Markets and business models
Consumer retention
Acquisition can be bought. Retention usually cannot. Cohort charts beat aggregate DAU, and payer retention drives LTV.
Retention is the consumer product's truth serum. Acquisition can be bought; retention usually cannot. A flat aggregate DAU can hide a leaky bucket refilled by paid spend. Cohort charts beat top-line actives.
Retention as a product discipline is how you fix the curve. This page insists you measure it as an economic input.
Concept
A cohort is a group who started in the same period (signed up in January, started a trial in week 12). D1 / D7 / D30 ask what share of that cohort returns on day 1, 7 and 30. Payer retention asks whether paying users stay — the number that drives LTV.
Useful lenses
Definitions vary — pick one and stick to it.
| Lens | What it asks | Rough intuition |
|---|---|---|
| D1 / D7 / D30 | Of a signup cohort, what % return on day 1 / 7 / 30? | Social/habit apps need strong early retention; tools can be weekly |
| W1 / W4 | Weekly actives returning | Useful when daily use is not the job |
| DAU/MAU | Stickiness of the active base | ~20%+ is often "good" for many apps; messaging/social aim higher |
| Paid retention M1/M3/M12 | Do payers stay? | For subscriptions, this drives LTV more than free DAU |
Vanity DAU from a launch spike or "AI tourists" can look like growth while paid retention is terrible. For monetized consumer products, track retained payers and revenue retention, not just top-line installs. Rebase after tourist churn.
Small changes dominate acquisition hacks. From the subscription sketch: moving paid lifetime from 8 to 12 months lifts LTV 50% with no new creatives.
Involuntary churn (failed cards, expired payment methods) silently erases paid cohorts. Dunning emails, account updaters and smart retries are revenue systems. Treat billing reliability like uptime.
Novelty converts well; habit retains poorly in some AI categories. Prefer products embedded in weekly workflows over one-shot demos that never reopen.
Common mistakes
- Watching aggregate DAU while paid cohorts melt. The leaky bucket looks full.
- Celebrating install spikes from tourists who never pay or return.
- Ignoring involuntary churn. A failed card is churn you can often prevent.
Check yourself
Aggregate DAU is flat. Why might the product still be dying?
Paid spend is refilling a leaky bucket. Cohort retention of each new group is falling.
Which retention number matters most for a subscription app?
Payer retention (M1/M3/M12). Free DAU does not pay the bills.
Key takeaways
- Measure cohort retention and payer retention, not just install spikes.
- DAU/MAU and D1/D7/D30 are complementary. Pick definitions and keep them.
- Lifetime is the highest-leverage LTV lever. Retention work is growth work.
- Involuntary churn is a billing-reliability problem, not a product-love problem.
Further learning
Retention Is All You Need (a16z)AI companies donAndreessen Horowitz
Lenny's Newsletter — consumer metricsWhich metrics matter most when building a B2C productlennysnewsletter.com
Next, hits and platform risk.