed.

The next day, the client wrote to say there aren’t many meetings there that reach full alignment so quickly, and credited how “strong, convincing and practical” the assessment and recommendations were.

Lovely feedback, obviously. But there’s a point here beyond me finding an excuse to quote someone saying nice things about our work.

You can’t ask a room to align. You have to do the work that makes agreement possible before anyone walks into it.

Every hypothesis gets tested against the survey, the interviews and the analytics until it’s either validated or binned. Every recommendation has to survive the question of what it would actually mean for the people expected to implement it. And if the evidence doesn’t support your preferred answer, you need to be prepared to abandon it, however lovely the slide.

It helps when every tool you’re recommending is one the organisation already owns. That shifts the conversation away from platforms and towards the much less glamorous business of ownership.

Pages nobody maintains. Content belonging to people who left years ago. Channels that exist because somebody once set them up and nobody has since had the heart to switch them off.

A new tool without new owners will end up in exactly the same place. Just with a nicer homepage and, if you’re particularly unlucky, a three-year licence agreement.

On Wednesday, another client demonstrated why agreement sometimes means deciding not to decide.

A platform decision has been paused, sensibly, until Comms, HR and IT can agree what they want. The difficulty is that IT doesn’t yet agree with IT. One part of the department wants to stay put; another wants a proof of concept.

We came away agreeing that pausing was right, which I suppose is alignment of a kind. At least everyone now agrees that they don’t agree.

Elsewhere, the live question is who checks a request for a new online community before it goes live: the person asking for it, or the team who’ll have to tidy up after it. And at the end of the week we kicked off with a new client whose brief had already ruled out choosing any tools before we’d asked a single question.

I could have kissed them. It’s never a channels problem.

The temptation, when people disagree about technology, is to assume they need more information about the technology. Another demo, another comparison matrix, another vendor explaining how their particular collection of rectangles will transform the employee experience.

Quite often, what they need is to agree who owns what, what they’re trying to achieve and what they’re prepared to stop doing.

You can spend a great deal of money avoiding those conversations. Entire industries depend on it.

Otherwise, as Íris so elegantly put it: HUH?

Also this week

My last few days in Morocco were considerably more chilled than the first few. The first half of the trip was work: the festival, my talk, a great many conversations. The second half had no agenda at all, and for once I managed to resist the urge to give it one.

I did all the things you’d expect. Rode a camel, got pummelled in a hammam, sandboarded down some dunes, averaged about 1.3 tagines a day.

And a few I didn’t expect. Like driving through the desert in the dark, sharing a quad bike with four other people, to a bonfire party in the dunes, where we danced to Darija tunes with a wonderfully random collection of people from around the world.

Five adults. One quad bike. Night. Desert.

Somewhere, a health and safety officer woke with a start and didn’t know why.

Then home, and straight back out again: a night with my best friends to see Christopher Hall at the Comedy Cafe, with slightly more wine than is wise on a school night.

Amsterdam in autumn is a magnificent thing. The canals, the trees, the golden light on the old brick buildings. After a week of Moroccan sunshine and desert adventures, I was reminded how much I love coming home to this ridiculous, beautiful city, and to the people who make it home.

I’ve also started work on a new talk for FutuHRistIC in London later this month. Having a whole book to draw from makes the process considerably easier. It’s almost as if spending several years writing down what you think about something turns out to be useful.

Who knew?

Consuming

I watched The Social Reckoning, Aaron Sorkin’s return to Facebook. I’m already familiar with the case, which may be why I found it a bit dull. It’s hard to sit on the edge of your seat for a story you already know the ending of, however briskly everyone walks down corridors delivering exposition at each other.

Its underlying point is a fair one, though. The algorithms that decide what billions of people see aren’t designed to serve those billions. They’re designed to keep us watching, clicking and reacting, because our attention is the product being sold.

Which made for an interesting contrast with the rest of my week, a good part of which I spent joyscrolling.

Hours and hours of it, watching the US midterm results come in.

There’s something deeply satisfying about watching extraordinarily wealthy arseholes discover that spending enormous sums of money doesn’t necessarily entitle them to the result they ordered.

I appreciate the irony of celebrating this while feeding the very algorithms I’ve just complained about.

Published

You can’t comms your way out of a bad deal

Three wifi-free hours on my flight home from Morocco gave me the opportunity to indulge in one of my favourite pastimes: taking a series of increasingly vicious swings at corporate bullshit.

The result was a piece about the deteriorating deal between employers and employees, and the increasingly desperate attempts of the communications profession to make everyone feel better about it. Engagement campaigns, pulse surveys and purpose statements are all very well, but they’re not going to persuade people that a shrinking pension, stagnant pay and the creeping expectation of unpaid labour constitute an exciting employee value proposition.

I had an indecent amount of fun writing it. And judging by the response, an awful lot of people enjoyed reading it. It’s now my best-read post of the year, which suggests either that I’ve struck a nerve or that the internet has a previously undiscovered appetite for watching me take a flamethrower to an employee engagement strategy.

Possibly both.

I published it on Wednesday morning, a few hours before Canva announced its restructure. Rarely has the news cycle been so obliging.

I couldn’t have timed it better if I’d been sitting in Canva’s boardroom with a copy of the redundancy plan and a stopwatch.

Coverage

Which is presumably why, when Canva’s memo landed, Liam Moran at In.Comms asked what I made of it. Happily, I’d already spent three hours on a plane sharpening my knives.

To give Canva its due, it got more right than most. The CEO put her own name to it. The severance terms were spelled out in actual numbers, sparing employees the usual treasure hunt through phrases like “generous package” and “support during this transition”. They even get to keep their laptops, which is more than some companies manage before they’ve switched off your login and had security escort you past the pot plants.

Some 1,271 people have been told they’ll lose their jobs, change roles or be relocated, without yet knowing which. Redundancy, a different job, or a new life in Sydney? That’s quite a range of possible outcomes to leave hanging over someone’s head while the leadership team finishes rearranging the boxes on its PowerPoint.

The one thing a memo like this ought to reduce is uncertainty. Canva has instead distributed it to more than a fifth of its workforce, with a helpful note about the company’s exciting future.

And the real damage may be among everyone who isn’t on the list. The people watching their colleagues spend the next few weeks wondering whether they’ll still have a job, a manager or a postcode. They’ll keep turning up, doing their work and collecting their salaries. But they’ll also be recalculating exactly how much loyalty, goodwill and unpaid effort this particular employer has earned.

No amount of empathetic messaging, carefully calibrated leadership visibility or pulse surveys with a reassuringly high response rate can undo what people have just watched happen.

The deal is the message.

Comms reacts to Canva’s restructure announcement, In.Comms

Travel

I’m home for a few days before a trip to the UK this week, then back to London the week after for FutuHRistIC.

This kind of geographical stability makes me uneasy, so I’ve dealt with it the only way I know how: by finishing the bookings for my six remaining trips between now and the end of the year.

I’ve also started planning some bigger adventures for the first half of next year, including the first part of a birthday present to myself. It will most likely end up being my 100th country.

Given my tendency to turn perfectly enjoyable hobbies into competitive endurance events with numerical targets, I suspect reaching three figures will do absolutely nothing to discourage me.

How do I keep it all organised? I’ve built myself a travel agent in Notion and Claude. It knows my preferences, pulls my itineraries together from my emails, and helps me plan routes that make logical sense.

There’s no Iceland on the list this time. But I took up knitting last year, I’ve made two scarves, and when I go back for a fifth visit, I’ve promised Íris I’ll bring one.

She may have to settle for HUH?

This week in photos

Weeknote 2026/44

Paradise Valley in full: a green river winding between palm trees and pale rock slabs, swimmers and sunbathers along the edge, café umbrellas on the bank, and dry mountains rising beyond.
Paradise Valley, Morocco

There are two ways to make something simpler. One is to understand it well enough to work out what matters. The other is to bin everything that makes it complicated and hope nobody notices.

I’ve spent much of this week attempting the first approach. Three discovery projects have reached the point where we stop collecting information and start deciding what it means. This involves taking a giant pile of evidence, much of it contradictory, and reducing it to something someone else can reasonably be expected to read, understand and act on.

Too much detail and you’ve produced an expensive filing cabinet. Too little and you’ve produced a confident-looking PowerPoint presentation with scant relationship to reality.

Outside work, I’ve been thinking about the stories we simplify, too. At a digital nomad festival in Morocco, I heard experiences that made the familiar narrative of location-independent work look embarrassingly narrow. I’ve started reading Cory Doctorow on what happens when we optimise work for machines and leave humans to deal with the consequences. And I’ve been writing about an AI industry that’s getting extremely good at making individual employees more productive without necessarily making organisations work any better.

Different examples of the same problem. When we simplify something, who decides what matters? And what happens to everything – and everyone – that doesn’t make the cut?

This week at work

At the start of every discovery project, somewhere around slide six of the kick-off deck, Jon and I show clients a diamond.

It’s the Design Council’s Double Diamond, or at least the first half of it. The shape widens as we explore the problem from every angle we can find: interviews, surveys, analytics, card sorts, the shared drive nobody has opened since 2019 but which contains the definitive version of the communications strategy. Then it narrows again, as we make sense of what we’ve found and work towards a decision.

We warn clients that the widening part will feel messy. It’s meant to. The urge to jump straight to a solution is strong, particularly when somebody has already seen a demonstration of a platform they quite fancy buying. Resisting that urge is much of the job.

What we don’t dwell on is the line down the middle.

This week, three projects reached it at once. Interviews wrapped. Platforms reviewed. Surveys and card sorts done. And the charts. My God, the charts. AI-assisted data visualisation of such exquisite filth that I briefly considered marking the report NSFW.

The widest point of the diamond is also the point where you’re holding the most information you’ll ever have. Then you have to put most of it down.

That’s what synthesis actually is. Not adding things up, but taking things away so the remainder makes sense.

Six themes become five, because two of them were the same theme wearing different hats. A category that everyone insisted was essential turns out to be needed by nobody, because something else already does the job. Another gets a new name, because the old one described the furniture rather than the reason anyone goes looking. Two slides get hidden because they say what the slide before them said, only less well.

Twenty-five people tell you, each with complete confidence, which channel is for what. Unfortunately no two of them agree.

And somewhere in there, a score out of five has halved itself on the journey from spreadsheet to slide. You catch it at the last minute and resolve, again, to stop doing maths after nine at night.

None of this shows up in the finished report. Clients see the five themes, not the sixth one that didn’t make it. They don’t see the hours spent deciding whether something is a finding, a symptom, a cause, an unresolved territorial dispute between two departments, a grudge dating back to the 2018 restructure, or just a thing somebody said very insistently in an interview.

You can put a cover page on a transcript, but it doesn’t make it insight. Although I’ve seen people try.

The right-hand side of the diamond ends in a word we put on the slide in the same size font as all the others, as if it were equally easy: decide. Next week the readouts begin, and the hard part becomes someone else’s.

Also this week

The rest of the week went the other way.

I went to my first digital nomad festival. I spend about half my life on the road, entirely by choice, so I suppose I qualify. But I’ve never quite fitted the digital nomad mould. The usual proposition is to earn a Western salary while living somewhere considerably cheaper. I maintain a home in Amsterdam and an office I use about 40% of the time, while spending half my life travelling. All the inconvenience of a nomadic lifestyle, with the economics of a particularly badly run property portfolio.

I’ve also avoided the events. Too many seem to involve digital nomads selling courses on marketing, personal branding and making money remotely to other digital nomads, who presumably intend to make their money by selling courses on marketing, personal branding and making money remotely to yet more digital nomads. It’s a business model with the structural integrity of a pyramid scheme but considerably better weather.

Arab Nomad Fest, in Taghazout, was different. Less about helping privileged graduates speed-run gentrification by taking their European salaries to low-cost beach towns. More about bringing opportunities to people who might otherwise be shut out of them.

On Tuesday I gave my talk, on working out loud and treating asynchronous work as a discipline rather than a hack. Afterwards, someone thanked me for reframing working across time zones as a strength to sell, instead of a problem to solve.

Me speaking at Arab Nomad Fest beside a slide titled “One day, three places”: me in Nagasaki, a client in Berlin and Jonathan in Bristol, with work handed between time zones. Caption: “The client never experienced a time difference.”
Me speaking at Arab Nomad Fest

Then I sat back and let other people widen the aperture.

Nasser Saqer Almohannadi has been to 218 countries. He managed this over forty years without quitting his job in Qatar’s energy sector: two big trips a year, carry-on only, printed copies of every document, and a wife who ran the logistics and has passed a hundred countries herself. I’m somewhere short of half his total, and had been feeling quite pleased with myself. He’s completed the game, and my immediate response was to start calculating how many countries a year I’d need to visit to match his score. I’ve turned someone else’s lifetime achievement into a personal performance improvement plan.

Maria Cuesta, who runs a community app for nomads, pointed out that nomads tend to cluster. They find other nomads, work alongside other nomads, and rarely get to know the places they’re actually living in. She talked about Facebook groups with 18,000 members, where the main thing people have in common is that they’ve joined the same Facebook group. That’s a flimsy foundation on which to build a community.

And then there was Arab Nomad Fest founder Muhammed Alard, who grew up in Damascus and talked about the invisible tax of a weak passport. A Syrian passport gets you into just 26 countries without a visa. Every application means weeks of paperwork. Payments don’t go through. Potential clients decide it’s easier not to bother. For most of the room, remote work is a lifestyle choice. For him, it was a means of survival.

“The internet didn’t ask for my passport,” he said. “It asked for my skills.”

Last year he returned to Syria for the first time in fifteen years, and made a film about it.

When he finished speaking, he didn’t try to sell the room a course. He asked us to hire people from places like the one he grew up in.

Consuming

I’ve started Cory Doctorow’s The Reverse Centaur’s Guide to Life After AI. A centaur is a human using a machine to extend their capabilities. A reverse centaur is a machine using a human to extend its own. Same ingredients, different employment prospects.

It’s a useful way of thinking about the increasingly fashionable promise to keep a ‘human in the loop’. This sounds reassuring, conjuring up an experienced professional exercising judgement, asking difficult questions and occasionally pressing a large red button marked STOP. In practice, the human may have neither the authority to challenge the machine, nor the expertise to understand what it’s doing, nor the time to check before the next automated decision arrives.

