See it work

Two clubs came for matchmaking. Both found the thing they needed in the Community Brain instead.

A members' club running curated dinners across several cities. A community built around shared physical challenges, whose members bond on expeditions and then scatter.

The club brought a third of its membership in for a two-week test and decided not to continue: at its size, nothing was being surfaced that its members did not already know. The second never got as far as a rollout. Once its leadership saw both layers side by side, the whole conversation moved to the one members never see.

In both, the person running the club could still name every member. That is the test, and it is a better one than any membership number. Where a leader can still do that, there is nothing left to discover. And a club that convenes often enough is already making the introductions itself, better than software would.

What both clubs were actually describing was a different problem. Not who should meet whom, but: I can no longer hold what is going on with each of them. Who is between raises. Who just moved. Who said something six months ago that matters this week. One of them put it as plainly as it can be put. We just need to know them.

That problem sits with whoever runs the club, not with its members, and it gets harder as the membership grows rather than easier. It is also the one worth solving at this size.

How to tell if this is you

• You or one other person could still list every member from memory.
• Your members see each other often enough that they are not short of contact.
• What you lose track of is not who is in the room, but what is going on with them right now.
• The people who would use a system daily are the two or three running the club, not the members.

What would break it

• Putting a member-facing tool in front of this membership. It will underperform at this size, and your members will read that as the club wasting their attention.
• Nobody whose job includes opening it. If it belongs to everyone on your team, it belongs to none of them.
• Members noticing it at all. At this size the layer that helps you should be invisible to them.
• Testing with a slice of the club. A third of a small membership is a small share of the possible pairings, and the hardest conditions you could choose.

If your club is here, the useful question is not who should meet whom. It is what your team can no longer hold in their heads.

More than nine in ten introductions that reached a member were accepted. Reach is where the work is.

A founders network running parallel local groups in different countries. A retreat organisation with cohorts stretching back several years, spread across continents. Between them, the clearest picture we have of what happens at this size.

Where members engaged, the introductions landed. About half of accepted connections produced a real conversation in the first weeks. Across the full period, counting only member-to-member activity, just under a third did.

Of the members who responded to an introduction at all, more than nine in ten accepted it. Almost nobody said no. Roughly four in ten of the people given accounts ever signed in. Eighty-seven percent of introductions expired without any response. Not declined. Never opened.

Two more patterns are worth knowing before you plan your own rollout. The in-person launch week produced ten times the activity of any other week, and around eight in ten of the members who ever became active came in around it. And the ten most engaged members generated about two thirds of everything that happened, which is normal at this stage and worth expecting rather than being alarmed by.

At this size the matching works. What decides the outcome is whether your members are somewhere you can actually reach.

How to tell if this is you

• Your members joined something ongoing, not a single event, retreat or course.
• You have an active channel of communication with your members regularly.
• There is a date already in your calendar where a good share of your members will be in one room.
• Someone's job includes the community, rather than doing it alongside everything else.
• Your cohorts are large enough or spread across countries that members cannot know everyone in them.

What would break it

• A list of people who came for an event and never agreed to more. The single largest determinant, and it is settled before anything is switched on.
• One live launch and no second one. The spike is real and it decays over roughly six weeks.
• Local cohorts too small to match within. If your groups are thirty, the value is across them, not inside each one.

When members choose to come in, engagement roughly doubles.

A large international business community, several thousand members, which opened a space around its annual gathering that members could enter if they wanted to. An invitation-only online membership of individuals across time zones, who join one at a time. In neither was anybody added to a list.

Roughly six in ten members used search, against about one in four in a comparable first week where members had been added rather than opting in. Just under six in ten accepted connections became real conversations, and not one request was declined.

In the online membership, people become active within hours of a single conversation, without a rollout or a pilot. Introductions arrive continuously, because there is no annual moment for them to cluster around.

The difference in both cases is that membership was chosen. Self-selection is doing real work in those numbers, and it is worth saying so rather than letting you read them as what a whole list would do. It is also the point at this scale: you will not activate everybody and you do not need to. A fraction of a large membership is still a great many people.

About six in ten connections in the larger community were started by a single administrator. That looks like health for roughly a month. A membership this size cannot be carried by one person. Where it is, the numbers are measuring that person's energy rather than the community's, and they fall when their attention moves.

How to tell if this is you

• Your members are dispersed enough that the useful connection is one they could not have made themselves.
• You have a way for members to opt in, rather than a list you would push to.
• Your members joined for the people, not only for what you publish.
• Growth in the membership would create value you could earn from, not only cost.

What would break it

• Uploading in bulk the entire list. Let members opt in. The ones who choose to enter are the ones who stay and engage.
• Members who joined for content rather than for people. A library and a network are different promises.
• No moment that recurs. Without a rhythm there is nothing to bring people back, and introductions land in inboxes rather than in anticipation.
• Nobody reading what members are asking for. There is no room in which to overhear it, so what members search for and do not find is simply lost.