Retention Benchmark Ladder
Concrete week-one cohort retention floors: ~80% friends/family, 60% free at scale, 20-30% paid.
- Difficulty
- Moderate
- Time to result
- ~weeks to results
- Steps
- 4
- Confidence
- 90%
A set of concrete cohort-retention thresholds for judging whether a product genuinely solves a job. Judge cohorts by whether week-one retention hits its floor and then flattens: friends-and-family should be near 80%, a free product at scale should hold at least 60% week-over-week, and a paid product 20-30%. If you can't retain the people who care about you, the product won't solve the job for strangers.
Origin
Widjaja's operating benchmarks from Gojek (which saw ~60-70% early retention) and advisory work; she frames it as cohort analysis, not a rolling rule of thumb.
Core principles
- 01Retention is the truest signal that a product solves a real job — if people keep coming back, it just needs to work
- 02Benchmarks scale with audience warmth: friends/family highest, then free-at-scale, then paid
- 03Week-one is the key reading; healthy cohorts flatten (stop dropping) within 2-3 frequency periods
- 04If you can't retain people who care about you (~80%), you won't retain strangers
- 05Don't be fooled by a launch spike — you may have just pulled forward everyone who could ever convert
How to run it
- 1
Measure by cohort, not rolling average
Track each signup cohort's return rate at your natural frequency (weekly, monthly). Week-one is the anchor reading you judge against the floor.
Pro tip Use cohorts so you can see the curve flatten rather than blending new and old users into a misleading rolling number.
- 2
Apply the right floor for your audience and model
Hold friends-and-family cohorts near 80%; a free product at scale should retain at least ~60% week-over-week; a paid product ~20-30%.
Pro tip For free products, a 60% week-one that flattens is the 'we're going to be fine' signal — Gojek saw 60-70% early because it solved a huge problem.
Watch out Retention only goes down from your earliest cohorts because early adopters are the most excited — so start high.
- 3
Confirm the curve flattens within 2-3 periods
Give teams two to three frequency periods for the cohort to flatten; it should settle around the floor (e.g. ~60% for free) rather than continuing to decay.
Watch out Don't mistake a launch uptick for durable retention, especially in new markets — you may have pulled forward all reachable demand and now have to work far harder for the rest.
- 4
Fix the step before conversion when short
When retention lags, the culprit is usually the step right before conversion — often ineffective copy or an unclear reason the user opened the app or reached checkout — so fix that friction rather than blaming the whole product.
In the wild
In Gojek's early days, weekly cohort retention ran 60-70% because the product solved a huge transportation and delivery problem. Widjaja took that as the signal the company would be fine — if people keep coming back, the product just needs to work.
→ High early cohort retention became the confidence signal for the business's durability.
When Netflix and Spotify expanded internationally, they saw a first-week signup spike and assumed it would only improve. In reality, few users in those markets held credit cards those services accepted, so the spike had pulled forward everyone who could subscribe.
→ A launch uptick that flattered the numbers while masking a much harder road to acquire the rest — a warning against reading spikes as durable retention.
Common mistakes
Reading a launch spike as durable retention
A big first-week uptick, especially in a new market, can just be pulling forward all reachable demand (e.g. everyone with an accepted credit card). Assuming it 'only gets better' hides that you now must work much harder for everyone else.
Setting the retention bar too low for the audience
Expecting modest retention from friends and family is a red flag — if the people who care about you don't return at ~80%, the product likely won't solve the job for anyone else.
Is it for you?
Best for
Founders judging whether early usage indicates real product-market fit versus early-adopter enthusiasm.
Not ideal for
Inherently low-frequency products where weekly cohort framing doesn't match the natural usage cadence.
From the transcript
“if it's a free product 60 percent right it has to be at least 60 if it's a free product week over a week if…”
“if you are much smaller like your friends and family that better be near close to 80 no matter what because if you can't even…”
“60 should be your week one and then it should flatten”
“that's actually what we saw at gojek early days it was like 60 70 retention rates”
“you just pulled forward everyone who could have possibly subscribed to you now you're going to have to work a lot harder”
From the episode
How to scrappily hire for, measure, and unlock growth
Crystal Widjaja, Gojek and Kumu