The Team Size Where Meeting Memory Stops Being One Person's Job

The founder used to sit in on everything

At four or five people, the founder is in almost every meeting that matters: sales calls, planning sessions, customer conversations, hiring debriefs. They personally remember what was promised and what was decided, and the company effectively runs on that memory. Nothing about this is designed. It works because the founder happens to be present nearly everywhere, not because anyone built a system for it.

That stops working the moment two meetings that matter happen at the same time, which is a schedule problem before it is a headcount problem. Most teams do not notice the shift as a gradual decline. They notice it as one specific event: a commitment made in a call the founder was not in, that nobody else in the room thought to write down, because writing it down had never been necessary before.

Why the breakdown follows a schedule, not a feeling

Fred Brooks gave this a formula in The Mythical Man-Month: the number of communication channels a group needs to stay coordinated grows as n(n-1)/2, where n is the number of people. The book's own example: 50 people produce 50 times 49 divided by 2, or 1,225 channels. The same formula gives 6 people 15 channels, 12 people 66 channels, and 25 people 300 channels.

The part of that formula that matters for meeting memory is what it does to any single person's coverage. Channels grow roughly with the square of headcount while headcount itself grows linearly, so the number of two-person conversations that any one individual, including the founder, is not part of grows far faster than the size of the company. That is a property of counting pairs, not a claim about anyone losing focus or getting lazier.

Small teams run an informal memory system without naming it

Psychologist Daniel Wegner described transactive memory in 1985 as a mechanism through which a group collectively encodes, stores, and retrieves knowledge, built on members knowing who knows what and being able to reach that expertise through ordinary conversation. Wegner's original research looked at couples and close-knit groups, where that shared map stays current because the people holding it talk to each other constantly, before the theory was extended to teams, larger groups, and organizations.

A small company runs exactly this kind of system without ever naming it. If you need to know what was decided last week, you ask the founder, because everyone already knows the founder was probably in the room. The founder becomes the group's transactive memory index by default, not by design.

That default only holds up as long as the channel count stays small enough for one person's presence to cover most of the conversations that matter. Once the channel count outpaces headcount, keeping the who-knows-what map current through ordinary conversation gets structurally harder, a mechanical consequence of how fast the number of pairs grows, independent of how well anyone on the team communicates.

The signal to watch is meetings the founder was not in

A specific headcount number is the wrong thing to watch for, because it depends on how many parallel meetings a given company runs and how concentrated decision-making already is. The more reliable signal is the share of substantive meetings happening without whoever has been the company's default memory holder present.

A concrete version of this: an account manager and a customer settle a renewal detail on a call the founder never joins. Three weeks later someone asks what exactly was promised, and no one wrote it down, because until recently someone would simply have remembered, and that assumption had been true often enough that nobody built a habit around checking it.

What has to change is not the org chart

The instinctive response is to add a layer, a chief of staff, an ops hire, someone whose job is to sit in on more meetings. That helps, but it does not solve the underlying math, because one more person is still one person, and the same channel formula applies to them the moment the team grows past their coverage too.

What actually needs to change is what previously lived in one person's head: it needs to become a record that anyone who was not in the room can retrieve, on their own, without asking around first. That does not require minuting every meeting formally. It requires that the commitments and decisions that used to survive by luck now survive by being written down and searchable, before the gap shows up as a missed promise instead of as a planning exercise.

Where a tool like Driffle fits

Driffle is built for exactly the point where informal memory stops covering a growing team. It transcribes a computer's audio directly, so no bot joins the call and no recording notice goes out, and it also holds screen context, so decisions made while looking at a doc or a deck are captured along with what was said. That record becomes searchable work memory: instead of asking around to find out what was decided in a meeting you were not in, you ask the record.

None of this replaces the judgment that made the founder a good memory holder in the first place. It replaces the assumption that one person's presence in the room is a reliable substitute for a record everyone can reach, an assumption that the channel math shows was only ever going to hold for a while.