Airbnb steered by nights booked because it captures both sides of the market at once. A night only counts if a host listed a place and a guest paid to stay there. That makes it a better signal of real use than site visits or signups. A team could game it with heavy discounts, cheap one-night stays, or looser quality checks on listings. Bookings would go up, but guests would have worse stays, hosts would earn less, and trust in the platform would drop.
Facebook’s early growth team found that users who added 7 friends in their first 10 days were far more likely to stick around. Friends were the proxy for a feed worth coming back to. A team could game it by auto-adding loose connections or spamming “People You May Know” prompts. Users would hit 7 friends, but their feed would fill with people they don’t care about, so the habit never forms.
Slack noticed that teams who sent about 2,000 messages almost always kept using it. Hitting that number meant a team had actually moved its conversations into Slack. A team could game it with bots, automated alerts, or nudges to post filler. Message counts would climb, but users would get noisy channels and notification overload, which pushes them away.
Goodhart’s Law says that when a measure becomes a target, it stops being a good measure. All three numbers worked because they tracked a real behavior. Once a team chases the number instead of the behavior, the metric looks great while the product gets worse. That’s why good teams pair a north star with guardrails like retention, reviews, or satisfaction.