Congratulations. You’ve created an accountability sink with a login.

I’ve only read the opening chapters, but it’s already made me think differently about the workplace AI conversation. We spend an enormous amount of time asking what can be automated, and considerably less asking who gets to decide, whose work gets better, and who inherits the mess when the machine has finished being efficient.

Coverage

For Strategic, I wrote about why comms has a cope problem.

Within three weeks I attended two very different conferences. At Unily Unite in London, the reassurance was everywhere: on panels, in presentations, over coffee. AI might change the tools, but the fundamentals of communication would endure. We’re special smol beans, you see, and our spectacular people skills will protect us from disruption.

At UNLEASH in Paris, among 7,000 HR people, I didn’t hear anything remotely like that. When SAP said its recruiters were now doing 95% less admin, the room applauded.

Same technology. Two professions. One discussing what was already changing, the other reassuring itself that it probably wouldn’t have to.

Of course the purpose of communication doesn’t disappear because we’ve invented new tools. But confusing the enduring purpose of a profession with the permanence of its current jobs, methods and skills is a dangerous mistake. And it’s one thing to tell yourself that comforting story. It’s quite another to stand on a conference stage and encourage thousands of other professionals to believe it too.

We’ve somehow managed to turn professional development into a group therapy session for people who don’t think they need to develop.

The editor told me to let rip. I did, within the bounds of professional decency. But I seem to have kept some in reserve.

Published

Nobody buys shared. Workplace AI is being built to make individuals more productive and processes more efficient. What’s missing is the layer underneath: the shared facts, decisions and understanding that allow an organisation to function as something more than a collection of individually optimised employees, each enjoying a bespoke relationship with their own little robot.

There’s a perfectly good commercial reason for this, if by ‘good’ you mean ‘entirely predictable and deeply unhelpful’. Vendors sell to functions. Functions buy with their own budgets. Shared understanding belongs to everyone, which means nobody has a cost centre for it, a business case to fund it, or a vice-president whose bonus depends on its existence.

Meanwhile, we’re building systems that can give every employee a personalised answer without ensuring they’re all working from the same reality.

Extremely clever parts. Increasingly questionable whole.

Travel

Still in Morocco. Outside the festival, this has been my first proper experience of the country. Sleepy seaside surf towns, cats everywhere, camels, street dogs, and a quantity of tagine that must surely put me on some sort of national register.

One evening I was reading on the roof terrace of my little hostel when the owner came over and announced that I must have dinner with them. I explained that I’d already eaten. He considered this information entirely irrelevant. I’m now familiar enough with Moroccan hospitality to know that further resistance would be both futile and impolite. I was having another tagine whether I liked it or not.

I liked it.

On Saturday I went to Paradise Valley. Not quite as paradisiacal as the name suggests, but pleasant enough. I swam in the river, slid along the rocks with childlike, leisure-centre glee, and was deposited in a pool at the bottom. Ate another tagine, obviously.

Shortly after this goes out, I’ll be face down in a hammam, being pummelled by someone who presumably knows what they’re doing. If I emerge with any skin left, I’ll consider it a success.

The day after giving a talk on how to make working across time zones a competitive advantage, I sent the same meeting invitation three times because the clocks had gone back and I’d lost track of what day it was. In desperation, I asked Google what time it was in Morocco. It gave me three different dates, two different times, and a confident AI-generated explanation of a time-zone change that hadn’t happened.

Google search results giving three different dates and times for Morocco, with an AI-generated summary describing a time-zone change that never happened.

So that’s me and a trillion-dollar technology company, both defeated by the concept of Tuesday.

This week in photos

Nobody buys shared

A surrealist painting of a crowd of identical men in navy overcoats and bowler hats, seen from behind and stretching to the horizon, each sealed under his own glass bell jar. Inside each jar is a different private sky (sunny clouds, a moonlit night, rain, an orange sunset), while the real sky above them all is a flat, empty lavender.

On the first day of UNLEASH in Paris, I asked a seven-figure AI orchestration platform how many people worked for the company.

Two minutes and 36 seconds later, it told me I wasn’t authorised to know.

In the intervening 156 seconds, this marvel of enterprise technology had rummaged through Workday, interrogated a knowledge base, consulted assorted other systems and burned through enough tokens to make a small woodland nervous. All to establish that I lacked the necessary permissions to access a number published in the company’s annual report.

I was watching a machine capable of coordinating multiple autonomous agents, navigating complex workflows and integrating with virtually every enterprise system known to humankind. Yet ask it how many people work here (something I could Google in seconds) and it behaves like a particularly officious receptionist at a Soviet-era government ministry. Your request has been considered and, alas, you do not have the appropriate clearance to know how many of you there are.

I laughed (obviously). But the more I thought about it, the less funny it became.

Because the agent hadn’t simply got an answer wrong. It had exposed something fundamental about how we’re designing workplace AI.

It understood the individual. It could identify me, retrieve my records, check my permissions and help me complete a task. It understood processes, too: the workflows, approvals and transactions that keep an organisation functioning.

What it seemingly couldn’t recognise was a fact about the organisation itself.

It had a model of me. It had a model of the machinery. It had no model of us.

After several days wandering around UNLEASH, watching vendors demonstrate increasingly sophisticated ways to optimise individual employees and automate the processes connecting them, I began to wonder whether this wasn’t the defining blind spot of the industry.

We’re building extremely clever systems for managing the parts. But who’s making sure the whole still works?

Two layers, both sensible

Walk the exhibition floor at an HR technology conference in the Year of our Lord 2026 and you’ll find an industry smearing AI all over two parts of organisational life.

The first is the individual. Copilots that draft emails, agents that book leave, skills engines that plot your next career move, platforms that personalise your working experience down to your role, location, circadian rhythm and presumably your Myers-Briggs type, if somebody can find a way to charge for it.

The promise is productivity. Save everyone a few hours a week, multiply by salary and number of seats, and, BINGO, you’ve got an ROI calculation that bangs in PowerPoint. Whether anyone actually gets to go home earlier is another question.

The second is the system. Orchestration, workflow automation, agents handing tasks to other agents, the slow extinction of the ticket queue. Nobody will mourn filling in a form to request permission to fill in another form (except possibly the person who designed the forms, who’s now Head of AI Governance).

The promise here is efficiency: faster resolution, fewer handoffs, lower costs. Ker-ching.

I’m not dismissing either. Some of what I saw in Paris was seriously impressive, and a depressing proportion of working life consists of exactly the administrative nonsense these tools can eliminate.

But both layers have a buyer, a budget and a business case. And somewhere between the productivity gains and the efficiency savings, we’ve mislaid the organisation.

The layer nobody owns

Between the individual and the system sits a third layer: the shared one. It’s what allows thousands of people to behave as an organisation rather than as independent contractors with matching lanyards.

Part of this is factual. How many people work here? What do we actually do? What do we sell, and to whom? How do we do it? Where do I find the current policy? What are we trying to achieve this year? These are questions any reasonably curious new starter might ask in their first week. And questions any AI claiming to understand the organisation ought to be able to answer.

These aren’t facts about me, or instructions for completing a transaction. They’re facts about us.

The other part is identity. What are we trying to achieve together? Why does this organisation exist, beyond providing a convenient administrative arrangement for distributing salaries? Facts establish a common picture of reality. Identity gives people a reason to care about it.

Consider my unfortunate orchestration agent. It had access to an impressive range of systems, but apparently no understanding of what constituted common organisational knowledge. It treated a published headcount figure as though I’d asked for somebody’s confidential personnel file.

That was a design failure. But it points to a wider organisational problem: in many companies, nobody has been given the job of deciding what the organisation knows about itself, where that knowledge lives and which version is authoritative.

Headcount sits in an HR system behind permissions designed for personal records. Strategy exists as a presentation, a town hall recording and fourteen departmental interpretations, each progressively more convenient for the department concerned. Decisions disappear into email threads. Policies survive in three versions, two technically obsolete but still enjoying a devoted following.

Gallagher’s 2026 Employee Communications Report offers a useful indication of the problem. For five of the seven foundational communication assets it examined, including strategic narratives and employee value propositions, more than half of responding teams lacked an agreed, documented standard. The asset either didn’t exist or lived in someone’s head. Just 15% reported an EVP that was active and socialised.

Having a strategy and having a shared, operational understanding of it are two very different achievements.

And you can’t ground an AI agent in a story nobody has written down.

Nobody buys shared

Why is the shared layer so neglected? Not because anyone is actively opposed to shared understanding. You’d struggle to find a CEO willing to announce that organisational coherence was being discontinued as a cost-saving measure.

It’s because the economics favour everything else.

Software vendors sell seats, transactions and measurable efficiency. The individual productivity story is easy to demonstrate. The process automation story is easy to price. Both produce numbers that can be entered into a spreadsheet and presented to a finance director without requiring an extended philosophical discussion about the nature of collective endeavour.

Shared understanding is harder to package. There’s no obvious licence for everyone knowing the same thing. No straightforward metric for the value of employees understanding why the organisation has changed direction, or recognising that another team’s priorities are connected to their own.

That doesn’t mean these things have no economic value. Their value is distributed across the organisation, while purchasing decisions tend to be made within functions.

There’s a principal–agent problem lurking here: vendors are rewarded for satisfying the people who sign the contracts, whose priorities don’t necessarily match those of the employees using the technology, still less the organisation as a whole.

But workplace AI adds another twist. Even when the buyer is happy and the users are happy, the organisation as a whole may be worse off.

HR buys employee experience. IT buys orchestration. Operations buys efficiency. Communications buys channels. Everyone gets a dashboard showing their investment is working. Nobody gets a dashboard showing whether the organisation still makes sense.

We pour millions into optimising the parts and leave the whole to chance.

And personalisation introduces another problem: it makes organisational inconsistency harder to spot.

Imagine two employees asking their AI agents what the company’s most important priority is. One receives an authoritative answer about reducing costs. The other is told that growth is paramount.

Both answers draw on genuine company documents. Both are plausible. Both are fluent, confident and appropriately personalised.

Neither employee knows the other has been told something different.

In the old world, they might have read the same badly written intranet page, disagreed about what it meant and had a row. That row was useful. It exposed the ambiguity, giving somebody an opportunity to resolve it.

Twenty thousand private conversations with twenty thousand AI agents don’t necessarily produce that moment of discovery.

Every employee gets a perfectly coherent version of reality. Unfortunately, they’re not all living in the same one, and nobody’s checking.

For decades, organisations have tried to break down silos. Now we’re building a bespoke, AI-powered silo around every employee, and congratulating ourselves on the user experience.

The cost of being individually right

As anyone who’s ever had the misfortune to disagree with me will attest, I fucking love being right. But even I can see that twenty thousand people being individually right is no guarantee of an organisation getting anything right.

We’ve known for decades that collective performance depends on more than individual competence. In a 2010 meta-analysis of 65 studies, Leslie DeChurch and Jessica Mesmer-Magnus found that shared understanding within teams contributed to performance over and above how people behaved and how motivated they were. Being good at your job is one thing. Knowing how your job fits with everyone else’s is another.

Jody Hoffer Gittell found something similar in her research on airline operations. At American Airlines, employees understood their own tasks but had limited awareness of the wider departure process. They could do their individual jobs without necessarily understanding how those jobs fitted together. Her subsequent work on relational coordination found that shared goals, shared knowledge and mutual respect help people coordinate work across organisational boundaries.

Which is worth remembering the next time someone demonstrates an AI agent that can complete an employee’s entire workflow without requiring them to speak to another human being. Removing unnecessary handoffs is excellent; removing the understanding of how your work connects to everyone else’s is something else entirely.

You need shared understanding to coordinate. Without it, one team is cutting costs, another is investing in growth, and a third is explaining to the board that both are happening simultaneously. Everyone hits their targets and nobody can explain why the organisation is going backwards.

You need it to disagree productively, too. There’s a massive difference between arguing about what to do and spending three meetings establishing what was actually decided in the first place. The former is how organisations make progress – the latter is how you end up with a steering committee.

And there’s another dimension that productivity metrics don’t capture: belonging.

People need to understand what they’re part of, how their work connects to their colleagues’ and why any of it matters. A personalised AI agent might make my working day easier. It might even make me happier. But it doesn’t necessarily help me understand what we’re doing together.

Nobody raises a service ticket saying, ‘We appear to have lost our collective sense of purpose.’ And if they did, they’d receive an automated response confirming that their sense of purpose was important to us, followed by a link to the employee assistance programme. The individual productivity dashboard can be glowing glorious green while the organisation becomes progressively less capable of acting as one.

We know how to measure whether AI makes an employee faster. We’re much less interested in whether it makes twenty thousand employees better at being an organisation.

Building for the whole

Before we spend another seven figures teaching AI agents to think, perhaps we could establish whether the organisation has managed to think anything through itself.

This doesn’t require a new category of enterprise software (which I know comes as devastating news to several hundred people currently preparing pitch decks). It requires someone to take responsibility for what the organisation knows, what it’s decided and which version of reality employees are expected to inhabit.

Not every document, conversation and half-remembered decision. That way lies an enterprise knowledge management transformation programme, three reorganisations, a consultancy-led maturity assessment, an entirely preventable exodus of competent people, and a steering committee that eventually decides to buy SharePoint.

I mean a relatively small set of facts, decisions and priorities that ought to be consistent wherever an employee encounters them, with clear authority over who can change them.

Those facts need authoritative sources, named owners and a process for keeping them current. And if we’re expecting AI agents to answer questions about the organisation, they need to know which sources to trust. Giving an agent access to seventeen systems isn’t the same as giving it the ability to distinguish an approved policy from a PowerPoint somebody abandoned in 2023.

We also need to distinguish between information about an individual and information that belongs to the organisation.

My salary, performance review and medical records are personal. They should be protected accordingly. The company’s published headcount, strategic priorities and approved policies are shared knowledge. They need appropriate access controls too, but they shouldn’t be treated as though every question is a request to rummage through someone’s personnel file.

This is partly a technical problem of classification, permissions and retrieval. But it’s also an editorial one. Someone has to decide which version of a fact is authoritative, what it means, when it changed and how employees should understand it. AI can’t resolve a disagreement the organisation hasn’t bothered to settle.

And finally, we need to stop treating personalisation as an unqualified good.

