Can You Measure a Community?
The three metrics seem to measure different things, but I think they share an underlying idea: a product becomes valuable when people form a community around it. The numbers capture traces of those relationships, not the relationships themselves.
For Airbnb, nights booked measured how often guests and hosts successfully found each other. Unlike revenue, it was less distorted by prices and exchange rates. But if a team chased nights with steep discounts or pushed unnecessarily long stays, bookings might rise without strengthening trust between hosts and guests.
For Facebook, seven friends in ten days was a proxy for reaching a social network worth returning to. The team could game it by aggressively suggesting weak connections. The graph would become denser, but users might feel no more connected.
For Slack, 2,000 cumulative team messages marked the point at which a group had genuinely tried the product. Stewart Butterfield reported that 93% of teams crossing that threshold were still customers. But encouraging trivial replies or excessive notifications could push teams past 2,000 while making collaboration worse.
What these examples make me wonder is whether we confuse the activity of a community with the health of a community. A booking is not trust; a friend count is not friendship; a message is not cooperation. The metrics are useful signals, but Goodhart’s Law warns us that once we reward the signal, people can produce it without producing the value behind it.
