Returning Customer Rate vs Cohort Analysis: Key Difference
Returning customer rate and cohort analysis sound alike but answer different questions. Learn the formulas and 5 good and bad cohort patterns.
By Nitish KumarOct 21, 2026 • 5 min read
Returning customer rate and cohort analysis both talk about repeat customers, so they get mixed up all the time. But they look in opposite directions, and using one in place of the other leads to wrong decisions.
The difference at a glance
Looks at one month
Returning customer rate
- Starts with all orders in a month
- Asks how many came from old customers
- Tells you who is buying now
Looks forward
Cohort retention
- Starts with one month's new customers
- Asks how many came back later
- Tells you customer health
| Returning customer rate | Cohort retention | |
|---|---|---|
| Starting point | All orders in one month | Customers who made their first order in one month |
| Question | How many of this month's orders came from old customers? | How many of this month's new customers came back later? |
| Looks | At one month | Forward, across many months |
| Tells you | Who is buying now | Customer health |
1. Returning customer rate
This is the share of a month's orders that came from customers acquired in the past.
| Metric | Formula |
|---|---|
| Returning rate | Orders from returning customers ÷ total orders in the month × 100 |
- Example: October has 1,000 orders and 200 are from returning customers. 200 ÷ 1,000 × 100 = 20%.
- Check the label: reporting tools define this differently. Some report the share of customers with two or more orders. Confirm the definition before you compare numbers.
2. Cohort retention
Take the customers whose first order was in one month. Then track how many of them order again in M1 (next month), M2, M3 and onwards.
| Metric | Formula |
|---|---|
| Cohort retention (month N) | Cohort customers who ordered in month N ÷ cohort size × 100 |
- Example: 500 customers joined in October. 70 of them ordered in November. 70 ÷ 500 × 100 = 14% in M1.
| Cohort | New customers | M0 | M1 | M2 | M3 |
|---|---|---|---|---|---|
| Aug | 400 | 100% | 6% | 5% | 14% |
| Sep | 450 | 100% | 6% | 15% | 4% |
| Oct | 500 | 100% | 14% | 5% | – |
| Nov (sale) | 900 | 100% | 3% | – | – |
3. Why one number can fool you
In the example above, November's returning rate jumps from 20% to 30% (750 of 2,500 orders). That looks great. The cohort table tells another story:
- The green cells form a diagonal. Every older cohort came back only in November, during the sale.
- The 900 customers acquired during the sale returned at just 3% in December.
The returning rate said old customers love the sale. The cohort said sale-acquired customers do not stay.
4. What each one tells you
Snapshot
Returning rate
- How much revenue depends on existing customers
- Whether a campaign to old customers (email, WhatsApp) worked this month
- Jumps in sale months, which are expected
Long view
Cohort retention
- Customer health: do new customers come back at all?
- Whether repeat buying is consistent, or green only in sale months
- Which acquisition month or channel brought better customers
- When customers drop off, for example a big fall after M1
5. How to read a cohort table
- Green only on the diagonal (sale months): customers repeat only for discounts.
- Green spread across M1, M2 and M3: healthy, consistent repeat behaviour.
- Weak M1 in every row: your post-purchase journey needs work.
- One cohort clearly weaker: check what you did that month (offer, product, channel).
6. Good and bad cohort patterns: 5 scenarios
In each table, rows are cohorts (the new customers of that month) and columns are M0 to M3. Green means strong repeat, grey means weak.
Scenario 1: Consistent cohorts Good
| Cohort | M0 | M1 | M2 | M3 |
|---|---|---|---|---|
| Jun | 100% | 14% | 11% | 10% |
| Jul | 100% | 15% | 12% | – |
| Aug | 100% | 14% | – | – |
- The curve dips after the first month, then flattens. Every cohort looks alike.
- What it means: customers return because of the product, not because of offers.
- What to do: scale acquisition, because new customers stay.