Of course a warehouse worker and a finance director need different information. Of course a colleague in Austria needs different employment policies from someone in Australia. Nobody is arguing for the return of the homepage carousel, where the CEO’s latest thoughts on quarterly earnings compete for attention with the lunch menu.

But some things need to be shared. A strategy isn’t much use if every employee receives a bespoke interpretation that happens to flatter their department’s existing priorities. Nor is a policy if the answer changes depending on which agent you ask.

Personalisation should change how people receive the truth, not give everyone their own version of it.

This ought to be familiar territory for internal communication professionals. Establishing a common understanding of priorities, decisions and organisational identity is – or should be – a big part of the job.

Let’s not pretend that spending two decades publishing leadership announcements nobody asked for has somehow qualified us to become the guardians of organisational knowledge. Earning that role requires editorial judgement, content governance, the ability to challenge contradictory leadership messages and the willingness to tell someone that their lovingly crafted departmental strategy is incompatible with the one the organisation supposedly agreed.

It’s hard to sell a product that tells a senior executive their strategy contradicts another senior executive’s strategy. Much easier to sell them both an AI agent that will explain each strategy with total, glassy-eyed, Elizabeth Holmes conviction.

The organisation-shaped hole

We may be heading towards a world where employees rarely visit an intranet, navigate a page tree or even know which system holds the information they’re using. They’ll ask a question, an agent will find an answer, and the machinery behind it will become invisible.

Fine by me. I’ve spent twenty years watching people struggle to find things on intranets. Contrary to popular belief, I’m not emotionally attached to the experience.

But somebody still has to maintain the common knowledge that makes those answers trustworthy: resolving contradictions, keeping decisions current and making sure there’s an authoritative version of what the organisation knows.

That task doesn’t disappear because we’ve replaced the interface. If anything, it becomes more important when employees stop visiting a shared destination and start receiving answers through thousands of private conversations with AI agents.

Otherwise, we’ve just built a more sophisticated system for distributing organisational nonsense. Faster, more personalised nonsense, delivered with the confidence of a machine that has never once wondered whether it might be talking bollocks.

An organisation exists because people working together can achieve things they couldn’t achieve separately. Right now, every meat puppet is being optimised for productivity and the machinery connecting them for efficiency. Both are worthwhile.

But the value of an organisation isn’t simply the sum of its employees’ completed tasks. It’s what those people can understand, decide and achieve together.

We risk building workplaces where every employee has a brilliant AI assistant, every workflow runs beautifully, and the organisation itself has absolutely no idea what it’s doing.

Weeknote 2026/43

Me in a black fluffy coat at the top of the Eiffel Tower, with the curve of the Trocadéro and the towers of La Défense behind me under low cloud.
Up the Eiffel

Every organisation has a woman who knows where everything is. Let’s call her Anya.

Anya knows which version of the expenses form actually works, why the procurement process described on the intranet bears only a passing resemblance to the procurement process in practice, and who to call when something needs doing by Friday. She’s survived three restructures, two digital transformations and a knowledge management programme designed, in part, to make her unnecessary.

None of this is in her job description. Nobody has allocated her any time to do it, and it certainly won’t feature in her performance objectives. But she’s helpful, she’s competent, and she’s a woman, so naturally she’s become the organisation’s unofficial helpdesk, institutional memory and emotional support animal.

The place would fall apart without her. Her manager thinks she’s excellent at teamwork.

The organisation, meanwhile, has spent a small fortune trying to make Anya unnecessary. It has built knowledge bases, implemented enterprise search, launched chatbots and is now enthusiastically investigating whether an AI agent can answer the questions that currently land in her inbox. The business case is compelling: think of all the time Anya could save if people would just stop asking her things.

Except the reason people ask Anya isn’t simply that she knows where the information lives. It’s that she knows what it means, whether it’s right, and what to do when the official answer doesn’t quite fit the question. She also knows the person asking, which turns out to be extremely useful.

This week I found myself working on several versions of the same problem, across organisations with almost nothing in common except employees, information and an impressive collection of digital tools.

The question wasn’t really how to get people to stop asking Anya. It was how to make sure everyone has an Anya.

This week at work

Some weeks I do one thing. This was emphatically not one of those weeks.

On Monday I was helping a firm switch on a new community platform: posters, desk drops, content plans, senior leaders posting on day one, and the niche, nerdy joy of watching membership climb into the hundreds before lunch. NGL, I love this bit. Months of planning for the worst, hoping for the best, and then watching theory turn into practice, as people start finding answers, finding each other and, inevitably, finding the GIF button.

The best bit is what happens next. You can spend months plotting use cases, building adoption plans and modelling benefits in Excel, but give people the means to connect and they’ll invariably find uses you hadn’t imagined. Sometimes those turn out to be more valuable than anything you’d think to put in the business case.

Over the rest of the week, Jonathan and I interviewed people at a global organisation about how work and communication actually move around a distributed network of smart, engaged people. Elsewhere, we tested how researchers expect to find things on a university intranet. And on Thursday afternoon I read back our discovery findings to an international NGO from a corner of a conference centre in Paris.

Different sectors, different sizes, different tools, different problems. And yet, everywhere we looked, people kept asking for the same thing: other people.

At the university, the most common request in the open comments wasn’t a better menu or a smarter search. It was: who do I ask? At the NGO, almost nobody uses search, because colleagues are the search engine. At the conference, a European telco showing off its new intranet admitted that more than half of what people typed into the search box was, essentially, the name of someone in HR.

And at another organisation, people would rather message a colleague directly than post in a channel set up specifically for asking questions. Because asking in public feels a lot like being caught not knowing, and nobody wants four hundred colleagues to witness them asking where the expenses form lives.

If you’ve ever watched Who Wants to Be a Millionaire, you’ll recognise the dilemma.

On paper, Ask the Audience is the better lifeline. In the studio, the crowd is usually right. Wisdom of the crowd and all that. And at work, a well-functioning community can be extraordinary: someone, somewhere, knows the answer, and a good network makes it possible to find them. But putting four hundred people in an Ask channel doesn’t magically produce collective intelligence. Particularly when most have muted it, two reply ‘following’ and the only person who knows the answer left in 2022. Nor does a search box offering you a policy from 2019, three versions of the same form and a page called Useful Links containing no useful links.

So people phone a friend.

The catch with Phone a Friend is that you don’t pick your friend when the question comes up. You nominate them before the show starts. The whole lifeline depends on knowing, in advance, The Person Who Knows Things.

And there’s the problem. If you already know the right people, the system works fine. If you’re new, remote, in a different country, or simply outside the right WhatsApp group, I’m afraid you’re on your own. The organisation has a wealth of expertise, but access to it depends on whether you happen to know someone who knows someone. Less a knowledge management strategy than the Freemasons with Microsoft 365 licences.

And even if you do know Anya, there’s only one of her. She’s got an actual job, alongside her unpaid secondment as the organisation’s institutional memory, unofficial helpdesk and human workaround for three business processes nobody has bothered to document. Making her easier to find is useful. Making her personally responsible for explaining the procurement policy to eight hundred people is not a sustainable operating model, even if she’s been given a £15 Love2shop voucher in recognition of going above and beyond.

That’s the task for the digital workplace. Not to replace the friend with a page, nor to turn the friend into a permanently available human helpdesk. It’s to make expertise discoverable, answers reusable and routes to help clear enough that knowing the right person isn’t a prerequisite for getting anything done.

Sometimes the answer should be a well-written page. Sometimes it should be a conversation with someone who knows what they’re talking about. And sometimes the most useful thing the technology can do is help the first person who asks a question leave an answer that saves the next fifty people from having to ask it again.

Digital feeds the personal. It doesn’t replace it. But it ought to make the personal a little less dependent on luck, proximity and whether Anya happens to be on holiday.

Which is why, at a mock funeral for the intranet in Paris this week, I was the one at the back quietly suggesting the deceased had been misdiagnosed. We’ve been burying intranets for twenty years, and every time they turn up at their own wake, asking who’s responsible for updating the staff directory.

Also this week

Paris, for UNLEASH. On Tuesday evening I was on a panel at People Connect Club, talking about communication for people who don’t spend their working lives sitting in front of a laptop. On Wednesday I spent an hour and a half on the Flip AI stand signing copies of the book for people who, astonishingly, wanted them signed.

I’m still not entirely accustomed to the idea that people are sufficiently interested in my niche professional obsessions to read a whole book of them, let alone queue up to have me write something in the front. It remains both deeply flattering and faintly surreal.

Full disclosure: my time in Paris was a paid partnership with Flip AI. Cheers, guys. It was a blast.

The Eurostar home was late, and deposited me back in Amsterdam at 11pm, in the middle of Amsterdam Dance Event. There was a rave on the tram. I declined to participate, preferring to trundle home with my suitcase like a woman whose idea of a wild night now involves remembering to put the bins out.

My own contribution to ADE came at 7pm the following evening: a spin class, stone cold sober. Forty-six is going brilliantly, thanks.

Before I left Paris, I had lunch at Chez Fernand Christine. I go every time I’m in town, because it tastes like my Nan’s cooking.

She was Parisienne, and came to London at the start of the 1950s. By the time my mum was at school, British food was still largely beige and boiled. So mum’s friends treasured an invitation to Chez Marguerite’s, where they could eat proper cordon bleu.

While Elizabeth David was busy introducing Britain to French provincial cooking, my mum’s friends were getting the real thing round Nan’s kitchen table. And for my entire childhood, I did too.

Nan died four years ago, just shy of 100. Every time I go to Paris, I make a beeline for this restaurant, order the boeuf bourguignon, and think about her. It’s not quite her cooking, of course. But it’s close enough to bring her back for an hour. Grief’s a funny old thing.

Connections

In Paris I caught up with Liz Hartley from HSBC. It had been long enough since we last met that she’s had two babies in the meantime. I’ve had a book. Hers involved less editing but considerably more cuteness.

UNLEASH was also a chance to catch up with friends from the HR world, including finally meeting Allie Nawrat, my editor at UNLEASH, in person. I’ve been writing for them for five years, during which we’ve managed an entire professional relationship without ever occupying the same room. A pleasingly analogue development at a conference devoted to the future of work.

Me and [name?] raising espresso martinis at the Recognition Café, a lilac-walled bar on the UNLEASH show floor.

Coverage

For inComms, I joined Rachel Miller and Chris Love in answering a deceptively simple question: how do you prepare leaders for a town hall?

My answer started with what a town hall can do that nothing else in the channel mix can: put a leader in the room, speaking in their own voice, with people who can actually respond. If all they do is read out the financials, they’ve thoroughly earned the eye-roll every office worker knows. This meeting could have been an email. An expensive one, given the cost of gathering hundreds of people to hear someone read aloud.

It ended somewhere a little more uncomfortable: not every leader should automatically get a town hall. An audience isn’t a perk of the pay grade. And if you’re going to put a leader in front of people, brief them on what those people have already heard, what they’re actually worried about, and which questions everyone in the rehearsal room is desperately hoping nobody asks.

I also wrote up both days of UNLEASH on LinkedIn. Day one featured an AI agent spending two minutes and thirty-six seconds searching every connected system for a fact available on Wikipedia, before informing me I wasn’t authorised to know it. Day two featured a funeral for the intranet, whose supposedly deceased body was demonstrating 84% monthly active users an hour later.

The future of work is in excellent hands.

Published

Your frontline communication problem isn’t a frontline problem.

Around 80% of the world’s workers don’t sit at a desk, yet most internal communication is still designed for the ones who do. When it doesn’t land, we call the people on the shop floor “hard to reach”, as though they’ve deliberately hidden themselves behind the stockroom shelves to avoid the CEO’s latest thoughts on transformation.

But the problem isn’t them. It’s design: communication built around the sender’s convenience rather than the conditions of the work, from the device to the time to the shift pattern. Which is why the reflex fix, another app (possibly with badges, because apparently we’re all eight), so rarely works.

An app can put information in someone’s pocket. It cannot put five spare minutes in their shift.

Travel

This weeknote comes to you from Taghazout, Morocco. Country 95, a very normal number of countries to have visited.

I’m here for ten days, for Arab Nomad Fest, working as a digital nomad for clients back in Europe while giving a talk on how to make that work: working out loud, treating asynchronous communication as a discipline rather than a hack, and accepting that flexibility requires rather more planning than simply taking your laptop somewhere with better weather.

So if you need to phone a friend, I’m here. I’ll answer eventually.

This week in photos

Weeknote 2026/42

Golden late-afternoon light on a tree-lined canal. An arched brick bridge crowded with parked bikes leads to a red-roofed gabled house hung with yellow festival banners. Small boats are moored on water that mirrors it all.
I love this town, part 26,995

We seem to have decided that the ultimate purpose of technological progress is to eliminate the inconvenience of being alive. Every click is one too many, every pause a productivity leak, every interaction an opportunity to shave another three seconds off the crushing administrative burden of existing. Why read a book with your own eyes when a machine can extract five key takeaways, sparing you the tedious business of encountering an argument, considering it, and perhaps even changing your mind? Why have a conversation when you can have a summary? Why experience anything at all when you could receive a concise, actionable digest of what you would have experienced, had you not been so terribly busy optimising your life?

The logical endpoint is presumably a world in which machines do all our reading, writing, thinking and talking, freeing us to devote our remaining hours to watching strangers reorganise their fridges on TikTok. Somewhere, a management consultant is preparing a slide deck explaining why this represents a 40% improvement in human flourishing.

Except some of the best things about this week were spectacularly inefficient. I travelled to an actual office to meet actual people, schlepped across the Netherlands to watch three Japanese men make an unconscionable amount of noise, and spent an evening with old friends without generating a single piece of content to prove it happened. Meanwhile, at work, I found myself arguing for deliberately making a digital experience less effortless. Not because inconvenience is inherently virtuous — anyone who’s ever tried to change a password on a government website knows otherwise — but because sometimes the bit we’re so desperate to eliminate is the bit that makes the whole thing worthwhile.

Perhaps the problem isn’t that we haven’t yet managed to make everything effortless. Perhaps it’s that we’ve become so obsessed with removing the effort that we’ve stopped asking what any of it was for.

This week at work

One client asked a question this week that sounds like a policy question but is really a design one: can I post this?

The answer we landed on fits on a Post-it. If it could go in the newsletter, it can go in the community. If it couldn’t, it can’t. No flowchart, no twelve-page acceptable use policy, no appendix clarifying whether a thumbs-up emoji constitutes formal endorsement. Just a small, well-placed moment in which someone has to stop and think about who’s going to read what they’re about to publish.

