Airbnb’s nights booked
Airbnb treats a night booked as the single unit of transaction on its platform. Every night means a host got paid and a guest got a place to stay so it captures value on both sides of the marketplace at once. A team chasing it could subsidize cheap stays or loosen listing standards, and as a result nights would rise but bad stays would erode the trust that makes people book a stranger’s home in the first place.
Facebook’s 7 friends in 10 days
Chamath Palihapitiya called getting any new user to seven friends in ten days the growth team’s keystone since past that point people got enough value that they rarely churned. I think of seven friends as roughly where the feed becomes worth opening. A team could hit it with aggressive contact imports and suggestions of near strangers but the feed would fill with weak ties and invites would start to feel like spam.
Slack’s 2,000 messages
Stewart Butterfield found that 93% of teams that had exchanged 2,000 messages were still using Slack. For a ten-person team that’s about a week of chatter so it signals a team that actually moved its conversations over. The easy hack is bots and integrations flooding channels. The count would jump but people would then get buried in noise and mute Slack which is the opposite of the habit the number was meant to track.
So what?
All three numbers came from studying users who stuck around, so they’re correlations rather than levers you can pull directly. I wonder if they still mean much now that AI agents can send messages, friend requests and even bookings on our behalf
Sources
– Airbnb Form S-1 (2020): https://www.sec.gov/Archives/edgar/data/1559720/000119312520294801/d81668ds1.htm
– Chamath Palihapitiya on Facebook’s keystone metric: https://cur.at/MjWdcQX?m=web
– Butterfield quoted in Andrew Chen’s The Cold Start Problem: https://www.goodreads.com/notes/56962799-the-cold-start-problem/84411704-mikko-ikola/435bc888-c6a7-45f1-83c2-6c66617e1ab2
