NSM is a measure of accumulated user value. It decomposes multiplicatively.
NSM = Breadth x Depth x Frequency x Efficiency
Airbnb: Nights Booked
Having a booked night as the NSM is genius because both sides win: a guest gets a stay, and a host gets paid. It’s a proxy for matched supply and demand, and it decomposes as follows:
Nights Booked = Active Guests x Trips per Guest x Nights per Trip
Facebook: 7 friends in 10 days
Chamath Palihapitiya’s (former VP of Growth at Facebook) growth team replayed new users’ first days and found that people who reached seven friends within ten days stuck around, and most below it churned.
Slack: 2k messages
They looked at customers who stayed and found that once a team had exchanged 2,000 messages, 93% were still using Slack. About a week of chat for a team of ten. This metric serves as a clean proxy for a whole team moving its conversations in instead of a single person.
Messages sent within Active Spaces = Team Size x Days Active per week x Messages per day
Impact on Goodhart’s Law
These numbers work wonderfully for these big companies because they measure a behavior that the team cannot directly cause or influence. For instance, Airbnb subsidizing long, cheap stays would reduce the quality of trust and hosts, eventually causing the next booking to worsen. Facebook would have a mandatory friend-adding step at signup or perhaps recommend strangers, leading to duller feeds. Slack spams join notifications and inflates the channels with meaningless messages. The threshold stops predicting retention and is rendered useless.