That moment is friction. And friction has had a terrible press.

Anything that slows a user down is a defect to be eliminated, preferably by a product team with a reassuringly expensive job title. Which is all very well until the thing you’ve made effortless is sending the board pack to the wrong Sarah.

Hans Monderman, a Dutch traffic engineer, spent much of his career taking things away. In Drachten, he removed the traffic lights from the town’s busiest junction, which carried around 22,000 cars a day, and replaced them with a roundabout, a cycle path and somewhere for pedestrians to cross. By one account, crashes fell from thirty-six in the preceding four years to two in the following two. “The trouble with traffic engineers,” he said, “is that when there’s a problem with a road, they always try to add something.”

Which is a devastating assessment of several other professions I could name.

Monderman hadn’t made the road frictionless. He’d moved the friction out of the signage, which drivers had long since learned to obey without thinking, and into their heads, where it could do some useful work. Without a green light assuring you that the universe has granted you right of way, you have to look at the cyclist. The cyclist looks at you. Somebody slows down. Everybody has to acknowledge that other people exist, which is a radical proposition in traffic engineering and, increasingly, online.

Amsterdam (where I spent an entire week, for once), offers a working demonstration of the full range. The red tarmac of a cycle lane tells you where you are without demanding anything of you. A tram bell interrupts you just long enough to prevent you stepping off the kerb into several tonnes of public transport. A speed bump on a bike path makes you reconsider your enthusiasm. The barrier at a level crossing dispenses with persuasion altogether. Cue, pause, confirm, gate. And for most of the journey, nothing, because most of the journey should be nothing.

The skill is knowing which moment deserves which intervention.

Get it wrong in one direction and you end up with the security warning: an interruption so routine that people have trained themselves to dismiss it before they’ve finished reading. In one study of more than 25 million browser warnings, users clicked straight through 70% of Chrome’s SSL warnings and around a third of Firefox’s. Same risk, different design. A useful reminder that the presence of a warning is not the same thing as the presence of thought.

Get it wrong in the other direction and people simply go somewhere else. At another client, there’s an entirely sensible push to lock down a video tool, because AI note-takers can wander into meetings unannounced, like a wedding guest nobody quite remembers inviting but who appears to be recording the speeches. The same team is wisely leaving WhatsApp alone for now, because the people who are rarely at a desk practically live in it. You can block a tool with the best of intentions, but if you’ve made someone’s job harder without giving them a workable alternative, the work doesn’t stop. It just moves somewhere you have even less control over. Congratulations: you’ve improved compliance by making the non-compliance invisible.

So perhaps frictionless was the wrong ambition all along. The goal isn’t to remove every obstacle, or to introduce a confirmation dialogue every time someone so much as looks at a button. It’s to make the ordinary things easy and the consequential things just difficult enough that someone has to engage their brain.

Monderman put it more bluntly: treat people like idiots, and they’ll behave like idiots.

An observation with applications well beyond Dutch roundabouts.

Also this week

Last week I promised an unsettling experiment in geographical continuity. The results are in: seven consecutive nights in the same bed, and the faint suspicion that I was getting away with something. No airports, no baffling hotel lighting, no trying to remember which country’s plug sockets I was dealing with. An almost indecent level of domestic stability.

I even had a face-to-face meeting with a client. In their actual office. Here in Amsterdam. I signed in at reception, acquired a visitor badge and sat in a room with other human beings, like it was 2018. Going to client offices used to be the tedious bit of consulting, an occupational hazard to be endured in exchange for the interesting work. Now it happens so rarely that I found the whole thing rather thrilling. I came away more energised than I have from any video call this year, which is perhaps something we ought to consider before commissioning another eighteen-month study into the future of hybrid work.

Without an airport to get to, the week filled up with the things I keep meaning to do when I’m not living out of a suitcase. I went to boxing. I went to HIIT. I bought tickets for two gigs at Paradiso in March 2027, which is either the mark of a settled, organised life or a spectacular act of optimism from someone whose travel diary routinely resembles a game of Risk. Apparently there are people who make plans five months in advance and simply expect to be in the country when the date arrives. Fascinating way to live.

And at the weekend, my best friend Marie turned fifty.

There’s something rather disconcerting about watching your friends reach their half century. Partly because I am quite certain we all turned thirty approximately five years ago, and somebody needs to explain where the intervening two decades have gone. But mostly because it’s wonderful. Here we all still are, with decades of shared history, a frankly unreasonable quantity of stories and absolutely no intention of behaving like respectable middle-aged people.

It was glorious. There are almost no photos, because we were all too busy having a lovely time. Imagine that. An entire social occasion that wasn’t organised around producing evidence of itself.

Consuming

Electric Eel Shock, Japanese garage metal, at Willem Twee in Den Bosch. On a Sunday afternoon, because the live music industry has finally discovered its most lucrative demographic: people who still want their faces melted off by loud guitars but would also quite like to be in bed at a reasonable time.

The band came on at half four. Ninety minutes of filthy riffs at a volume sufficient to rearrange your internal organs, and a drummer who still performs wearing nothing but a strategically placed sock. That’s a level of commitment to the bit most of us couldn’t sustain through a single quarterly business review, never mind an entire international touring career. I can only assume the venue’s risk assessment made for interesting reading.

And then, at six o’clock, it was over. Even after schlepping home from Den Bosch, I was on the sofa with a takeaway by quarter to eight and tucked up in bed by nine. All the joy of a proper metal gig, none of the inconvenience of being awake when the pubs close. No queuing for the night bus, no standing on a platform at midnight wondering whether the last train is a philosophical concept rather than an actual service.

Getting old is fucking brilliant.

Coverage

For UNLEASH, I wrote about AI’s latest change of name, because apparently the technology hasn’t generated quite enough confusion without somebody deciding it needs rebranding. My instinct is to roll my eyes and get on with my life, but unfortunately the people responsible for enterprise technology procurement don’t always have that luxury. My argument: name the tool, not the technology. A useful name tells people what something does. A grand one mostly tells you what somebody would like you to think of them.

Jon and I also joined the Digital Workplace Group’s Digital Workplace Impact podcast to talk about designing channels people actually use. A deceptively simple ambition which, judging by the state of the average digital workplace, remains surprisingly controversial.

Published

The residue has a price.

Diane Coyle has published a paper arguing that as AI makes knowledge increasingly cheap to copy, relationships become the asset that matters. Meanwhile, the archive of bankrupt Spirit Airlines, including 100 million emails and 500 million Teams messages, has attracted bids from Google and an AI start-up. The accumulated detritus of working life is now a commodity, and a potentially rather valuable one.

Put those two developments together and you have an uncomfortable problem for everyone who spent the past decade encouraging employees to work out loud, share what they know and leave a useful record of their working relationships. We sold people the benefits of transparency without necessarily explaining who might eventually own the transparent bits.

The court hearing has since been pushed to 6 November, so this one has some way to run.

An organisation can own the archive. It can’t compel the candour that made it worth having.

Travel

I’m writing this on the train to Paris, which is certainly an interesting time to visit. The French are once again demonstrating their commitment to making political instability a participatory national sport, though I suppose nobody can accuse them of being indifferent.

It’s People Connect Club tonight, then UNLEASH World, where I’ll be signing copies of the book with Flip. If you’re there, come and say hello. I promise not to make you discuss friction, unless you start it.

This week in photos

The residue has a price

Surrealist painting of an abandoned open-plan office at dusk. On a pedestal, a glass bell jar holds a small lavender cloud made of faint speech-bubble shapes, with an orange auction tag hanging beneath it. Wisps of vapour drift towards it from the empty desks.


In August I wrote a piece for Reworked about the extraordinary afterlife of Spirit Airlines. The airline had gone bankrupt, leaving behind 100 million emails, 500 million Microsoft Teams messages, and the files, calendars and other records of its working life. Google bid $10 million for the archive. Then micro1, a startup building AI agents designed to work inside simulated companies, made a late offer of $12.5 million.

The airline was dead. Its employees had lost their jobs. But the corporate corpse still had a few organs worth harvesting.

My argument was that an archive holds the residue of work, not the work itself. And that you don’t need to believe the residue is valuable for it to change how people behave. You only need to believe that someone else does.

This week, Diane Coyle, Bennett Professor of Public Policy at Cambridge, published a working paper that arrives at roughly the same place from the opposite direction. I came at it from twenty years of watching people decide what to put in writing. She comes at it from the economics of intangible assets.

Coyle explains why relationships become more valuable as AI makes everything else easier to replicate. Spirit shows what happens when someone decides to strip-mine the evidence of them. And in a particularly elegant piece of economic self-sabotage, the prospect of extraction may make people rather less willing to leave anything worth mining.

Relationships are the moat

Coyle’s paper, The business of business is relationships, opens with a number most comms people will find faintly vindicating. By WIPO’s estimate, intangibles accounted for between 77% and 90% of the enterprise value of large companies in the top ten economies in 2024. Not factories, not stock. Stuff you can’t drop on your foot.

She proposes distinguishing between two kinds of intangible capital.

Knowledge capital is the codified, transferable stuff: software, data, brands, patents, designs. It can be owned, licensed, protected and sold. Its value often depends on being able to prevent other people from using it.

Organisational capital is harder to pin down. Management know-how, internal routines, the relationships that make things happen, and the vast body of knowledge about how the organisation actually works, as opposed to how someone in Transformation thinks it works after spending six months drawing boxes. It’s tacit, contextual and difficult to separate from the organisation that created it.

Her argument is that generative AI will put pressure on the value of codified knowledge while making organisational capital relatively more important.

When everyone has access to the same models, the same automation and the same ability to produce suspiciously familiar PowerPoint decks, competitive advantage has to come from somewhere else.

She models work as a chain of complementary tasks, where the weakest link determines what the whole system can achieve. Making one link spectacularly efficient doesn’t help much if the next one still involves persuading someone in Procurement to reply on Teams.

Who knows whom. Who can get something unstuck. Who understands why a particular process exists, even though nobody has documented the reason since 2014. Who you actually need to speak to in Finance.

These aren’t decorative features of organisational life. They’re part of how economic value gets created.

It’s the most rigorous economic case for internal communication and collaboration I’ve read in years. And it never once uses the phrase “internal communication”.

The afterlife of an airline

Coyle recognises that some organisational knowledge can be digitised and transferred, although much of its value depends on the context in which it was created. AI complicates that distinction because it can learn from digitally recorded activity, acquiring something that looks remarkably like tacit knowledge.

Spirit offers a rather grisly demonstration.

Google’s $10 million bid was selected at auction, although the sale still requires court approval. micro1’s higher late offer has not displaced Google as the selected buyer. And the approval hearing, originally scheduled for 14 October, has now been postponed until 6 November, following objections about employee privacy.

In a letter to Spirit and Google, 121 US lawmakers warned that removing names and other direct identifiers might not be enough to prevent individuals being identified from the records. Unions have also challenged the proposed transfer. Google says personal information will be excluded or de-identified by an independent third party before it receives the data.

There is a particularly revealing distinction in the proceedings. A court-appointed privacy ombudsman has recommended approval following additional safeguards for consumers. Employee privacy, however, fell outside the scope of the assessment.

The passengers have a privacy assessment. The people whose conversations helped run the airline are still fighting over what happens to theirs.

The hearing has been delayed, but the commercial proposition remains: records of an organisation’s working life have attracted substantial bids from companies interested in what AI might learn from them.

That doesn’t mean Spirit’s organisational capital was worth $12.5 million. It means a bidder believed it could extract that much value from the records. Those are different calculations.

An archive might tell you that everyone went to Jane in Finance whenever a procurement decision got complicated. It can’t necessarily tell you why. Was Jane the expert, the person with authority, the person everyone trusted, or simply the only human being who knew how to make SAP behave?

A model might learn to reproduce the pattern without understanding what made it work. Which, come to think of it, is a reasonable description of several corporate reorganisations I’ve witnessed.

This is the distinction that matters. The relationships were valuable because they enabled people to do things together. The records may be valuable because they allow a machine to simulate how those people behaved.

Relational value, recorded value and extractable value are not the same thing. But they overlap, and AI is making the boundaries between them commercially interesting.

The Enron email archive has been used by researchers for more than two decades. The idea that old workplace communications might be useful to outsiders isn’t new. What Spirit makes unusually visible is the market price attached to those communications, and the question of who gets to decide what happens to them.

Which is a problem for everyone who has spent the last decade persuading employees to put more of their working lives into the record.

The bit that goes first

In my Reworked piece I argued that when people suspect their words might become someone else’s asset, the first thing they stop writing down isn’t the decision. It’s the reasoning.

The trade-off. The reservation. The objection that got overruled. The explanation of why an apparently sensible proposal was rejected because somebody tried it three years ago and the results were catastrophic.

You end up with immaculate archives of decisions, stripped of every doubt, disagreement and inconvenient fact that might explain how anyone arrived at them. A corporate memory consisting entirely of the minutes, with none of the meeting.

Read through Coyle’s framework, this is more than a records-management problem. The reasoning is part of what allows organisational knowledge to survive beyond the people who originally developed it.

And when people stop recording it, the reasoning doesn’t necessarily disappear. It goes back into phone calls, coffees, private messages and the heads of people who happened to be in the room. Which is fine until one of them leaves, another gets promoted, and a third decides they’ve explained the same thing to six successive managers and can no longer be bothered.

What disappears is the organisation’s ability to inherit that knowledge.

Access becomes a matter of proximity again: who you know, which meetings you’re invited to, whether you happen to sit next to the person who remembers why the policy says something apparently insane.

Digital working was supposed to make that knowledge more widely available. Instead, we risk rebuilding the old hierarchy with better software.

And there’s a nasty feedback loop here. The more organisations try to extract value from the records of work, the more reasons employees have to keep the interesting parts of work off the record. The archive becomes cleaner, the organisation becomes less capable of learning from itself, and everyone congratulates themselves on their information governance.

Of course, this isn’t an argument for recording everything. Quite the opposite. A permanent, searchable record of every workplace conversation would be an extraordinary resource, provided your ambition is to create the world’s most comprehensive database of things people wish they hadn’t said.

People need room to think aloud, change their minds, disagree and occasionally say something stupid without wondering whether it will become training material for a future product.

