By Teresa Torres · producttalk.org · @ttorres on X · LinkedIn
Torres argues that all-time totals, like lifetime registered users, are rarely useful because they only go up and hide whether a product is improving or declining. The right measure depends on choosing a time increment that reflects how often customers actually use the product and that gives the team something to act on. Monthly, weekly, or daily active users may not fit products people use only a few times a year, such as textbook rental sites. Her practical suggestion is to compare each week's new signups as a cohort against prior weeks while the team is still learning what success looks like. The point matters because metric choice shapes which product decisions teams can make and how quickly they learn.
01Key takeaways
- Avoid all-time totals, since they only rise and tell you nothing about whether you are improving.
- Pick the time increment that matches how customers naturally use your product.
- Ask how a metric will change your behavior before you adopt it.
- Monthly feedback is often too slow for teams that ship changes almost daily.
- When you do not yet know the right engagement measure, compare cohorts week over week.
02Key sections
- Why all-time totals mislead
- Cumulative figures like lifetime signups always rise, which creates a false sense of progress. They reveal nothing about current momentum or decline.
- Vanity metrics and the Lean Startup lens
- Building on Eric Ries, the author shows how vanity metrics feel good but offer nothing to act on. Companies that chased registered users without engagement illustrate the danger.
- Choosing the right time increment
- The best interval is the one that matches the product's natural usage rhythm and that is short enough to give quick feedback. Monthly active users are often too slow for teams that ship frequently.
- Products with infrequent usage
- For products used only a few times a year, daily, weekly, and monthly activity metrics are mostly meaningless for much of the year. Teams need measures tied to the actual usage cycle.
- Weekly cohorts as a practical default
- While uncertain about the right engagement metric, the author's team compares each week's signup cohort with prior cohorts. This gives a usable signal for product decisions.
03From the post
“Do you know how much money you've made over your entire lifetime? Probably not. Or how many miles you've ran since the day you were born? Maybe. If you aren't a runner. I can only think of one person who might know the answers to both of”
04Frameworks mentioned
Summary and takeaways written by PM Atlas; quotes are short excerpts. © the original author.