Scenario 2: Improving cohorts Good
| Cohort | M0 | M1 | M2 | M3 |
|---|---|---|---|---|
| Jun | 100% | 6% | 5% | 4% |
| Jul | 100% | 9% | 7% | – |
| Aug | 100% | 13% | – | – |
- Each new cohort beats the one before it (M1: 6%, 9%, 13%).
- What it means: something is working, such as a new WhatsApp flow, better packaging or a product fix.
- What to do: find out what changed and do more of it.
Scenario 3: Sale-only spikes Bad
| Cohort | M0 | M1 | M2 | M3 |
|---|---|---|---|---|
| Aug | 100% | 5% | 4% | 15% (Nov sale) |
| Sep | 100% | 6% | 16% (Nov sale) | 5% |
| Oct | 100% | 15% (Nov sale) | 5% | – |
- Green appears only on the diagonal, in the sale month. Every other month is cold.
- What it means: customers come back only for a discount.
- What to do: give customers a reason to reorder in non-sale months, such as a replenishment reminder, a new launch or a bundle. Do not mistake sale repeats for retention.
Scenario 4: Big drop after M0, no tail Bad
| Cohort | M0 | M1 | M2 | M3 |
|---|---|---|---|---|
| Jun | 100% | 4% | 2% | 1% |
| Jul | 100% | 4% | 2% | – |
| Aug | 100% | 3% | – | – |
- Every cohort is the same, and every cohort is weak.
- What it means: customers arrive, buy once and leave.
- What to do: check product quality, delivery experience and your post-purchase flow. First ask whether your product is a one-time purchase (furniture, for example), because low repeat is normal there.
Scenario 5: Worsening cohorts Bad
| Cohort | M0 | M1 | M2 | M3 |
|---|---|---|---|---|
| Jun | 100% | 15% | 12% | 10% |
| Jul | 100% | 11% | 8% | – |
| Aug | 100% | 7% | – | – |
- Each new cohort is weaker than the one before it (M1: 15%, 11%, 7%).
- What it means: your new customers are not like your old ones. Common causes are a new ad audience, more discount-led acquisition or a channel change.
- What to do: find the month the decline began and look at what changed in acquisition that month.
Quick summary
| Pattern | Verdict | Next move |
|---|---|---|
| Consistent | Good | Scale acquisition |
| Improving | Good | Find what changed and do more of it |
| Sale-only spikes | Bad | Build non-sale reasons to reorder |
| Big drop, no tail | Bad | Fix product or post-purchase experience |
| Worsening | Bad | Audit channel, audience and discounts |
Note: a "good" M1 number is not the same for every store. Products like skincare or food are bought again more often than furniture. Compare your cohorts with your own earlier cohorts first, not with someone else's benchmark.
7. Which one to use
| You want to know | Use |
|---|---|
| Did my retention campaign work this month? | Returning rate |
| Are the customers I acquired worth keeping? | Cohort |
| Was last month's sale a one-time spike? | Both: returning rate for the spike, cohort for what came after |
8. Common mistakes
- Treating a high returning rate as good retention. It can rise simply because new customers fell.
- Judging cohorts with very few customers (10 to 20).
- Mixing guest-checkout duplicates and refunded orders into the counts.
- Comparing sale-month cohorts with normal-month cohorts without saying so.
Pro Tip: always put the returning rate and the cohort table side by side for the same months. If they disagree, trust the cohort.
New to cohorts? Start with our guide to customer cohort analysis for ecommerce growth, then build your own table with the free Customer Cohort Analysis Tool. For the order-value side of repeat buying, see 7 practical ways to increase AOV with product bundles. To check your overall conversion, use the free Conversion Rate Calculator. To lift the conversion that brings in those new customers, see our ecommerce conversion rate optimization guide.
Final Thoughts
Returning rate is a monthly snapshot of who buys. Cohort is the long view of whether new customers stay. You need both, but they answer different questions.
Your next step: pull your last 6 months of orders and find the M1 retention of each month's new customers. Compare it with your returning rate for the same months. Want it done for you? Upload your order data to the free Customer Cohort Analysis Tool and see how many customers come back, month by month.