A community of practice rarely fails with an announcement. There’s no closing memo, no final meeting where someone says it’s over. It just goes quiet. The discussion thread that used to get a reply within hours starts sitting for a week. The monthly call becomes bimonthly, then stops appearing on the calendar at all. Nobody decided to kill it. It simply stopped being worth anyone’s attention, one skipped post at a time.
Most organizations have started more communities of practice than they currently have running. The pattern is familiar enough that it barely registers as a failure anymore, it’s treated as the expected lifecycle of a CoP: launch with energy, coast for a while, fade. But a look at what separates the communities that keep going for years from the ones that quietly stall shows the difference isn’t luck, topic, or the enthusiasm of the founding members. It’s structural.
What the current research actually shows
APQC’s 2025 Communities of Practice in Modern Organizations survey, discussed in a webinar by community researcher Stan Garfield and APQC’s Lynda Braksiek, offers one of the more current, grounded looks at where CoPs actually break down. Two findings stand out.
First, resourcing is thinner than most leaders assume. Roughly 30 percent of organizations surveyed said they don’t explicitly provide resources or processes to support participation, and only around a quarter reward or recognize people for showing up. Communities are frequently launched with enthusiasm and then left to run entirely on the goodwill of members who are already busy with their actual job.
Second, structure matters more than most organizations give it credit for. Only about 23 percent of organizations reported using a formal, enterprise-level community model, the structure APQC’s research identifies as most effective. The rest rely on informal, functional, or mixed setups, which the research links to a familiar problem: redundant communities that never manage to connect people across the organization, each one too small and too disconnected to sustain real momentum.
The founding theory still holds up
Etienne Wenger, whose original research on communities of practice remains the foundation of the field, described three elements that hold a community together: mutual engagement, a negotiated shared purpose, and a shared repertoire, the inside language, tools, and reference points that build up as members interact over time. Communities that die quietly are usually missing one of these three, most often the first. Without regular mutual engagement, there’s no ongoing relationship for a shared purpose or shared language to form around, and the group never becomes more than a mailing list with a name.
Wenger’s framework also points to why so many corporate CoPs feel hollow even when attendance looks fine on paper: reification (turning practice into artifacts like documents, wikis, and slide templates) is easy to measure and easy to fund. Participation, the actual back-and-forth exchange between members, is harder to see and easier to neglect. Organizations tend to invest in the artifacts and starve the participation, which produces a community with a well-organized SharePoint site and no pulse.
Psychological safety is a precondition, not a nice-to-have
Research on communities of practice in agile and technical environments has found that trust, social identity, and reciprocity are consistently linked to whether members actually share knowledge inside a community, not just whether they attend. A community can have an active calendar and still fail at its actual purpose if members don’t feel safe admitting what they don’t know or challenging an established way of doing things. In practice, this is the difference between a CoP where people ask real questions and one where people only post polished updates. The second kind looks alive and produces almost nothing.
What keeps the rare ones alive
Pulling together the research and the recurring patterns across organizations that sustain CoPs for years rather than months, a few practices show up consistently:
- A single home, not several competing ones. Communities that end up duplicated across regions, business units, or platforms dilute the exact critical mass they need to feel alive. One enterprise-level community per topic, discoverable from a single directory, outperforms five smaller informal versions of the same thing.
- A leader who is an active practitioner, not just an organizer. APQC’s research on CoP leadership consistently points to passion for the topic and active participation in the practice itself as the traits that predict a leader can sustain a community, more than facilitation skill alone. Members can tell the difference between someone running a calendar invite and someone genuinely working the problem alongside them.
- Explicit time and recognition, not assumed goodwill. Where organizations provide dedicated time or formal recognition for participation, engagement holds. Where it’s left to whatever time is left over after billable or deliverable-driven work, it’s the first thing that gets dropped when someone’s calendar fills up.
- A standing cadence members can predict. Momentum survives on rhythm. A community that meets reliably, even briefly, holds its shared context better than one that meets “whenever there’s enough to discuss,” because the second version quietly trains members to stop expecting it.
- Real questions welcomed over polished updates. Communities that function as a status-report channel rarely sustain engagement, because there’s no reason to check in between updates. Communities built around genuine problem-solving give members a reason to return between meetings.
The signal leaders should actually watch
Attendance is a lagging indicator. By the time a meeting is visibly under-attended, the community has usually already been quietly dying for months. A more honest early signal is whether members are asking each other real, unresolved questions between formal sessions, in the discussion thread, the chat channel, wherever the community lives day to day. When that stops, the structure is usually still intact, but the actual community, the mutual engagement Wenger described as the core of the thing, is already gone. Everything after that is maintenance on an empty room.