The challenge is to preserve the knowledge organisations genuinely need without treating every interaction as something to be harvested.

That requires judgement about what gets recorded, why, for how long and for whose benefit. Not merely another exhortation to work out loud.

Measuring the wrong thing

Coyle ends with a measurement agenda, because economists, like comms professionals, have an unfortunate tendency to want numbers for things that resist being counted.

Her proposed indicators of organisational capital include organisational structures, management practices, employee ratings and investment in technology. They’re a starting point for research, not a definitive framework, and she acknowledges the difficulty of measuring something so contextual.

The trouble with proxies is that they’re wonderfully easy to count, which makes it tempting to forget what you were trying to measure in the first place.

An organogram can tell you how many layers separate the CEO from the warehouse floor. It can’t tell you whether anyone on the warehouse floor trusts a word the CEO says.

We have more sophisticated instruments in the digital workplace: collaboration patterns, organisational network analysis, employee listening. But these have their own limitations. A Teams graph can show you who talks to whom. It can’t tell you whether they’re collaborating brilliantly or engaged in a six-month argument about who owns the spreadsheet.

Interaction isn’t necessarily collaboration. Activity isn’t necessarily value. And a record of communication isn’t necessarily a record of what mattered.

There’s another complication. Measurement changes behaviour, particularly when the people being measured know that their conversations might be analysed, retained or repurposed.

If people become more guarded about what they write, the resulting data may still look comprehensive. There will be messages, meetings, documents and beautifully coloured network diagrams. But the record will increasingly reflect what people are prepared to make visible, rather than how the organisation actually works.

The instrument isn’t necessarily broken. The behaviour it’s observing has changed.

That matters because we are increasingly tempted to treat digital traces as a complete representation of organisational life. AI makes it possible to analyse them at a scale that would have been unimaginable a few years ago. It doesn’t make them a neutral or complete account of reality.

And the more commercial value we attach to those traces, the less neutral the process of producing them becomes.

Who owns the residue?

There’s another question running through Coyle’s paper that Spirit brings into focus: who benefits from organisational capital?

Among the relationships she expects to become more important are those between employers and employees, including the negotiation over how productivity gains from AI are shared.

It’s a pleasingly civilised arrangement, assuming the employees still have jobs, somebody has invited them to the negotiating table, and the table hasn’t already been sold to Google.

At Spirit, the people who produced the archive are no longer working for the airline. The proposed transaction is being negotiated through bankruptcy proceedings, with the potential proceeds forming part of the estate. The people whose conversations make up the archive don’t individually control its sale, and there’s no reason to assume they will personally benefit from the value their records might create for an AI company.

Organisational capital is collectively produced, but the rights to its digital remains are not necessarily collectively held.

This isn’t entirely new. Employers have long controlled substantial amounts of the information produced in the course of work. What has changed is the range of things that information might be used for, and therefore what someone else might be prepared to pay for it.

Spirit makes that imbalance unusually visible. And the privacy objections demonstrate that ownership of an archive does not automatically settle the question of what may lawfully or legitimately be done with everything inside it.

For anyone running internal comms or a digital workplace, this creates an awkward problem.

We’ve spent years trying to persuade people to share what they know, explain their decisions and make work visible. We did so because organisations lose an enormous amount of knowledge when everything happens in private conversations and the heads of people who eventually leave.

We weren’t wrong about that. But we built the systems on an assumption that now looks increasingly quaint: that the record of working life would remain primarily useful to the people doing the work.

So the answer cannot simply be better transparency about retention policies. Telling employees that their conversations might one day be sold to train AI models is admirably candid. It is not necessarily going to encourage them to keep talking.

We need to make deliberate choices about which knowledge deserves a lasting record, how long informal communications should survive, what secondary uses are acceptable, and what meaningful limits apply to the exploitation of workplace data.

Trust isn’t something you can retrofit with an engagement campaign after you’ve designed a system that gives people every reason to keep their mouths shut.

Just because storage is cheap and someone in IT thinks the archive might be useful to an AI model in 2031 doesn’t mean every half-formed thought deserves to be preserved for posterity.

We also need to recognise that employees have legitimate interests in what happens to the records of their working lives, even when those records sit on corporate systems.

Coyle makes a compelling economic case that relationships will become more valuable as AI commoditises knowledge. Spirit shows that the digital remains of those relationships may have a quite separate commercial value. And the people creating both are beginning to notice.

The challenge now isn’t simply to persuade employees to write more things down. It’s to give them good reasons to believe that doing so won’t eventually be used against them, or turned into somebody else’s product.

Because an organisation can own the archive. It can’t compel the candour that made it worth having.

Your frontline communication problem isn’t a frontline problem

Surrealist painting: an empty wooden office chair floats in an electric-blue sky among white clouds. On the pale sandy ground far below, a line of small uniformed workers, including a nurse, a cleaner and a delivery rider, walk past in dark-blue silhouette without looking up.

Around 80% of the world’s workers don’t sit at a desk. That’s roughly 2.7 billion people, per Emergence Capital’s much-quoted estimate. They drive, build, nurse, serve, stack, fix and deliver.

And yet almost everything we call “internal communication” was designed for the other 20%. The intranet, the all-staff email, the leadership video, the Teams post: each assumes a screen, a login, a reliable connection and a working day with a few spare minutes in it.

We then describe the majority of the workforce as “hard to reach”, as though the problem were their working conditions rather than our design choices. The workers themselves have noticed: in Microsoft’s survey of 9,600 frontline employees, 63% said messages from leadership simply don’t reach them.

There’s a particular kind of organisational optimism that believes this can be fixed with another bit of tech.

Frontline employees aren’t reading the intranet? Give them an app. They’re not opening emails? Push notifications. They’re not engaging with leadership messages? Perhaps some gamification. A leaderboard, possibly. Everyone loves a leaderboard, particularly when they’re finishing a twelve-hour shift.

The frontline communication problem isn’t really a frontline problem, it’s a design problem. Specifically, we’ve designed communication around the convenience of the sender rather than the circumstances of the person expected to receive it. And the sender, almost without exception, sits at a desk.

It’s a pattern Jonathan Phillips and I kept running into while writing Digital Communications at Work, in hospitals, retailers, charities and manufacturers alike. The sector changes. The assumption doesn’t.

The desk is doing more work than you think

Most internal communication practices carry a set of assumptions that are so deeply embedded we barely notice them.

Employees have an email address. They have a device on which they can access comms. They can check messages during the working day. They can follow a link to the intranet, answer riddles three to authenticate, sacrifice a goat to the god of multifactor authentication, and spend a few minutes reading something that mattered enormously to the person who commissioned it.

For many office workers, these assumptions are broadly reasonable. Even if they don’t always have the inclination to read the latest leadership update, they have the ability to do so.

Now consider someone driving a bus, serving customers, working on a production line or halfway through a hospital shift.

“Check the intranet” is not a neutral request. Neither is “we sent an email”.

Both make assumptions about the employee’s device, access, connectivity, attention and, crucially, who gets to decide what they do with the next five minutes of their working day.

An office worker can usually glance at a notification between meetings. A nurse cannot necessarily interrupt a medication round because Corporate Communications has published an exciting update about the organisation’s strategic priorities.

And this isn’t simply a question of access. It’s also about what people actually need to get their jobs done.

The employee on the production line may need to know that a safety procedure has changed, that tomorrow’s shift starts earlier or that a particular piece of equipment is out of service. They may be less interested in a video of the CEO discussing the organisation’s transformation journey, however handsomely produced.

That doesn’t mean they are disengaged. It could just mean they’ve correctly identified which information is useful to them.

Utility, like beauty, is in the eye of the beholder. And we’re often too happy to mistake our own priorities for everyone else’s needs.

An employee app solves access, not opportunity

The employee app market has done a good job of recognising that traditional corporate channels don’t work equally well for everyone.

Mobile access can be transformative. It can put shift information, payslips, policies, operational updates and support services into people’s hands without requiring a desk or corporate laptop. For employees who previously relied on printed notices or information passed down through managers, that’s a genuine improvement.

But there’s a difference between making communication technically accessible and making it practically usable. In the same Microsoft research, a third of frontline workers said they lacked the tools to do their jobs, rising to 41% among those without management responsibilities, and more than half had been handed new digital tools with no training at all. The access problem hasn’t been solved, it’s just been handed to the employee.

You can give a warehouse employee an app. You cannot assume they are allowed to take out their phone while operating machinery. You can send a notification to a retail worker. You cannot assume they can read it while serving a customer. You can publish a beautifully accessible article for a care worker. You cannot conjure up ten spare minutes in an already overstretched shift.

An app can put information in someone’s pocket. It cannot put five spare minutes in their shift. So an app only succeeds if it’s designed for the ninety seconds someone actually has between tasks, not the fifteen minutes we wish they had.

This distinction gets lost when organisations measure the success of frontline communication primarily through downloads, registrations and monthly active users.

As we say in the book, there is no ROI in launch. It’s all in sustained adoption and use, and use only pays off when it helps people do something. So the question isn’t how many people have the app, but whether they can find what they need, at the moment they need it, in a form they can use. That changes what you measure. How long does it take someone to find tomorrow’s shift pattern? How many “where do I find…” questions still land on supervisors? Did the safety update reach people before the shift it applied to?

A million app downloads would be a magnificent achievement… unless the million people who downloaded it still have to ask their supervisor where to find things.

Those are useful measures of whether the technology is being adopted. They tell you somewhat less about whether employees can access the information they need, at the moment they need it, in a form they can use. It’s the distinction we draw in the book between outputs and outcomes. Downloads tell you whether people have adopted the technology. Outcomes tell you whether it’s made their working lives any easier.

A million app downloads would be a magnificent achievement… unless the million people who downloaded it still have to ask their supervisor where to find things.

Stop confusing communication about work with communication that helps people work

In Digital Communications at Work, Jonathan and I distinguish between broadcast communication and communication aligned with the work people actually do. We separate news from operational and reference content, because each has a different purpose, a different audience and a different lifecycle, and treating them as one undifferentiated stream serves none of them well.

Both have a place. Employees need to understand their organisation, its direction and the decisions that affect them. But they also need information that helps them complete a task, solve a problem, make a decision or respond to something happening right now.

Corporate communication tends to be organised around the former. Frontline employees often have much more immediate needs relating to the latter.

We have reached the peculiar point where an organisation can burn electricity, consume water and summon the full might of generative AI to ensure a warehouse worker receives a hyper-personalised video of the CEO explaining the company’s purpose, while remaining apparently unable to tell that same worker whether they’re on the early shift tomorrow.

The answer isn’t to abandon organisational communication. It’s to recognise that communication which helps people do their jobs can be an important foundation for everything else. A channel that reliably solves real problems earns a place in someone’s working day. A channel that primarily creates additional demands on their attention has a harder case to make.

Means, motive and opportunity: the Fletcher Test

In Digital Communications at Work, Jonathan and I introduce the Fletcher Test, borrowing the detective-story formula of means, motive and opportunity to ask three questions about any piece of communication.

Does someone have the means to receive it? A motive to engage with it? And an opportunity to do so within the reality of their working day?

Frontline communication is where the distinction becomes particularly important. Organisations have a habit of treating the first question as though answering it automatically settles the other two.

At Unily Unite recently, we put the test to work with more than 100 communications and digital workplace professionals. We gave them a fictional organisation, a week’s worth of internal communications and employee personas ranging from lorry drivers to head-office finance staff. Their job was to judge each message from the employee’s perspective, not the sender’s.

You can download the workshop toolkit and run the exercise with your own team.

Means: can they actually receive it?

Does this person have access to the channel? Do they have the right device, permissions, connectivity and language support? Can they authenticate without navigating a sequence of systems designed for someone sitting in front of a company laptop?

And does the format work in their environment?

A long article may be appropriate for someone who spends much of their day reading documents. It’s considerably less useful to someone who needs a short operational instruction between tasks.

Means is where technology can make a real difference. But it requires an understanding of the working environment before choosing the technology.

Motive: why would they care?

This is the question communicators sometimes find uncomfortable.

Not “why does the organisation want them to know this?” but “what reason does this employee have to engage with it?”

Those are not necessarily the same thing.

An update to a safety procedure has an obvious connection to someone’s work. A change to shift arrangements may affect their pay, childcare or journey home. Information about a new employee benefit may have direct personal value.

The announcement that the organisation has launched a refreshed set of corporate values may be very important to the people who spent six months developing them. Whether it deserves three minutes of a warehouse worker’s attention is another question. Particularly if the warehouse worker wasn’t consulted about the values, doesn’t recognise them in their own experience of the organisation, and has rather more pressing concerns about next week’s rota.

And motive is shaped by more than the message itself. It’s shaped by the relationship between the employee and the organisation.

In the UK alone, 1.2 million people are on zero-hours contracts, close to the highest figure on record, and over half a million of them are under 25. Asking people in that position to demonstrate commitment to an organisation that has gone to considerable lengths to avoid committing to them is quite an ask. Particularly when the proposed solution is a push notification about company values.

This doesn’t mean every communication must offer an immediate personal benefit. Organisations have legitimate reasons to explain their strategy, culture and decisions. Employees also deserve to understand the wider business they are part of.

But attention isn’t something an organisation is automatically entitled to because it pays someone’s wages. And if the employment relationship gives people little reason to trust what they’re being told, no amount of channel optimisation will manufacture that trust.

We should be able to articulate why a particular audience needs a particular message, rather than assuming that because it matters to the sender it must matter equally to everyone else.

Opportunity: is there a realistic moment when they can engage?

This is the one I think we neglect most.

Even when someone has access to a channel and a decent reason to care, there must be an opportunity to encounter and act on the information.

That might mean building communication into shift handovers, team briefings, existing operational routines or the systems employees already use to do their jobs. It might mean providing paid time to complete essential training or read important updates. It might mean recognising that some information is best communicated by a manager in person, while other information should be available on demand.

And it might mean deciding that a message doesn’t need to be sent at all. As we put it in the book, the most powerful distribution decision is often not sending.

This is the awkward part of the exercise. Communications teams are usually asked to improve reach, increase engagement and make messages more compelling. Much less often are they given permission to question whether the communication deserves to exist in the first place.

Yet attention is a finite resource, and frontline employees have particularly little discretion over how they spend theirs.

