Retention rate
Retention rate is the percentage of customers from a defined starting group who are still buying after a set period. It is calculated as customers retained ÷ customers at the start × 100, and ecommerce operators use it to measure whether the store keeps the customers it paid to acquire.
| Variable | What it covers |
|---|---|
| Customers at period start | Active customers when the window opens. |
| New customers added | First-time customers acquired inside the window - excluded so acquisition cannot masquerade as retention. |
| Customers at period end | Active customers when the window closes. |
Conventions differ. The formula above suits contractual or subscription revenue. Ecommerce purchases are episodic, so the more honest measure is cohort repurchase retention: the share of one period's first-time buyers who purchase again within a set window, read through a cohort analysis.
Worked example
Example numbers. Lifting this cohort's retention from 28% to 33% keeps 50 more customers - roughly 95 further orders, A$9,025 of revenue and A$3,610 of contribution margin from customers already paid for.
What is a good retention rate?
It depends, and any single benchmark hides more than it shows. What drives it: the category's natural purchase cycle (coffee reorders monthly; a mattress may not repeat for years), the window you measure over (a 90-day view punishes long-cycle categories), whether you measure a cohort or the whole base, and whether the customers you retain are margin-positive at all. Retention drives the frequency and lifespan terms of lifetime value, so the way to price what a retention point is worth in your own margin structure is the free Margin LTV calculator.
Retention rate vs related metrics
| Metric | What it measures | How it differs from retention rate |
|---|---|---|
| Repeat purchase rate | Share of the whole base with two or more orders. | Base-wide and cumulative; retention is time-boxed on a specific cohort. |
| Cohort analysis | Behaviour grouped by first-purchase period. | The method that makes retention readable, not a rate itself. |
| Customer lifecycle | States such as active, at risk and lapsed. | Replaces a single percentage with where each customer sits right now. |
| Win-back | Recovering customers who already lapsed. | Starts where retention failed; retention is keeping them from lapsing at all. |
Common mistakes
- Measuring the whole base instead of cohorts. Heavy acquisition floods the denominator with new one-order customers and reads as collapsing retention when nothing changed.
- A window shorter than the purchase cycle. Judging a long-cycle category on 90-day retention marks perfectly healthy customers as lost.
- Binary retained-or-lost with no lifecycle states. A customer three weeks overdue and one lapsed for eight months look identical in a single rate - they need opposite actions.
- Retaining unprofitable customers. Retention of margin-negative customers compounds the loss; a retention rate only means anything on customers who leave margin behind.
Retention rate FAQ
Related
Blufire S4 Customer Value & Segmentation tracks lifecycle states and migration for every customer, so retention shows up as movement between active, at-risk and lapsed states instead of a single average. The Math teaches the full method free.
Updated July 2026