How does retention compare across markets, and are recent expansion countries holding up?
№ 050 · Cohort retention by country
Definition
MetricBuyer cohort retention curves split by market or country
Unit% of cohort active at month N, per country
Rows are countries instead of acquisition months, columns are months since acquisition. Newer expansion markets have shorter rows. Mature markets retain better but recent expansion is the leading indicator. Build by tagging users with country and tracking retention per country per month. Example: US 81% at M5; BR 69% at M3.
Benchmarks
| Bottom 30% | Median | Top 30% |
|---|---|---|
| expansion cohorts plateauing at less than half the home-market level at the same tenure after the launch period, or degrading cohort over cohort within a market | home market outperforming young markets at equal tenure is normal; app-layer averages for shopping and marketplace apps (D30 approx 2-8.7% depending on sub-category) vary by region with APAC mobile-first markets typically showing lower D30 but different web and superapp repeat paths | expansion-market cohorts converging to home-market curves within 2-3 quarters of launch at the same tenure; all markets clearing the consumer-transactional bar of approx 30% m6 on transacted buyers |
consumer transactional retention survey (2020) for the m6 bar plus 2024-2026 app network aggregates for D1/D7/D30 by category; country-level splits in public benchmark data are directional rather than tabulated, so the operative comparison is internal cross-market at equal tenure · Public country-level retention tables at marketplace granularity are scarce; network reports publish category and regional aggregates, so this card benchmarks primarily against your own home market at equal tenure. Regional differences also reflect measurement (app vs web vs superapp mix), payment infrastructure and category mix, not only product quality.
Category split omitted: No two independent Tier 1-2 sources publish marketplace retention tables by country; regional aggregates exist only at app-category level.
When it looks bad
The per-country small multiples show the home market flat-lining healthily while one or more expansion markets decay without plateau at every tenure, and successive cohorts within the weak market start lower each month.
Home market m6 sits at 34%; a market launched three quarters ago shows m6 at 12% with m1 dropping 41% to 29% across its monthly cohorts.
What to do about it
- Compare markets only at equal tenure (months since that market's launch and cohort age) with the home market's early curves as the reference, so a young market is judged against the home market's own year-one shape rather than its mature curve.
- Diagnose weak markets in order: liquidity first (fill rate and supply coverage by city), then payment and delivery infrastructure completion rates, then category mix vs home market; retention gaps abroad are usually supply and rails, not brand.
- Localize the retention engine, not just the storefront: local payment methods, local supply curation and market-specific lifecycle timing; measure each localization by its lift on m1-m3 retention in that market's next cohort.
- Set a convergence deadline per market (expansion cohorts within a set band of home-market curves at equal tenure by quarter N); markets missing it twice get investment paused rather than blended into the global average.
Sources
- Casey Winters and Lenny Rachitsky consumer transactional m6 bar (approx 30% good, 50% great) applies per market, not blended; blended curves hide market variance caseyaccidental.com ↗
- UXCam (aggregating AppsFlyer and Adjust) shopping and e-commerce D30 median approx 2% with strong performers 3-6% across 50+ countries; category benchmarking beats a global average uxcam.com ↗
- Sendbird (aggregating AppsFlyer, Adjust, Statista) marketplace apps D30 approx 8.7% vs shopping apps approx 4.8-5.6%; regional and sub-category composition drives spread sendbird.com ↗
- Phiture cross-source 2025-2026 aggregates D1 25-26%, D7 11-13%, D30 5-7% with wide category variance; benchmark within category and geography phiture.com ↗
- Andreessen Horowitz (Olivia Moore) cohort comparison discipline: newer cohorts (here, newer markets) should be compared to older ones at the same tenure; divergence flags value-prop transfer failure a16z.com ↗
- ClickPost (Dynamic Yield derived) regional commerce baselines differ structurally (Americas vs APAC vs EMEA basket and device mix), so cross-country retention gaps partly reflect market structure clickpost.ai ↗