If you’re asking someone to stop doing the job they’re paid to do so they can consume something you’ve been paid to produce, the burden of proof is very much on you.

Stop treating frontline employees as one audience

There’s another problem with the phrase “frontline workforce”: it describes a relationship to the organisation, not a coherent set of information needs. And given it covers most of the working population, it’s a remarkably unhelpful one.

A brain surgeon and the person cleaning the operating theatre are both frontline workers. So are an airline pilot bringing 300 people into Heathrow, the bloke delivering your Ocado order, the woman checking tickets on the 7.42 to Waterloo and the person trying to persuade a drunk man to leave Wetherspoons. Their working lives have almost nothing in common, beyond the fact that none of them spends the day sitting in head office being invited to webinars about employee engagement.

They differ in what they need to know, how often it changes and what happens if they don’t know it. They differ in skills, expectations and digital confidence. They differ in the devices they’re issued (or not issued), and in how much say they have over their own time.

And they differ in their relationship with the organisation, even when they work in the same building.

Take the surgeon and the cleaner. One may be a permanent employee with considerable professional autonomy, access to clinical systems and a well-established relationship with the hospital. The other may work for an outsourced facilities contractor, on a different set of systems, with different reporting lines and little access to the organisation’s internal channels. They may work on the same ward, encounter the same patients and be affected by the same operational decisions, yet belong to entirely different information ecosystems.

This is all extremely pertinent when you’re designing communication. Who is responsible for telling the cleaner that a ward has closed? Does the hospital communicate directly with them, or through the contractor? Can they access the relevant information without a hospital login? And if the answer is that their supervisor will tell them, what happens when that supervisor is off sick?

The surgeon needs the safety alert before the operating list starts. The cleaner needs to know which areas require enhanced infection-control procedures before beginning their shift. Both are operationally important. Neither is particularly well served by a generic weekly newsletter announcing that the hospital is delighted to unveil its new strategic vision.

Some frontline workers operate alone. Others are constantly surrounded by colleagues. Some work predictable shifts in one location; others move between sites or spend their days travelling. Some will download an app the moment they’re asked; others haven’t been issued an email address, let alone a login.

The opportunities to communicate are different. So are the consequences of getting it wrong.

A retail worker might receive information through a daily team briefing. A field engineer might need an update integrated into the job-management system. A driver might need something available at a depot or during a scheduled break.

Treating all of them as a single audience because they don’t sit at desks is about as useful as treating everyone who does sit at a desk as having identical needs. Nobody writes a communication strategy for “people with chairs”.

The distinction between desk-based and deskless work matters, but it is only the beginning of segmentation.

Good communication design starts with the work: what people do, where they do it, which systems they use, who they interact with, who actually employs them, how much control they have over their day and what information helps them make decisions.

Only then should we start discussing channels.

Your managers are not a free distribution network

When digital channels struggle to reach frontline employees, organisations often fall back on managers.

“Cascade this to your teams.” Corporate shorthand for making someone else responsible for the bit where communication has to work.

Managers are expected to receive corporate messages, interpret them, decide what matters, translate them into something meaningful for their teams and deliver them consistently, usually alongside the small matter of running the operation.

This assumes a great deal about the people we’ve made managers. Many got there because they were excellent at their previous jobs. The best engineer became the engineering manager. The most experienced nurse became the ward manager. The person who could keep a warehouse running through a minor apocalypse was rewarded with responsibility for twenty other people. Whether any of them could lead, coach or communicate effectively was sometimes a secondary consideration, if it was considered at all.

Some turn out to be excellent managers. Others are indifferent, inept or actively dreadful. Plenty have never received meaningful management training. Yet we construct elaborate communication cascades on the assumption that every one of them possesses the judgement, confidence and interpersonal skills to translate corporate strategy into something useful for their teams. This seems an optimistic foundation on which to build anything, let alone an organisation’s primary means of reaching its workforce.

Even the good ones are stretched. Gartner puts it bluntly: the average manager has 51% more responsibilities than they can effectively handle. And in Microsoft’s frontline survey, 69% of frontline managers said their own leaders don’t communicate with them effectively, which rather undermines the idea of the cascade before it has started. Information is expected to flow effortlessly down a management hierarchy in which the people halfway down don’t know what’s going on either.

Managers can be an exceptionally effective communication channel because they provide something an app cannot: context, interpretation and the opportunity for a conversation. A good manager can explain what a decision means locally, answer questions and spot confusion before it becomes a problem.

But that is precisely why they should not be treated as human forwarding services. In Digital Communications at Work, we treat managers as a channel with its own requirements, including briefing, timing, capability and the authority to adapt the message, rather than the default fallback when digital reach runs out.

If a message requires local interpretation, give managers the information, training, authority and time to provide it. If it simply requires employees to know a fact, consider whether making that fact directly accessible would be more reliable. And if the same information must pass through four management layers before reaching the person who needs it, perhaps the problem isn’t that managers need another briefing pack.

Perhaps it’s that you’ve built a distribution model which only works if every manager is a competent communicator, despite having done remarkably little to ensure that they are.

What a frontline communication strategy should actually do

Before commissioning another platform, I’d suggest establishing what problem you’re actually trying to solve. An unfashionable approach, admittedly, but cheaper than discovering afterwards that you’ve solved the wrong one.

What information do different groups of employees actually need to do their jobs? Which messages require immediate action, which need to be available for reference, and which are primarily intended to build understanding or connection?

Where does information currently move successfully? Where does it get stuck? Which informal routes have employees developed because the official ones don’t work?

What access do people genuinely have during paid working time? What constraints are imposed by safety, security, location, language, devices and working patterns?

And perhaps most importantly, who has the authority to decide what gets communicated, through which channels and how frequently?

These are the questions discovery is meant to answer, which is why Digital Communications at Work opens with discovery rather than with platforms. I’ve also written about how to start with discovery, including a real example of why giving frontline workers access to an intranet isn’t the same as making it useful.

Because a communication strategy that doesn’t make choices about volume isn’t really a strategy. It’s a distribution plan.

The answer may involve an employee app. It may involve digital signage, improved intranet access, operational-system integrations or something as unfashionable as a well-placed noticeboard. It may mean investing in the communication skills of managers and supervisors, giving them proper training, support and time to do the job, rather than another briefing pack and a cheery instruction to cascade.

Most likely it involves a combination, because different information needs call for different approaches.

The important thing is that the channel follows the need, rather than the organisation buying a channel and then searching for enough content to justify it.

There is a particularly expensive form of organisational self-deception in which a new communication platform becomes successful because people are persuaded to use it, rather than because it makes their working lives materially easier.

Adoption is not the same as value. A launch is the starting line, not the victory lap.

Design around the work, not around the sender

An employee can be brilliant at their job, trusted by colleagues, knowledgeable about customers and proud of what they do, while remaining blissfully unaware of whatever the CEO had to say about the organisation’s exciting transformation journey last week.

This is not an engagement crisis. It may simply mean they’ve got a job to do, and are getting on with it. Which, given that you’re paying them to do precisely that, seems a peculiar thing to hold against them.

We’ve spent years measuring whether employees engage with our communication, and less time asking whether our communication engages with their work. When the numbers disappoint, we reach for another platform, another campaign, another nudge. Anything, apparently, except the possibility that what we’re asking people to pay attention to isn’t worth the interruption.

Perhaps the problem isn’t that your frontline employees are hard to reach. Perhaps it’s that you’ve designed communication around a working day they don’t have, priorities they don’t share and time nobody has given them. Then, when they fail to respond as expected, you’ve diagnosed a problem with their engagement rather than your assumptions.

Your frontline employees aren’t hard to reach. You’ve just spent an extraordinary amount of time, money and effort trying to reach them where they aren’t.

A note on interests, and an invitation: this post is part of a paid partnership with Flip AI. I’m working with them as a creator partner around People Connect Club Paris and UNLEASH Paris next week. Flip AI reviewed this piece before it went live, but the argument is mine, including the bits about apps. I’d have written it anyway; the subject has been on my mind for years.

If you’re an in-house communications professional, come and join the argument on the Seine on 20 October, where I’ll be in conversation with Flip AI’s Luke Talbot. I’m particularly interested in hearing where you think I’ve got this wrong. Writing is how I work out what I think; finding people prepared to disagree with me is how I find out whether any of it survives contact with reality.

And if you’re at UNLEASH, come and find me at Flip AI’s booth on 21 October, where I’ll be signing copies of Digital Communications at Work from 14.00-15.30

Build for change: communication that holds when the ground moves

Surrealist painting of a gnarled old tree with its roots wrapped in hessian and twine, standing on a floating slab of earth high above a distant landscape. Soft white clouds drift across a blue sky, with hills and water far below.

Most internal communications planning operates on the charming assumption that organisations are fundamentally stable places, occasionally interrupted by change. There’s business as usual, with its newsletters, intranet stories and meticulously colour-coded content calendars, and then there’s The Crisis, during which everyone abandons the calendar and runs around looking for the emergency distribution list. Half the people on it have left, the other half haven’t opened an email all week, and someone from Legal would like to review the holding statement.

This was always a deranged way to think about organisations. Now it’s bordering on wilful denial. Instability isn’t an unfortunate interruption to business as usual. It is business as usual. Yet we continue to talk about transformation as though it’s a journey with a destination, rather than a series of wearyingly unpredictable events punctuated by steering committees. Nobody knows what’s coming next, and the reassuringly precise roadmap presented at last month’s town hall isn’t going to change that.

The job, then, is to build communication that can flex: channels people already use and trust, in a digital workplace designed to adapt as the organisation changes around it. That work happens in the good times. You can’t build trust once the crisis has started, however enthusiastically you update the CEO’s talking points.

This was the argument of my recent keynote at Dan PR in Ljubljana, organised by Marketing magazin under the bracingly honest theme A World Without Stability. I can’t give anyone stability back. Nobody can. What we can do is build communication that holds anyway.

The future arrived wearing a knitted hat

In Nihonbashi, Tokyo, I was once shown around by a tour guide who sat on my shoulder. It was a small robot in a knitted hat, from a café called DAWN, and the person guiding me was a pilot working remotely: someone living with a disability that would make a conventional job difficult or impossible.

Japan has a labour shortage. It also has people who want to work and have been shut out of work. DAWN connects those two problems. Walking around Tokyo with a robot on my shoulder should have felt futuristic. It didn’t. It felt completely ordinary, which is generally how the future turns up: not with a fanfare, but on your shoulder, wearing a hat.

Sharon O'Dea, in a blue t-shirt and jeans, walks on a Tokyo street with a small robot on her right shoulder
Being shown around Tokyo by a robot guide (photo: Sharon O’Dea)

I wrote much of Digital Communications at Work, the book I co-authored with Jonathan Phillips, in Japan. Many of the forces starting to reshape work in Europe — ageing populations, labour shortages, automation — aren’t the future there. They’re already the present. Japan is a useful corrective to the sort of futurism that involves putting 2035 on a PowerPoint slide alongside a stock photo of someone prodding a holographic interface.

And these forces aren’t arriving one at a time. Demographic change, geopolitical volatility, AI, eroding institutional trust and the fragmentation of work are accumulating and colliding. None is going to wait for the previous transformation programme to finish, or for the steering committee to approve the next phase.

Look at who counts as ‘internal’ now. Employees, yes, but also contractors, frontline workers, supply-chain partners, remote workers and people who move between organisations and projects. Access to information becomes uneven. Context becomes inconsistent. Some people feel close to the organisation; others barely feel part of it at all. Yet plenty of internal communication still assumes an employee sitting at a desk, logged into a company laptop, with enough control over their working day to read a message whenever we decide to send one.

As we put it in the book, “A communication strategy that assumes equal access will, inevitably, exclude.”

The robot in Tokyo wasn’t interesting because it was a robot. It was interesting because somebody had designed a way for a person to participate who would otherwise have been excluded. The technology was infrastructure, not the purpose. Any industry rather more interested in selling technology than asking who it actually helps might want to reflect on that.

Build something that holds

Many organisations still approach communication as a publishing operation. They have messages to distribute, channels to fill and calendars to maintain. The success measures tend to reflect that: how many things went out, how many people opened them, how many clicked. We’ve become tremendously sophisticated at measuring the production and distribution of material without really establishing whether any of it made a blind bit of difference.

But when the ground moves, nobody cares how many intranet stories you published last quarter. They care whether they can find out what’s happening, understand what it means for them and reach someone who can help. An impressive open rate tells you very little about whether someone understood the message, trusted it or could do anything useful with it. You can open an email halfway through a shift, discover you need to complete a task on a system you can’t access, and still count as a communications success story on somebody’s dashboard.

That’s an infrastructure problem, not a copywriting problem.

In Digital Communications at Work, we describe five layers of a communication ecosystem: Plan, Collaborate, Publish, Distribute and Discuss. It’s a way of seeing how information moves around an organisation, rather than looking at each channel in isolation, which is what happens when every department has its own platform, its own budget and a persuasive account manager.

Once you’ve mapped those layers, you can ask an uncomfortable question: what if one disappears tomorrow?

We worked with a human rights charity whose people operate across countries run by authoritarian regimes. They know that social networks can be switched off, sometimes along with the entire internet. For them, relying on a WhatsApp group wasn’t just a question of convenience; it was a vulnerability.

They reduced their dependence on platforms most likely to be withdrawn and made regular local backups of essential materials, so people could still access what they needed without a connection.

Most organisations aren’t operating under those conditions. But the underlying question applies to everyone. If email and Teams disappeared tomorrow morning, where would your people go? If your intranet was unavailable, how would you reach your people? And how many of your supposedly business-critical processes turn out to depend on a spreadsheet maintained by someone who’s currently on holiday?

If the answer is ‘I’m not sure’, you’ve found something worth fixing (ideally before the next incident review).

Resilience isn’t something you buy during an emergency. It’s built through the dreary work of mapping dependencies, agreeing fallbacks and testing whether they work. Nobody gets excited about a backup distribution list until the day they need one, by which point it’s a little late to discover that nobody owns it.

Holding isn’t enough

It’s relatively easy to notice when something breaks. Far harder to notice when it stops being useful. Communication also has to adapt to changes that happen slowly enough for everyone to convince themselves that nothing much is changing.

Working patterns shift. New technologies appear. Organisations merge, restructure and acquire new populations of employees. The channels that worked brilliantly five years ago become less useful, while the people responsible for them continue filling them with content because that’s what the communications plan says they should do.

