Cohort GMV Value

Measure success by total value from a cohort over years, not retention or per-user metrics

Archie Abrams
Growth and experimentation at Shopify

Cohort GMV Value

"It's really looking at the total value, but on that GMV basis. And GMV is a power law based metric. And so it's really that power law that drives the success of each cohort. Again, going back to investing, same thing there. Each vintage from a fund, how much did that return as a fund? And it's really driven by the few really successful outliers." - Archie Abrams

What It Is

Cohort GMV Value is a measurement approach where you track the total gross merchandise value (or equivalent business metric) generated by a cohort of users over multiple years, rather than focusing on retention rates, average revenue per user, or other per-user metrics.

This approach embraces power-law dynamics: most users in a cohort may churn or produce little value, but the outlier winners generate so much value that the entire cohort becomes profitable. Think of it like venture capital vintage returns—individual deal success rates matter less than whether you found the big winners.

How It Works

The Shopify model:

  1. Acquire a cohort of merchants in a given quarter
  2. Track total GMV that cohort produces over 1, 2, 3, 4, 5+ years
  3. Don't worry about per-merchant averages or retention rates
  4. Let the power law work—the Allbirds and FIGS of the cohort make the whole thing profitable

Why this works for platform businesses:

  • Shopify's revenue comes primarily from payments (percentage of GMV), not subscriptions
  • One merchant doing $100M GMV is worth more than 1,000 merchants doing $10K each
  • Lowering barriers to entry means more at-bats for finding the next big winner
  • Most businesses fail anyway—entrepreneurship has inherent churn

The mindset shift:

  • Traditional SaaS: "Keep every customer, maximize per-customer value"
  • Cohort GMV: "Maximize total value from winners, accept that losers will churn"

How to Apply It

  1. Define your cohort metric - What's your equivalent of GMV? Total transactions? Total content created? Total value facilitated?

  2. Track cohorts over long horizons - Build systems to see cohort performance at 1, 2, 3+ years, not just 90 days

  3. Reframe churn narratives - High churn isn't necessarily bad if your winners are big enough

  4. Lower barriers to entry - If power-law dynamics apply, you want more at-bats, not higher-quality average users

  5. Calculate payback on cohort basis - "Cohort value over cost and payback"—not per-user CAC/LTV

When to Use It

  • Platform and marketplace businesses with transaction-based revenue
  • Businesses with power-law user value distributions
  • When your "winners" are dramatically more valuable than average users
  • When lowering friction creates more potential winners
  • When retention metrics encourage over-qualifying users

Why Not Per-User Metrics?

Per-user metrics can be misleading in power-law businesses:

Metric Problem
Retention rate Encourages qualifying users out, missing potential winners
ARPU Averaging destroys power-law signal
LTV Individual LTV matters less than cohort total
CAC/LTV ratio Can improve by excluding promising-but-risky users

"We don't think tons about churn, we almost optimize for churn... We want to lower the barriers to getting started and get as many people in the door trying their hand at entrepreneurship."

Source

  • Guest: Archie Abrams
  • Episode: "Growth and experimentation at Shopify"
  • Key Discussion: (00:10:18) - How Shopify measures success through cohort GMV
  • YouTube: Watch on YouTube

Related Frameworks

Back to Growth