We noticed that aggressive push notification frequency was driving a short-term spike in session opens, but our user satisfaction scores were declining. I flagged this to the team and made the case that we were burning user goodwill. We reduced notification frequency and saw satisfaction recover over the following quarter. It was a clear example of choosing the user experience over a vanity metric. The team was initially hesitant but came around once I explained the reasoning, and ultimately it was the right call for the product.
On our notifications team, we were hitting a daily send cap that lifted 7-day re-engagement rate by 14% — a number growth loved. But our cohort data showed 30-day notification opt-out rate climbing 22% among weekly actives, the users who drive long-term retention. I brought this to a cross-functional review with growth, engineering, and the data science lead. Growth pushed back hard — that 14% number was in their OKRs. I held the position using opt-out velocity as the leading indicator: at current trajectory we'd lose opt-in permission from a meaningful share of our most engaged cohort within two quarters. We ran a 6-week experiment reducing frequency for that segment, accepted a 9% short-term re-engagement drop, and saw opt-out rate fall 18%. Six months later, 90-day retention for that cohort was up 11 points.