This is where organisations tend to confuse the channel with the job it was supposed to perform. The intranet, the newsletter, the employee app: these are ways of meeting people’s information needs, not ends in themselves. Yet we become extraordinarily attached to them, particularly after spending eighteen months and a substantial amount of money getting them approved.

In Digital Communications at Work, we argue for separating publishing from distribution. Keep a reliable source of truth, with content that’s accurate, owned and maintained. Then allow the channels carrying that information to change as people’s needs and working habits change. The information should be dependable. The means of reaching it should be adaptable.

That matters even more as AI assistants and agents become another way people access workplace information. Your content may increasingly be consumed and interpreted by machines before it reaches a human. If your policies are duplicated, contradictory or three years out of date, AI won’t magically fix the problem. It may simply repeat the wrong answer more efficiently, with the sort of confidence previously reserved for a senior manager explaining a system they’ve never used.

Flexibility isn’t an annual technology strategy exercise. It’s a habit of asking whether the channels you have still serve the people using them, and being prepared to stop doing things that no longer earn their place. Which sounds straightforward until you discover how many apparently indispensable communications exist primarily because somebody important once asked for them.

Resilience without adaptability leaves you maintaining an increasingly obsolete system. Adaptability without resilience gets you chaos with a new logo.

You need both.

Prepare the ground before you move the tree

On the Goto Islands, I stayed at Ako House, an old Japanese house restored by a man who bought it for one reason: to save the enormous old ako tree in the garden. He didn’t buy a house and happen to get a tree. He bought the tree and took on the house to protect it.

Japanese has a word for preparing a tree to survive being moved: nemawashi. Literally, it means digging around the roots, loosening the soil and preparing the tree so it has the best chance of surviving the move.

In organisations, the same word describes the work of consulting people before a decision becomes official. Talking to the people affected, understanding their concerns and allowing their input to shape what happens next. Not the ceremonial consultation that takes place once the decision is effectively irreversible, but the kind where somebody might actually have to change their mind.

We tend to do this backwards. Leaders make a decision, communications is brought in to prepare the announcement, and the people whose working lives are about to change hear about it once everything is settled. Then everyone is terribly surprised when they resist, and comms is asked to develop an engagement campaign to address the negativity. Perhaps with a colourful graphic explaining the benefits.

The groundwork isn’t the announcement. It’s the conversations that should have happened before it.

This is where trust becomes infrastructure too. People are more likely to believe what they hear during uncertainty if they’ve had reasons to trust the organisation when things were relatively calm. That means channels that work, information they can rely on and leaders who communicate even when they don’t have a polished answer. It also means being prepared to tell people things they might not want to hear, rather than treating every difficult message as a problem to be solved with more positive language.

Three sentences do an enormous amount of work: here’s what we know, here’s what we don’t know yet, and here’s when you’ll hear more.

‘We don’t know yet’ isn’t weakness. In a period of instability, it can be one of the most reassuring things a leader says, provided they do follow through on the third sentence. It’s certainly more useful than an elaborately worded assurance that the organisation remains committed to its people while declining to explain what is actually happening to them.

The same principle applies to the communication systems themselves. Agree what people absolutely must know, how you’ll reach them and who’s responsible for making sure it happens. Do it before the emergency, because when everything is moving, there won’t be time to build the relationships, the governance or the distribution mechanisms from scratch.

And no, buying an emergency notification platform is not a substitute for people believing the messages it sends.

Our robots are not AI

Back in Tokyo, there’s one detail about the robot in the knitted hat that I haven’t mentioned. The café’s website makes a point of saying: “Our robots are not AI.”

In an industry currently determined to put AI into absolutely bloody everything, including things that were working perfectly well before, this feels positively subversive.

Everything valuable about that tour was deeply human: the planning, choosing what to show me, the conversation we had, the connection we built. The technology’s job was simply to carry a person to where they were needed.

That’s a useful way to think about workplace communication, too. Technology can carry information further, faster and in more formats than ever before. AI is making that easier still. But carrying information isn’t the same as helping people make sense of it, and producing more content isn’t the same as building trust. If it were, the average organisation would have achieved perfect mutual-understanding nirvana by now.

I can’t tell you which change is coming for your organisation. Neither can your CEO, your IT department or your favourite LinkedIn futurist, although the latter will probably sell you a framework for it. What you can do is prepare for a range of possibilities, rather than betting everything on one prediction.

Build channels people actually use. Keep the information behind them reliable. Make sure the system can survive the loss of a platform and adapt when working habits change. And establish the relationships that will carry communication when people most need to trust what they’re hearing.

None of this can be bolted on after the ground has moved. It’s built in the calm, through the deeply unglamorous groundwork that rarely makes it into a conference keynote but determines whether communication works when it matters. It’s considerably less exciting than announcing a new platform or launching a transformation programme, which is probably why organisations keep doing those instead.

The ground will keep moving. The question is whether what you’ve built can move with it.

This essay draws on my keynote at Dan PR 2026 and ideas explored in Digital Communications at Work, co-authored with Jonathan Phillips. If you’d like me to explore these ideas at your event or with your leadership team, see my speaking page.

The produce button

A surrealist painting of a vast, pale office. On a pedestal, a small elegant machine with a single orange button pours out an endless flood of paper, screens and charts that sweeps across the room and up the walls. On the far right, a lone office worker calmly types on a laptop, almost buried.
Yes I burned a tree to generate this pic in CanvaAI, the irony is not lost on me.

At Unily’s Unite conference this week I watched someone turn a set of source documents into a podcast, an infographic and a video in the space of a few minutes.

This was Content Studio, Unily’s new AI-powered tool for taking trusted company information and turning it into different formats for different audiences: podcasts, videos, infographics and more, grounded in the original sources and with humans still in control of what goes out.

It was impressive. Genuinely impressive.

And I could immediately see the use case. Take a health and safety update written centrally, turn it into a visual explainer for factory workers, translate it into the languages spoken at that site, make an audio version for people who spend their working day away from a screen. Do in minutes what would previously have required designers, translators, video editors and several increasingly desperate emails to an agency.

There was, though, a rather lovely irony to the timing.

At the very same conference, I was running a workshop called Signal Over Noise: Winning Attention in an Age of Infinite Content. We’d expected about 40 people. More than 100 turned up.

Either internal communicators have suddenly developed an unexpected passion for information theory, or having too much stuff and nobody paying attention to it has struck a nerve.

So there we were. In one room, demonstrating how astonishingly easy it has become to produce more content. In another, communicators were crammed around tables trying to work out how to stop drowning people in the stuff.

Welcome to internal communications in 2026.

Because the interesting thing about Unily’s new Content Studio isn’t whether it’s good. It is. The interesting thing is what happens when tools like it collide with a communications system that already produces far more than people can reasonably consume.

Unily’s own framing gets to the heart of it: content is abundant, impact isn’t. And now, having identified the abundance problem, we have invented a machine for making abundance virtually free.

This is what we’ve been asking for

Credit where it’s due, because there is a lot to like here. The central argument of Digital Communications at Work is that communication needs to be designed around how people actually work. Not around the org chart, and certainly not around what’s convenient for the communications team.

A frontline employee on a twelve-hour shift does not have the same information needs, available time or working context as someone sitting at a laptop in head office. Yet for years we’ve routinely given them the same 800-word intranet article and called the job done because technically everyone could access it.

Congratulations. It’s responsive.

The problem with doing this properly has always been partly practical. Adapting communication for different roles, languages, locations, channels and contexts takes time and money. Most internal comms teams simply don’t have the capacity to make twelve versions of everything.

AI changes that equation. One source can become an infographic for a factory screen, an audio summary for someone on the move, a short article for the intranet and a translated version for a local team. The underlying information remains grounded in approved source material while the presentation changes to fit the person receiving it.

That’s not gimmicky personalisation. That’s good communication design.

For years we’ve said we should meet employees where they are. Now we actually can.

Which makes the next question rather important: How often should we?

We are already drowning

The communications problem inside most organisations isn’t that employees need more information.

It’s that they are surrounded by it.

Emails. Teams messages. Intranet news. Town halls. Newsletters. Notifications. Communities. Videos. Podcasts. Viva Engage posts. PDFs attached to emails summarising PDFs attached to other emails, with Copilot in the same window offering to summarise the summary.

Then we look at falling engagement figures and ask how we can persuade people to read more. Wrong question.

Information overload isn’t a personal failing by employees who need to manage their inbox better. It’s a design outcome.

Organisations have built systems in which hundreds, sometimes thousands, of people are independently authorised to demand everybody else’s attention. Every function has something important to say. Every project needs visibility. Every leader has an update. Every initiative has a launch. Every programme has a newsletter.

Individually, almost all of these decisions are reasonable. Collectively, they amount to being trapped in a lift with the entire organisation barking its priorities at you.

And then along comes AI and removes one of the few remaining constraints: making the stuff.

Attention is a budget.

Every communication spends some of it. Every notification asks someone to stop doing one thing and pay attention to another. Every beautifully personalised two-minute video is still two minutes in which they aren’t doing the work they’re ostensibly employed to do.

Personalisation can make each individual item more relevant.

It does nothing, by itself, to reduce the number of items.

And when the cost of producing those items approaches zero, I have a nasty feeling we’re about to discover that the natural volume of organisational communication, when freed from all physical constraints, is infinite.

The problem with removing friction

The first risk is obvious: volume.

Production cost has always acted as a crude form of governance.

Making a video used to be sufficiently expensive and annoying that someone, somewhere, had to believe the subject warranted a video. You needed a camera, an editor, perhaps an agency, a budget, and the particular form of psychological collapse that comes from hearing the same thirty seconds of Intro by The XX for the twenty-fourth time while someone adjusts a crossfade.

That’s not an especially sophisticated editorial process, but it was a brake.

Remove the friction and suddenly almost everything clears the production bar. Why wouldn’t we make a podcast version? It takes three minutes. Why not an infographic too? Why not a video? Why not six localised versions, translated into eight languages and personalised for four employee groups?

Soon the organisation hasn’t solved information overload. It has achieved information overload in twelve beautifully targeted formats, each carefully optimised for the precise demographic characteristics of the person ignoring it.

And I don’t think employees will become cynical about AI-generated communication simply because a machine made it. Most people have better things to worry about.

They’ll become cynical if there’s more of it, especially if all of it starts to sound eerily similar.

The signal that somebody bothered

The second risk is trust.

Part of the signal carried by communication is that somebody bothered. The medium is the message and all that.

Someone wrote the note. Someone chose the words. Someone stood in front of a camera and said the uncomfortable thing. Someone decided this mattered enough to spend some time on it.

Effort itself communicates importance.

If the CEO records a five-minute message after a difficult restructuring, that act carries information beyond the words themselves. They showed up. They put their face to it. They had to look into a camera and say “I know today’s news will be hard for many of you” out loud, while arranging their face into the internationally recognised expression for I am not, in fact, a sociopath.

If software turns their announcement into a jaunty three-minute podcast, the information may be perfectly accurate.

But something has changed.

Communication isn’t simply information transfer. If it were, internal comms would have been solved by SharePoint in about 2003 and we’d all be doing something else now.

Who said something, why they said it, what they chose to emphasise and how much effort they were prepared to expend saying it are all part of the message.

The cheaper communication becomes to manufacture, the more valuable those signals of human intent may become.

There’s an odd possibility here that the more polished content we can generate, the more we’ll value the slightly awkward video of an actual person saying actual words into a camera without three keywords sliding smoothly into view beside their face while ambient electronica burbles underneath.

Human imperfection may yet become a premium feature.

Month six is the bit that worries me

The third risk is complacency.

I’m actually less worried about hallucination here than I am about habituation.

In month one, everyone will scrutinise the generated storyboard. They’ll check every sentence against the source. They’ll debate the emphasis. They’ll inspect the images. Someone will zoom in to 400% and discover the reassuringly diverse group of colleagues in the background includes a woman with seven fingers and a security pass apparently growing directly from her sternum.

By month six?

Yeah, you know how this works.

You’re late for another meeting. Teams is flashing. Somebody needs this published by four. You’ve generated versions of this sort of thing dozens of times before and they’ve always been fine.

You glance.

Looks alright, dunnit?

Publish.

Grounding a system in approved sources is important because it reduces the risk of invention. But a communication can contain no invented facts whatsoever and still be wrong.

It can emphasise the wrong thing. Bury the caveat. Miss the thing employees actually care about. Turn a complicated and politically sensitive decision into something technically accurate yet completely tone-deaf.

Those are judgement problems, not generation problems, and no amount of source grounding removes the need for judgement.

Governance needs to happen before the produce button

This is why I’m not convinced the most important governance question is who approves AI-generated content.

By that point, we’ve already made it.

The more useful question comes earlier: Should this communication exist at all?

In my Signal Over Noise workshop at Unite, I used what I call the Fletcher Test: Means, Opportunity, and Motive. I stole the method from the iconic Jessica Fletcher, because I did a sufficiently wishy-washy arts degree to spend three publicly funded years of university watching an enormous amount of trashy daytime television. Murder, She Wrote taught me that every murder needs three things: the means to do it, the opportunity to do it, and a motive.

(It also taught me that Cabot Cove had a murder rate that really ought to have attracted federal attention, but this is not the time, nor is it the place).

The same test turns out to be surprisingly useful for internal communications, with fewer suspiciously well-dressed corpses.

Do we have the means to reach this audience effectively?

Do they have the opportunity to pay attention to it?

And, most importantly, do they have a motive to care?

AI makes the Means bit astonishingly easy.

That’s precisely why Motive and Opportunity become more important.

If I can make you a personalised video in your preferred language, beautifully formatted for the device you’re holding, I have solved the delivery problem. But I have not magically made you care.

I’d go further. Communications teams should start thinking in terms of attention budgets for audiences. Not simply how much content a team produces, but how much attention it’s reasonable to demand from the people receiving it.

Approval shouldn’t just ask whether something is accurate, on-brand and compliant. It should ask whether this thing is worth interrupting people for.

What if we’ve got the direction wrong?

There’s another possibility I find much more interesting. Most enterprise personalisation still starts with the organisation.

We have a message.

We have twelve audiences.

We will create twelve versions and push the appropriate one towards each person.

