How Communities of Practice accelerate data & AI capability

Even the well designed data literacy programmes that are delivered well, trail off as a distant memory once learning ends. Most people return to their desks, pick up where they left off, and within a few weeks the new language, the new mindset and the new habits can start to erode. Six months later, the executive sponsor asks what changed. The answer, usually, is: less than expected.

This is the reinforcement problem, and it is where most organisations underinvest. Communities of Practice are one of the most effective tools for solving it. This webinar breaks down what they actually are, how they work in practice, and what Jonlee Lockwood, Head of Data at Engie UK, learned from building one.

Where communities of practice fit in the change model

At Data & AI Literacy Academy, we use the ADKAR model as our framework for thinking about behaviour change. ADKAR maps the five stages an individual moves through: Awareness, Desire, Knowledge, Ability and Reinforcement.

The first two, Awareness and Desire, are about mindset. People need to understand that change is happening and have a personal reason to engage with it. The middle two, Knowledge and Ability, cover the actual learning: knowing how to do something is different from being able to do it under real working conditions. Those four stages are what a well-designed literacy programme delivers.

Reinforcement is the fifth stage, and it is where programmes most often fall short. ADKAR is explicit on this: without reinforcement, people drift back into old behaviours regardless of how strong the first four stages were. The change doesn’t hold. Communities of practice are one of the most practical levers available for that fifth stage, because they keep the conversation alive after the formal learning ends.

What a community of practice session actually involves

Communities of practice sessions are structured touchpoints designed to sustain momentum between learning modules and after a programme wraps. At Data & AI Literacy Academy, we think of them as the mechanism that brings a cohort back together during the down months.

In practice, a session typically includes:

• Learner demonstrations, where participants stand up and talk through what they have applied. “I used what I learned to solve this problem. It saved me four hours a week.” That kind of specificity matters far more than an abstract progress update.
• Programme highlights and module impacts, so people can see how their learning connects to wider progress.
• Awards and recognition, which Engie used particularly effectively. Monetary vouchers and public acknowledgement shift how people feel about showing up.
• Open discussion of real business problems. This is where the most valuable material tends to surface. When people feel safe to speak honestly, you find out what is actually broken. Jonlee found that community sessions regularly closed feedback loops that would otherwise have sat unresolved for months: someone would raise an access issue, and he could respond immediately that it was already being addressed.

The format is intentionally informal. That informality is the point. It creates the conditions for honest conversation that a formal project meeting rarely does.

The Engie approach: starting with champions

Engie’s communities of practice didn’t begin with a training rollout. They started with a deliberate decision to build ownership and stewardship within the business before scaling the programme.

Jonlee identified people across the organisation who were already engaged with data: people in finance doing their own Excel work, people tinkering with tools, people who asked the right questions in meetings. Data enthusiasts exist in every organisation; the work is finding them and giving them a reason to gather.

The early community wasn’t just a discussion forum. Data & AI Literacy Academy designed workshops specifically for these champions, with the goal of giving them a short-term vision, a sense of purpose and some concrete skills to bring back to their teams. The focus was on articulating value, so that when champions spoke to colleagues, they weren’t just relaying training talking points. They were describing real outcomes.

That foundation mattered when it came to leadership engagement. Engie started their first phase with a Data for Leaders programme, bringing in executives, directors, heads of function and team leaders simultaneously. When the community conversations started happening at all levels of the business, the language was already aligned. Leaders weren’t hearing unfamiliar concepts from their teams; they were hearing the same things they’d discussed in their own sessions.

Why people actually turn up

Getting people to attend community sessions when they’re already stretched is one of the most consistent challenges we see. Jonlee is candid about it: in the early days, some champions did see the commitment as a burden.

Motivation works through two channels. Extrinsic motivation includes recognition, awards, vouchers and public credit. These have a real effect, particularly in the early stages when the intrinsic rewards haven’t yet materialised. Intrinsic motivation comes later, as people start to see their own impact: “I raised that problem and it got fixed.” “The CFO used a definition I introduced.” “My team is asking better questions.”

The practical implication is that you can’t rely on goodwill alone to sustain a community. Tie attendance to OKRs and team objectives. Get line managers to signal that this isn’t side-of-desk. Make the outcomes visible and specific.

One of the patterns Engie observed was what one participant described as the drip-feeding effect: data definitions and ways of thinking were spreading to colleagues who hadn’t been part of the programme at all. That kind of cultural diffusion is hard to manufacture directly. It happens as a consequence of having engaged champions who talk about their work. When those champions are motivated and visibly recognised, the effect accelerates.

