How to Turn a Service Into a Community

A well-run walking group meets every Tuesday, has a full register, and gets good satisfaction scores. Ask the attendees to name three other people in the group, and half of them can’t. That group is a service. It delivers a session to twelve individuals, competently, forty-two weeks a year. It is not a community, and no amount of running it well will turn it into one, because the design was never aimed at that outcome.

This distinction matters because most funders and most programme managers optimise for the service metrics — attendance, satisfaction, session count — because those are what’s easy to measure. The result is a lot of well-attended groups where the ties run through the facilitator and nowhere else. When the funding ends or the facilitator leaves, the group evaporates, because there was never a network there. There was a hub, and spokes.

The hub-and-spoke problem

Name the failure mode so you can spot it: in a hub-and-spoke group, every relationship in the room passes through one person — the facilitator, the link worker, the coordinator. Members know the coordinator well. They know each other by first name, maybe, and nothing else. Robert Putnam’s distinction between bonding and bridging social capital is useful here: a hub-and-spoke group generates a thin form of bridging capital — access to the coordinator — but almost no bonding capital between members. Remove the hub and the structure collapses instantly, because there was no lattice underneath it.

A community, by contrast, is a network where the ties don’t depend on any single node. People show up because of each other, not only because of the programme. That’s a different design target, and it requires deliberately building in things a service doesn’t need.

The sequence

1. Map the ties, not just the attendance. Before changing anything, find out who currently knows whom. A short show-of-hands exercise — “who here has spoken to another member outside of session?” — will usually reveal a hub-and-spoke pattern immediately. This is your baseline, and you should repeat the question every few months.

2. Give members a reason to arrive early or stay late. Structured time — the session itself — produces attendance. Unstructured time before and after produces relationships. Eric Klinenberg’s account of libraries and parks as social infrastructure makes this point about physical space: it’s the incidental contact around the scheduled activity, not the activity itself, that does most of the relationship-building. If your group starts the moment people sit down and ends the moment it’s over, you have engineered out the part that builds ties.

3. Rotate who leads. If the same person opens every session, sets the agenda, and closes it, that person is the only node everyone is connected to. Handing small pieces of facilitation to members — bringing the biscuits, choosing next week’s topic, leading the warm-up — distributes the connective tissue of the group away from a single point.

4. Create a reason to meet that isn’t you. A WhatsApp group members set up themselves, a shared spreadsheet, a habit of two members walking the last stretch home together — these are signs the network is starting to hold its own weight. Programmes often discourage this kind of informal organising because it’s harder to track. Encourage it anyway; it’s the thing you’re actually trying to produce.

5. Deliberately introduce people who haven’t spoken. Ray Oldenburg’s concept of the third place describes settings where regulars mix freely with newcomers, and where the mixing itself is the point, not a byproduct. Facilitators can engineer this directly — pairing rotations, seating that isn’t the same every week, a five-minute exercise that requires talking to someone you haven’t yet.

6. Track network density, not just satisfaction. Satisfaction scores tell you whether people liked the session. They tell you nothing about whether the group could survive your absence. Repeat the tie-mapping exercise from step one and watch whether the number of member-to-member connections is rising relative to member-to-coordinator connections.

What the evidence actually supports

Claim	Evidence status
Social prescribing groups improve self-esteem and confidence	Reasonably well supported across qualitative reviews, though trial evidence is thin
Structured, purposeful activity works better than contact alone	Supported by qualitative synthesis; not yet quantified
Some social prescribing reduces GP or emergency service use	Reported in a minority of included studies, not the majority
Rotating facilitation or building peer ties increases group durability	Plausible extension of social capital theory; not directly tested in the social prescribing literature
Network density is a better predictor of long-term outcomes than attendance	Consistent with survey data on isolation predictors; not tested as a programme design metric

That last row is the honest caveat. The steps above are a defensible design procedure built from adjacent evidence — social capital theory, third place theory, and what social prescribing studies say does and doesn’t work. Nobody has run a trial comparing hub-and-spoke groups against deliberately networked ones. Treat this as an engineering approach informed by good theory, not as a proven intervention.

What this means in practice: if your programme’s success measure is attendance and satisfaction, you are measuring whether you ran a good service. If you want to know whether you built a community, you need a different measure — ask members to name each other, and track whether that number grows.

There’s independent evidence for why the underlying goal is worth the extra design work. In the AARP Foundation’s national survey of adults 45 and older, only 33% of people who had spoken with their neighbours reported loneliness, against 61% of those who never had. Network breadth predicts loneliness better than most other factors measured. That is an argument for building lattices, not hubs.

What this does not solve

This procedure only works on people already in the room. It does nothing for the person who never joined the walking group, never got referred, or dropped off after one session before any tie had a chance to form — and that population is larger than the one any group will ever reach. It also assumes a programme has enough runway to let ties develop; a twelve-week pilot that gets defunded before month four will produce satisfaction data and nothing else, because bonding capital takes longer to build than most commissioning cycles allow.