It’s cleverer broadcasting, really. But generative AI opens up another model entirely. Imagine a housekeeper sees that the CEO has published a long update and asks:

What does this mean for me?

The system knows enough about their role and context to answer usefully. It draws from the original, approved communication. It explains what’s relevant and leaves the rest alone. If they want more detail, they ask another question. The original remains there, intact, as the source.

That’s pull rather than push, and I think that’s potentially much more transformative.

Instead of communications teams trying to anticipate every conceivable information need and manufacturing twelve versions just in case, employees interrogate trusted information when they actually need it.

That doesn’t eliminate broadcasting. Some things genuinely need to be pushed. Emergencies exist. Mandatory information exists. Leaders sometimes need to speak to everyone. But most organisational information isn’t an emergency, however enthusiastically its owner filled in the comms request form.

Pull changes the economics of personalisation because we don’t need to manufacture every possible version in advance.

And if we’re going to personalise proactively, our understanding of audiences needs to get much better too. Personas shouldn’t simply be lovingly hand-crafted descriptions of what we imagine “frontline Fiona” wants, replete with a stock photo of a suspiciously cheerful woman in a hi-vis jacket.

We have years of behavioural and engagement data telling us what different groups actually use, ignore, search for and respond to. Use it.

The scarce skill is restraint

AI is going to have us hammering the produce button with the glassy-eyed enthusiasm of a Vegas gambler who’s three hours past bedtime and halfway through the children’s inheritance.

That’s inevitable. And, mostly, it’s useful.

But the interesting consequence is that production itself becomes less valuable. If everyone can turn a policy document into a podcast, infographic, video, article and six translations before lunch, doing so is no longer much of a skill.

Knowing whether to do it is.

Which brings me back to the bigger question I came away from Unite thinking about: what exactly is the internal communications job becoming?

A significant chunk of communications work has historically been production. We write the article. Build the newsletter. Edit the video. Create the campaign. Polish the CEO post. Turn the strategy into an infographic. Then, because we’ve made the thing, we need people to consume the thing, so we spend another chunk of time trying to persuade them to click on it.

It’s an entire professional ecosystem built partly around the fact that making things used to be difficult.

Now it isn’t. Or increasingly, it won’t be.

And I don’t think the answer is to reassure ourselves that AI simply frees everyone up to do more valuable, strategic work, as if there’s a vast strategic nature reserve somewhere waiting to receive several thousand newly liberated newsletter editors.

There may well be fewer traditional communications jobs. The production layer is getting thinner. Pretending otherwise won’t make it thicker.

But the work that remains becomes more judgement-heavy, not less.

What deserves attention?

What doesn’t?

When does personalisation make something genuinely more useful, and when does it simply manufacture more noise?

When does a leader need to show up themselves?

When should an organisation speak, and when should it make good information available and trust people to find it when they need it?

And perhaps most uncomfortably: if we no longer need quite so many people to make the content, are we prepared to define the value of communications by something other than the amount of content we make?

For a long time, communications teams were constrained by what they could produce. This week I watched that constraint disappearing in real time.

Then I walked into a room designed for 40 people and found more than 100 communicators waiting to talk about how to produce less. There’s probably something in that.

The produce button is the easy bit. The difficult bit is resisting the obvious next step: pressing it over and over until the entire organisation disappears beneath a perfectly personalised slurry of podcasts, infographics and three-minute videos explaining why nobody has time to get any work done.

Weeknote 2026/38

Sharon, in a cream cardigan and jeans, smiles as she stands in the bobsleigh track. The curved concrete wall beside her is covered in graffiti in pink, purple, green and teal, and tall pines rise behind.
In Sarajevo (Photo: me)

Above Sarajevo, at the top of the Trebević cable car, there’s a bobsleigh track built for the 1984 Winter Olympics. On Tuesday, I skived off work for the afternoon to walk down it.

When it opened, it was state of the art: a triumph of Yugoslav engineering, built for the country’s first Winter Games and beamed around the world as evidence of a modern, confident state that could pull off the technically difficult business of firing people down a mountain on a tiny sled at 80 miles an hour.

And it worked. That’s the important bit.

There was nothing wrong with the track. Nobody had cheaped out on the concrete or forgotten to consult the bobsleigh stakeholder group. It did precisely what it had been designed to do, extremely well.

Eight years later, the country it had been built to showcase was tearing itself apart.

The front line ran across Trebević and the track acquired a rather different use, as an artillery position for the forces besieging Sarajevo. The curves built to shave hundredths of seconds off an Olympic run turned out to offer useful cover in a war. You can still see the bullet holes in the concrete.

Now it’s more than a kilometre of cracked track curling through the trees. Moss grows over it. Graffiti covers almost every surface, new layers appearing over old ones. People walk it, cycle it, spray-paint it, photograph it. Nobody bobsleighs.

There’s something almost indecently neat about the timeline. State-of-the-art Olympic infrastructure. Artillery position. Ruin. Tourist attraction. All in forty-odd years. Most enterprise software doesn’t manage that many changes of purpose without a rebrand and a licensing review.

The people who designed the track weren’t wrong about what it was for. They just couldn’t design for what happened next.

I’ve been thinking about that a lot this week.

We have a habit of looking at old systems and assuming their current state tells us something about the quality of the original decision. Look at this mess. Who designed that? What idiot thought this was a good idea?

Plenty of today’s messes began as perfectly sensible answers to yesterday’s problems. The organisation changed. The technology changed. People found new routes through it. New requirements arrived; old ones developed the organisational equivalent of squatters’ rights. Somebody added a workaround, then a platform to fix the workaround, then another platform to join up the platforms, and eventually everyone forgot which bit was the workaround.

Nobody sat down with a blank sheet of paper and designed the resulting horror. It accumulated. That’s the bit we tend to miss. We judge inherited systems as if they were commissioned yesterday in their current form, rather than deposited gradually, layer by layer, by years of individually defensible decisions. Organisational geology.

Eventually you end up looking at something nobody would build today and wondering which idiot designed it like that.

Quite possibly, nobody did.

The bobsleigh track certainly wasn’t badly designed. It was exceptionally well designed for a world that ceased to exist.

This week at work

We kicked off a discovery with a new client this week. Their channel mix has grown the way most do: one thing at a time, each added for a perfectly good reason, almost nothing ever taken away. No villain. No single terrible decision. Just years of reasonable people solving the problem immediately in front of them until, collectively, they’ve created something nobody would ever design from scratch.

It works, sort of. It’s just not designed for what comes next.

Discoveries are my favourite kind of work. Many projects start with an answer: build this, launch that, write the strategy. A discovery starts with a question, and you’re being paid to take it seriously. You get to look at the platforms and the data, then at how people actually find things rather than how the org chart, governance deck and person who bought the software insist they do.

You nearly always find the thing that wasn’t in the brief.

That’s usually the thing that makes the difference..

Obviously I would say bringing in an outsider is a good idea. I run a consultancy. Please apply the appropriate discount for vested interest. But hear me out.

People inside an organisation have learned the routes. They know which channel really means it, which folder contains the actual version, which search result to ignore, and who to ask when all three fail. This knowledge is hard-won, enormously useful and almost completely invisible to the people who have it. After you’ve stepped over the same loose floorboard every day for five years, you stop thinking of the floor as broken.

An outsider walks in and immediately trips over it. That’s useful intel.

We can ask the obvious question without looking daft. We have no emotional attachment to the SharePoint site somebody spent eighteen months getting approved. People tell us things they’d never say in the meeting where its owner is sitting three seats away. And because we’ve looked inside a lot of organisations, we can tell a problem peculiar to you from one that practically everyone has.

It’s usually the second, which people find oddly reassuring.

Timing matters too. The moment to look is before the next thing gets added: the new platform, the new app, the new channel promising to tidy up all the other platforms, apps and channels. This is the workplace technology equivalent of buying another storage box because the cupboard is full.

You have not solved the storage problem, you have simply purchased more storage problem.

Another big piece of work this week sits at the opposite end of the lifecycle. This organisation’s intranet is relatively new. The paths are still clean and not many people have wandered off them yet. They’ve asked for advice on structure now, before it grows, which is rarer and smarter than it sounds.

Most organisations only ask for directions once they’re lost.

Partly this is for the humans. Increasingly, though, it’s for the machines.

Everyone wants AI that can answer staff questions. Fine. But the answer will only ever be as good as the information the machine is pointed at. Clear structure. A named owner on every page. A date somebody last checked it. An agreed source of truth, rather than six documents called FINAL_v7_USE_THIS_ONE.

None of this is sexy. Nobody is putting “we know who owns the expenses policy” on the keynote stage at an AI conference (but if you really want to, you know where I am). For years we’ve called it housekeeping, which is perhaps why organisations have been able to avoid doing it with such determination.

Now it’s AI infrastructure.

People eventually learn to stay on the paved path. They know Dave’s folder is ancient, that the policy on the old intranet is only there because nobody has the permissions to delete it, and that “FINAL” is, at best, an expression of optimism.

An AI assistant has none of this institutional scar tissue.

It will cheerfully wander off-trail, find the 2019 expenses policy in a folder everyone else knows to ignore, and read it back to you with the serene confidence of something that will never personally have to explain the rejected claim.

Also this week

Bosnia: Mostar, then Sarajevo. Country 93.

I was eleven when the war began and fifteen when it ended, which means Bosnia occupies a very particular bit of my mental filing system: the evening news of my childhood. Sarajevo was one of those place names I knew long before I had any idea where it actually was. Mostar. Banja Luka. Srebrenica. Snipers. Siege. UN peacekeepers in blue helmets looking variously concerned.

Occasionally the war escaped the television and turned up at school, when a new kid with little English arrived from a place I’d previously only heard about from a man standing in front of some rubble on the Six O’Clock News.

Thirty-odd years later, I went to five museums about it in three days. This is, even by my standards, quite a lot of museum-based misery.

Three were in Mostar: the Museum of War and Genocide Victims, the War Photo Exhibition and the Mandate of Memory. Two were in Sarajevo. Much of what they showed looked exactly as I remembered it: grainy and grey, all mud and smoke and rubble, filmed from a distance or shot through a long lens. War as it arrived in a British living room in the 1990s, somewhere between John Major’s cones hotline and the nation’s apparently endless wait for Wet Wet Wet to relinquish the number-one spot.

Then, early one morning in Mostar, before the day trippers arrived, I squeezed through a gap in a fence and climbed the old sniper tower.

It wasn’t built as a sniper tower. It was a bank.

The Ljubljanska Banka building was a modern office block on what became the front line, tall enough to command a view across the city. During the war, that made it useful for something its architects presumably hadn’t included in the requirements.

Today it’s a shell: cracked concrete, exposed floors, bullet holes, broken glass. No reassuring visitor-centre arrows telling you where to stand or laminated panel explaining what you’re looking at. Getting in involves climbing through a hole in a fence, which is generally a fairly reliable indication that your travel insurer would prefer you didn’t. Mostly, these days, it’s where people with nowhere else to go end up sleeping, and using.

The whole building is covered in graffiti now. Huge, lurid, constantly changing graffiti, in colours the footage never had. At the top there was nothing between me and a perfectly clear blue sky.

I stood where the guns had been and looked down at the streets I’d watched on the news, on a beautiful sunny morning.

It was troubling in a way I still haven’t quite found the words for. Perhaps because the pictures in my head had always been grey. History had helpfully stayed there, in the grainy footage, where I’d filed it. Finding the same streets sitting under a blazing blue sky, full of people drinking coffee and going to work, made the distance between something that happened on television when I was a child and something that happened to actual people in this actual place collapse rather abruptly.

Sarajevo kept doing this to me.

The War Childhood Museum is built around objects donated by people who were children during the war, each accompanied by a few lines of their story. No grand sweep of military history. No arrows moving across maps. Just the stuff children had.

Amina was born in 1980, the same year as me. For her thirteenth birthday, in 1993, her friend Jelena made her a soft toy out of old clothes, because you couldn’t just go to a shop and buy a present. They were both too old for toys by then, Amina wrote, but she kept it somewhere she could see it. Its bright colours made the war days more beautiful.

I turned thirteen in 1993 too. I couldn’t tell you what I got.

I was holding it together until the binder clip.

Damir was younger, born in 1986. One summer he got hold of a permanent marker, which gave him a certain status among his friends. His friend Mirnes, a year older, offered to swap it for a binder clip, the way their parents swapped a bag of flour for a lamp. Before handing the marker over, Damir used it one last time to draw a smiley face on the clip.

Mirnes died a few years later.

That’s when I started crying.

There is something particularly brutal about seeing a war through the things children decided were valuable. History deals in borders, armies, treaties and casualty figures. Children deal in permanent markers, homemade toys and the important business of whether a binder clip constitutes a fair swap.

Two days later I walked past the Genocide and War Crimes Museum and kept going. It turns out even I have a limit on misery museums.

Consuming

Whenever I travel I try to read something set where I’m going. For Bosnia, on Emma Duke’s recommendation, that was Black Butterflies by Priscilla Morris.

It’s set in Sarajevo in 1992. Zora, a painter, sends her family to safety in England and stays behind, sure it’ll all be over in a few weeks.

It isn’t.

Halfway through the trip I visited Vijećnica, the old National Library, which has an exhibition about its burning and rebuilding. A few days later I reached that part of the book: the library in flames, its books drifting across the city as scraps of black paper. That’s where the title comes from.

Normally reading something set in a place gives me a better picture of it.

This time it worked the other way round. I didn’t have to imagine the room, or the building, or where it sat in the city. I’d stood there, and that hit hard.

Connections

On Tuesday I hosted an IABC EMENA session on communicating with digital nomads. Tony Stewart joined from Hong Kong, Simon Cavendish from London, and I dialled in from an apartment in Sarajevo, which felt like either excellent commitment to the premise or an elaborate tax-deductible visual aid.

The audience was a mix of people already working on the move, the nomad-curious, and communicators trying to work out what happens when someone on their team announces they’re going to spend six weeks working from another time zone.

We’d set out to make it a conversation rather than a presentation, and it was. Thanks to everyone who came and asked questions. I’ve written up what we talked about, and so has Simon.

Travel

At home this week, prepping for two talks. After that I’m heading to London for Unily Unite, then Ljubljana for Dan PR, Slovenia’s top communications conference, where I’ll be talking about building communication that holds an organisation together when the ground keeps shifting.

This week in photos