The skill: Finding where and why users churn, then designing interventions. The #1 thing that separates products that compound from those that slowly die. Without retention, every other growth effort is a leaky bucket.
Benchmarks
Diagnose your retention problem
Start with the curves
Don't look at aggregate retention numbers — they hide everything. Plot retention by monthly cohort. Each cohort is a line. What you're looking for: does the curve flatten or tick up (smile curve = healthy, users who stay stay), or does it keep declining (frown curve = no stable base)? You can't grow out of a frown curve. Every new user you acquire leaks out the bottom.
Distinguish activation retention from long-term retention. Day-1 churn is an onboarding problem. Month-3 churn is a value problem. They have different causes and completely different fixes. Treating them the same wastes time on the wrong intervention.
One more thing: churn is a lagging indicator. By the time it shows up in your monthly numbers, the damage happened weeks ago. Hiten Shah tracks daily retention rate as his #1 metric for this reason — it surfaces problems while they're still fixable.
Find the biggest drop-off
If 50% of users leave on day 1 and 5% leave in month 3, fix day 1. Don't optimize month-3 retention while hemorrhaging on day 1. This sounds obvious, but teams routinely build features aimed at engaged users while ignoring the massive drop-off at the front door.
Andrew Chen calls this the Next Feature Fallacy: when retention is bad, the instinct is to build more features. But a feature buried behind day-7 engagement will only be seen by ~4% of your original visitors. Fix the landing page, onboarding, and first-run experience first. That's where the biggest drop-off lives, and that's where features have the most mathematical leverage.
Diagnose the cause
Before you pick a fix, identify whether you have a friction problem or a desire problem. Adam Nash identifies these as the two levers for engagement: lowering friction and increasing desire. Most retention fixes focus on friction — removing steps from onboarding, speeding up load times, simplifying flows. But if users complete onboarding fine and still don't come back, friction isn't the problem. They don't want to return. That requires different interventions: better content, stronger hooks, features that create anticipation. A faster onboarding won't help if the product doesn't give users a reason to open it tomorrow.
To find out which it is, interview churned users. But ask what they switched to, not why they left. What they switched to reveals the value you failed to deliver. "Why did you leave?" generates polite excuses.
For two-sided products, there's a third cause: the supply side isn't showing up. Emmett Shear built Twitch's retention by retaining streamers first — revenue tools, community features, customizable channels. He asked creators exactly what they wanted and gave it to them. If your product has a "hard side" that creates value for everyone else, their retention is your retention.
Watch for degrading cohorts
If newer cohorts consistently retain worse than older ones, the problem might not be your product — it might be who you're acquiring. Andrew Chen calls this the Law of Shitty Cohorts: as you scale past early adopters, each new cohort tends to retain worse. Early users came via word of mouth with strong intent. Later cohorts arrive through paid channels with weaker intent. You can't "fix" this with features — you need to fix your acquisition targeting.
Your next wave of users also retains differently because they're different people. Bangaly Kaba's Adjacent User Theory: they have different needs, different expectations, and different activation paths. Segment retention by acquisition source and user type. The fix for organic sign-ups churning might be completely different from the fix for paid-channel sign-ups churning.
Ship fixes in cohorts
Don't roll out retention improvements to everyone at once. Ship to a subset, isolate who saw what, and measure retention per cohort. If you can't attribute a retention change to a specific product change, you're guessing. Hiten Shah's approach: cohort exposure on every change, so you always know what moved the number.
Retention is the foundation of growth. Viral loops, content marketing, and paid acquisition all fail if users don't stick. Fix retention before investing in acquisition.
Do's and Don'ts
Written with ❤️ by a human (still)