Advanced plan & up

Cohort retention analysis

How to read Ordinary's cohort retention curves — what fraction of customers come back for a 2nd, 3rd, Nth order — and use them to benchmark loyalty.

Ordinary Written by The Ordinary Team · Updated

Cohort retention analysis

Cohort retention is available on Advanced and higher.

Shows, for each order number N (1 through 10), what percentage of customers have placed at least N orders. The curve describes how deeply your customers buy.

How to read the curve

The x-axis is order number. The y-axis is percent of the cohort.

  • N=1 is always 100% — every customer in the cohort has at least one order (otherwise they wouldn’t be in the cohort).
  • N=2 is your “do they come back?” rate. If 40% of customers place a second order, your N=2 is 40%.
  • N=3, N=4, … decay from there. The slope tells you whether customers who come back once keep buying, or whether most of your repeat revenue is from a small loyalist core.

Typical shapes

  • Flat decay — 50% → 30% → 20% → 15% — most of your revenue is concentrated in a repeat core; new customers drop off fast.
  • Steep then flat — 40% → 12% → 10% → 9% — high first-repeat gate; anyone who passes it stays for a while.
  • Slow decay — 60% → 50% → 42% → 38% — very loyal customer base; subscription stores often look like this.

Three modes

Above the chart, a mode selector:

  • Org-wide (default) — every customer who’s placed ≥1 order. The store-wide retention curve.
  • Product — Any — customers who’ve bought a specific product, bucketed by their total order count across your store. Answers: “when a customer buys this product, how deeply do they engage with my brand overall?”
  • Product — Same — customers who’ve bought a specific product, bucketed by how many times they bought THAT product. Answers: “do customers come back for THIS specific product?”

Product modes require picking a product from the selector next to the mode dropdown.

Worked example

You run a protein powder brand. Org-wide retention curve is 100% → 45% → 30% → 22%.

  • 45% at N=2 says: of everyone who ever ordered, 45% came back for a second order. OK for the category.
  • 30% at N=3 says: 30% came back a third time. The drop from 45% to 30% (a 33% “exit rate” at the second-to-third jump) is a signal — customers who ordered twice mostly ordered a third time. The second order is the harder gate.
  • Product — Same for your flagship flavor might be 100% → 55% → 40%. That’s higher than the org curve because this product has a specific reorder pattern.

Common questions

Why start at N=1? N=1 gives you the base size so you can sanity-check the cohort. It’s always 100% by construction.

How is the cohort defined? Your entire paid-customer base in the date range. Customers with 0 orders are excluded (they’re “Prospects,” not a retention cohort).

Is there a time axis? Not yet — this chart is depth-only. A time-bucketed version (e.g. “of customers acquired in Jan, what fraction repeated in Feb / Mar / Apr”) is the Cohort LTV report, tracked as a post-launch enhancement.

Can I export this? Not yet. CSV export is on the post-launch roadmap.

Where it feeds into other parts of Ordinary

The retention curve is also the denominator for the Offer calculator — it uses the implied probability that a discounted first-time buyer comes back.

Did this answer your question?

Thanks for your feedback! 🙌

Related articles