Nobody Asked Her to Organise a Community of Practice, So She Did It Anyway

Most organisations have a community of practice that nobody actually wants to attend. It sits in the calendar every two weeks, three people talk, the rest keep their cameras off, and everyone quietly agrees it’s a formality. Suzanne Lagerweij has a blunt explanation for why: people don’t feel like they matter there. “Why would they go the extra mile if they go unnoticed?”

In episode 28 of Stories on Facilitating Software Architecture and Design, Suzanne tells Kenny Schwegler how she turned that observation into something concrete at a client where AI adoption had stalled out in a spreadsheet. She has a background in psychology and social work — including group homes for troubled teenagers — and she is fairly direct about the parallels with software teams. Both involve stubborn people you cannot order around, and both require you to figure out what they actually need before anything moves.

The Spreadsheet Was the Whole Strategy

When Suzanne got interested in how her client’s department was adopting AI, she found exactly one artefact: an Excel sheet where people could log their experiments. A handful of early adopters were filling it in. Nothing else existed.

“No direction, no listening to the needs of people, no support by management at all.”

This is a familiar shape. A technology arrives, a few curious engineers start poking at it in their own time, and the organisation’s response is to create a place to record activity rather than a way to make sense of it. The experiments were happening. They just weren’t connected to anything or anyone.

Suzanne had some of those early adopters in her own team. She talked to them. And then she did something that was very clearly not in her contract.

An Internal Conference Nobody Had Budgeted For

She decided to organise an internal AI conference for the department. She went to management, who said yes — but the enthusiasm was thin. Their main concern was the hours. There was a roadmap full of work, and this was time spent not delivering it.

“There was not very much encouragement. There was mostly discouragement of the whole experimentation.”

Getting speakers was harder than getting permission. People kept saying they weren’t the type to stand on a stage. Suzanne had to reframe what a session even was — not a finished product, not a polished result, but the story of what you tried and what you ran into.

She was explicit about the audiences she wanted in the room:

• The people already experimenting, so they’d be seen

• Everyone else, so they’d get curious

• Management, so they’d finally understand what kind of support was needed

That last one mattered most. Management wasn’t withholding support out of malice. They genuinely had no idea what was happening in the workspace. The conference was a way of showing them.

About 100 people came. Twenty sessions. Management arrived hesitant and left having felt the energy in the room. The second edition in June had 150 participants and 35 sessions. A third is happening in November, which Suzanne will miss — she’s already handed it over to colleagues.

The Cake Is Not a Joke

Alongside the conference, Suzanne started something smaller: a biweekly AI café. It began inside her own team about a year ago as an internal conversation about what they were trying and what was changing. After a while she persuaded the team to open the door to others.

Her team, incidentally, was the reason she’d been hired in the first place. They were the difficult team. Two years later, they were the department’s frontrunners on AI. Suzanne’s account of why is unglamorous: “I was supporting, I was listening, I was moving things. I was having my team start to see why they mattered.”

Every session has a demo, then time to talk. And there’s cake, which she bakes herself. “You need to create an environment where people want to be.”

People asked her to make it hybrid. She refused. Not because remote is bad, but because she’d seen what happens: people stay at their desk, keep the headphones on, watch the demo, and skip the part where connection actually happens. “If you want to have an actual community that works, people have to be together.”

You can disagree with that call. It’s a real constraint on who can participate. But she made it deliberately, with a clear idea of what she was optimising for, which is more than most recurring meetings can claim.

When Management Discovered They Were Part of the Problem

Suzanne’s client had a fairly traditional view of roles and responsibilities. Her behaviour confused them. She wasn’t sticking to her assignment.

“They had mixed feelings about that, because they really like that I actually challenge things and I put things on the table. They could feel that things started to move. But they also realized that there where parts where they were sometimes a part of the problem.”

