How much of each cohort's spend do we retain over time?
№ 031 · GMV retention by cohort
Definition
MetricGMV retention: cohort spend in month N as % of the cohort's first-month spend
Unit% of first-period cohort GMV
Triangular heatmap where rows are buyer signup months and columns are months since signup. Values show % of M0 spend retained. Build by grouping orders by buyer cohort and summing GMV per month-since-signup, ratio'd to M0. Example: Oct 25 cohort 67% at M5; Mar 26 cohort 100% at M0.
- GMV
- Gross merchandise value. Total value of goods sold through the platform before returns, discounts and take rate.
Benchmarks
| Bottom 30% | Median | Top 30% |
|---|---|---|
| cohorts that never flatten: continued decay past m12, or newer cohorts plateauing visibly below older ones | supply side 80-95% through m1-m3, plateauing around 45-50% by m12; demand side materially lower than supply in most models | supply side at or above 100% at m12 (best-in-class); demand side flattening at a high level with curves that plateau or smile between m6 and m12 |
consumer marketplaces, seed through Series B, a16z dataset approx 2020-2022, monthly GMV retention indexed to m0; public IPO comps report annual year-2 vs year-1 GMV retention and sit above private averages · The a16z sample skews toward companies the fund engaged with, so likely above true market average. Best-in-class supply retention above 100% comes from 50-70% of suppliers retaining while expanding GMV 2-3x, not from universal retention. Corroborating investor bars: at least 80% quarterly GMV retention cited for Series A readiness.
| Category | Bottom 30% | Median | Top 30% |
|---|---|---|---|
| Supply side (consumer marketplaces) | below 40% at m12 and still decaying | 45-50% at m12 (80-95% in m1-m3) | at or above 100% at m12 |
| Demand side (consumer marketplaces) | no plateau by m12 | materially below supply side; typical curves plateau at a fraction of m0 spend | curves flatten by m6 at a level supporting LTV:CAC of 4x+ |
When it looks bad
Spaghetti lines fan downward with each newer cohort sitting below the last, and no line flattens; the heatmap shows color draining left to right with the bottom rows darker than the top rows at the same tenure.
Jan cohort at m6 retains 52% of GMV, April cohort 41%, July cohort 33%; none has plateaued by m9.
What to do about it
- Fix the first-sale experience on the supply side: time-to-first-sale and perceived market price drive supplier return behavior per a16z; instrument days-to-first-transaction per new supplier and clear listings that have not sold in the first 30 days.
- Build expansion mechanics for retained suppliers, since best-in-class 100%+ retention comes from 2-3x GMV expansion by the 50-70% who stay: demand analytics, listing optimization tools, and inventory financing for power sellers.
- On the demand side, engineer the match: filtered discovery and preference capture before showing options, which a16z credits for strong demand retention in managed service marketplaces.
- Compare cohorts at fixed tenure in a single view each month; if new cohorts underperform old ones at the same month, treat it as a supply quality or mix problem, not a marketing problem.
Sources
- Andreessen Horowitz (Olivia Moore) average supply-side GMV retention 80-95% in m1-m3 plateauing around 45-50% by m12; best-in-class at or above 100% through m12; retained suppliers expand GMV 2-3x; demand side lower than supply a16z.com ↗
- Everything Marketplaces (Marketplace Capital) current fundraising bar references GMV alongside retention; take-rate-dependent GMV thresholds by stage everythingmarketplaces.com ↗
- Qubit Capital Series A diligence bar cited at 80%+ quarterly GMV retention with a path to positive unit economics qubit.capital ↗
- Casey Winters and Lenny Rachitsky consumer transactional m6 user retention approx 30% good, 50% great; GMV retention sits above user retention when retained users expand spend caseyaccidental.com ↗
- Andreessen Horowitz cohort quality and fragmentation determine marketplace durability; retention curves are the core comp investors run a16z.com ↗
- BS&Co DTC reference point for the demand side: 18.8% of 156K customers placed a second order within 365 days, showing how low unaided repeat behavior is without marketplace liquidity bsandco.us ↗