Keeping communities from becoming a dumping ground

There is a version of a community of practice that nobody wants to attend: the session where every unresolved data quality problem gets tabled, nobody has the authority to act on any of it, and the same issues surface every month. Jonlee was deliberate about avoiding this.

The community session needs a clear purpose that isn’t “bring your problems.” The workshops Data & AI Literacy Academy ran with Engie’s champions early in the programme served partly to establish that purpose: to build a shared vision, create connections across teams and define what the community was actually for.

Problem-solving does happen in these sessions, but productively. The value is in having the right people in the room: someone who knows where the data quality issue lives, and someone senior enough to commission a fix. When that combination is in place, issues that would normally sit in a ticket queue for months can be resolved in a conversation.

Towards a self-sustaining community

One of the clearest signs that a community of practice is working is when it starts to run without the person who started it. Jonlee, as Head of Data at Engie, has driven the programme with significant personal investment. But the structure he has built means he can step back. Champions are beginning to shape the next phase of the learning programme itself, contributing their knowledge of the business to customise content for different cohorts. Some are helping design the programme delivery, not just participate in it.

That kind of ownership is what makes behaviour change durable. It distributes accountability, reduces dependency on a single champion and gives participants a stake in the outcome. Jonlee put it plainly: when people are brought in to help design what they’re learning, they’re building something, not going through something.

Data & AI Literacy Academy supports this progression through a lifecycle model. Some clients need help starting from scratch, building the first community structure, recruiting the initial cohort and establishing a regular format. Others, like Engie, arrived with something already in place and needed support to deepen and sustain it. The starting point differs; the destination is the same: a community that carries the data culture forward without depending on external facilitation for every session.

Beyond communities of practice: the wider engagement picture

Communities of practice are one part of a broader approach to keeping data literacy alive inside an organisation. Data & AI Literacy Academy runs several other community formats depending on what a client needs.

Data steward and data owner communities focus on accountability: bringing the people responsible for specific data sets together to share what they’re doing, showcase improvements and create peer accountability for data quality. Governance committees, which some clients have built from scratch, embed data thinking into existing decision-making structures rather than running parallel to them.

There is also the events side: graduations that mark cohort completions, roadshows that bring the programme to different parts of the business, and data escape rooms that bring a competitive, time-pressured format to skills practice. Hackathons and data literacy assessment drives serve a similar function. The point isn’t novelty for its own sake. Variety sustains engagement over time. A community that does the same thing every month loses people. One that mixes a graduation with an escape room with a structured challenge gives people a reason to keep showing up.

Common friction points and what to do about them

Three objections come up consistently when organisations try to build or revive communities of practice.

“I don’t know what I’m supposed to do.” Champions often join without a clear brief. Ready-made toolkits, suggested talking points and micro-learning resources solve most of this. People engage more readily when they know what they’re walking into.

“I don’t have time.” This is the most common one. The answer is structural: get community participation written into objectives, secure explicit management sponsorship, and make it clear that this is valued work rather than an extra. Jonlee was explicit that Engie’s success was partly a function of exec buy-in. Without it, community participation becomes a personal commitment that people quietly drop when pressure builds.

“What’s in it for me?” The answer changes over time. In the early stages, recognition and awards carry it. Later, the intrinsic rewards take over: access, influence, visibility, the satisfaction of seeing something you shaped actually working. Surface those wins in the community itself, regularly and specifically. Named people, real outcomes, concrete numbers where you have them.

Where to start

You don’t need a perfect environment to start a community. Engie didn’t start with one either.

Identify a handful of people who are genuinely interested in data, not just those with data in their job titles. Give them a small, defined purpose. Run one session, make it useful, and build from there. If the programme is already live, look at who’s most engaged in the learning and bring those people in. Some of the most effective champions at Engie were identified by Data & AI Literacy Academy through what they were saying and doing inside the programme itself.

The communities that sustain themselves took time, iteration and the occasional reinvigoration when energy dipped. Jonlee’s advice for those moments: go back to the why. Refresh the format. Get people to share not just what they’ve done, but who they helped doing it. That shift, from reporting progress to bringing the people whose work changed, is small in format and significant in effect.

Dani Arnold is Director of Customer at Data & AI Literacy Academy. Jess Moricca is Communities and Events Manager. This post draws on a live webinar conversation with Jonlee Lockwood, Head of Data at Engie UK.

‍

‍