The clearest example came from the AI adoption group she joined. Management wanted to measure the gains. Could they do more work? Could they do it with fewer people? Suzanne pushed a different question: where are we on the learning curve, and what is this actually like for the developers living through it?

She suggested a questionnaire and drafted one. Management then expanded it substantially, and organised a mandatory meeting where everyone would attend and fill it out.

“Well, that’s not how it works.”

A compulsory survey produces compliance, not information. Wouter Lagerweij, on the call as the next guest, called it a version of Goodhart’s law — and offered a second-hand story about a company so focused on using up its AI subscription that developers set agents running on invented tasks so they could get on with their real work. Kenny’s summary was tidier: it used to be lines of code, now it’s token spend.

What This Story Shows About Bottom-Up Change

• A logging mechanism is not a strategy. The spreadsheet gave management the comfortable feeling of having a view without any of the understanding. Recording activity and making sense of it are different jobs.

• Absence of support often means absence of information. Suzanne concluded management wasn’t blocking anything on purpose — they simply couldn’t see the work. Her intervention was to make it visible rather than to argue for it.

• The roadmap wins by default because it’s the only thing that gets discussed. Kenny made this point sharply: if management talks about the roadmap and not the community, the roadmap is what carries value. Attention is the currency.

• Intrinsic motivation still needs someone to create the conditions. Kenny asked the uncomfortable question — why do these things collapse the moment the organiser leaves? Suzanne didn’t pretend it doesn’t happen. She spent real effort finding successors, and admitted she’d waited too long trying the official route first.

• Mandatory participation destroys the thing it’s trying to measure. The questionnaire episode is a small, precise example of how a good instrument becomes useless once it’s enforced.

• Sometimes management in the room is the problem. Her advice for anyone starting a community of practice: it may be better if management isn’t part of it, because the conversation changes the moment they walk in.

The Thing She Didn’t Fix

Suzanne is leaving. The client wanted to keep her and extend the approach to other departments. She’s moving on anyway, and she’s honest about what’s unresolved: the management culture. She opened a few people’s eyes about intrinsic motivation and about looking at their teams differently. Culture moves slowly, and she knows it.

What strikes me about her story is how little of it was clever. No framework, no maturity model, no transformation programme. She asked people what they needed, gave them a stage, made the invisible visible to the people who could act on it, and baked a cake. The hard part wasn’t the design of the intervention. The hard part was doing it when it wasn’t her job, when people kept asking why she was bothering, and when the honest answer was: because if I don’t, nobody does.

So here’s the question worth sitting with. In your organisation, there is probably something like that spreadsheet — a place where activity gets recorded and nothing gets understood. Who is going to make it visible to the people who could actually help? And if the answer is nobody, is that because it isn’t anyone’s job, or because everyone is waiting for permission?

Further Reading

Mentioned in this episode

• The Psychology of Software Teams by Cat Hicks — Suzanne recommended it for anyone who wants to understand the group dynamics behind team change, which she sees as the core skill in this kind of work.

• Goodhart’s law — raised by Wouter Lagerweij to describe what happens when AI usage or productivity becomes the target rather than the indicator.

Worth exploring

• Cultivating Communities of Practice by Etienne Wenger, Richard McDermott and William Snyder — the foundational work on why communities need to emerge rather than be mandated, which is precisely Suzanne’s distinction between a chapter and an AI café.

• Drive by Daniel Pink — a readable treatment of autonomy, mastery and purpose that gives language to what Suzanne means when she talks about people needing to feel they matter.

• Team Topologies by Matthew Skelton and Manuel Pais — useful for anyone whose organisation has chapters and communities of practice that exist on paper but not in practice.

• The Culture Map by Erin Meyer — helpful if you’re trying to understand why the same facilitation approach lands very differently across teams and countries.

• Turn the Ship Around! by L. David Marquet — a first-hand account of what changes when a leader stops relying on strict roles and responsibilities, which is exactly the shift Suzanne was nudging her client’s managers towards.