How many sellers from each cohort stay active at month N?
№ 037 · Seller cohort retention
Definition
MetricSeller retention: % of a seller cohort still actively transacting at month N
Unit% of cohort sellers active at month N
Same triangular structure for suppliers. Watch the first-3-month cliff: are sellers getting at least one successful order? Build by tracking sellers with at least one sale per month, grouped by onboarding date. Example: Jan 24 sellers 38% active at M12; Apr 25 sellers 64% at M3.
Benchmarks
| Bottom 30% | Median | Top 30% |
|---|---|---|
| supplier logo retention below 50% by m12, or active-seller definitions loosened (listing but not selling) to hide it | 50-70% supplier logo retention among funded consumer marketplaces (a16z observed range), with supply GMV retention plateauing 45-50% by m12 for the average company | 60-70% of suppliers still active at extended tenure, with retained sellers expanding GMV 2-3x, producing supply GMV retention at or above 100% |
consumer marketplaces, seed through Series B, a16z 2020-2022 dataset; active seller defined as transacting, not merely listed; power-law context from Marketplace Pulse across Amazon, eBay, Etsy, TikTok Shop · Seller logo retention is rarely published directly; the 50-70% range is inferred by a16z from its dataset and sits above true market average. Because GMV follows a power law, blended logo retention can mask the loss of a few high-GMV sellers; pair this card with the supply concentration card. Define active as transacted in the period, and state the period.
Category split omitted: No two independent Tier 1-2 sources publish seller retention splits by category; only blended ranges and power-law shape data exist.
When it looks bad
Cohort lines step down sharply after month 1-2 and keep sliding, with the drop concentrated in sellers who never made a first sale; GMV-weighted retention diverges below logo retention when larger sellers start leaving.
m3 seller retention 38%; 55% of churned sellers never completed a sale; two top-decile sellers left in the quarter, taking 9% of GMV while logo retention fell only 4 points.
What to do about it
- Engineer the first sale: guarantee early liquidity for new sellers via boosted placement or house demand in their first 30 days; a16z identifies time-to-first-sale and perceived market price as the decisive retention drivers, and StockX-style seeding of early listings is the reference tactic.
- Track GMV-weighted seller retention beside logo retention, and run a named retention program for the head of the power law (top 1-5% of sellers) with dedicated support, fee tiers and tooling, since that cohort carries most GMV.
- Give retained sellers expansion tools (demand analytics, listing optimization, inventory finance) so the 2-3x GMV expansion that defines best-in-class supply retention has a mechanism, not just hope.
- Diagnose churned sellers by exit state: never-sold, sold-then-stalled, and multi-tenanting-out require different fixes (activation, demand allocation, exclusivity perks).
Sources
- Andreessen Horowitz (Olivia Moore) 50-70% of suppliers retain and expand GMV 2-3x in best-in-class cases; first-sale experience (price, time to sale, process) is the primary retention driver a16z.com ↗
- Casey Winters and Lenny Rachitsky supply-side retention is typically the stronger side; viable demand retention depends on it, and vice versa via transaction volume per seller caseyaccidental.com ↗
- Marketplace Pulse power-law context: 1.6% of active US Amazon sellers produce 50% of 3P GMV; retention of the head cohort matters disproportionately marketplacepulse.com ↗
- Marketplace Pulse power-law distributions consistent across Amazon, eBay, Etsy; a marketplace loses GMV unevenly when sellers churn marketplacepulse.com ↗
- Everything Marketplaces (Marketplace Capital) supplier switching costs are a structural category property that fixes the retention ceiling; category selection dominates tactics everythingmarketplaces.com ↗