A retention number moves for two reasons. Customers changed what they do, or someone changed how the number is counted. A report can tell the two apart only when its definitions are written down. The headline rate and the strategies that move it are covered in our customer retention guide. The full program scorecard, holdouts included, is in our loyalty program KPIs blog.
What are the main customer retention metrics?
Retention metrics fall into two groups. Count metrics ask whether customers stayed: the retention rate, churn rate, active member rate and cohort retention. Value metrics ask what the kept customers are worth: repeat purchase rate, purchase frequency, lifetime value and revenue retention. Pair one of each, so a rising count cannot hide shrinking spend.
| Metric | Formula | Inputs you need | Best for |
|---|---|---|---|
| Customer retention rate | Customers active at the start who are still active at the end ÷ customers active at the start × 100 | Customer IDs on both dates and a fixed definition of active. | Any business. It is the headline number. |
| Churn rate | Customers lost in the period ÷ customers at the start × 100 | The same group and period as the retention rate. | Contracts and subscriptions, where customers cancel on a known date. |
| Repeat purchase rate | Customers with two or more purchases in the window ÷ customers with at least one × 100 | Purchase dates per customer and a window length. | Retail, restaurants and ecommerce, early in the relationship. |
| Purchase frequency | Orders in the period ÷ unique customers who bought in the period | Order count and unique buyer count. | Businesses where customers never cancel. |
| Customer lifetime value | Average order value × purchases per year × gross margin × expected years retained | Margin, frequency and a retention estimate. | Deciding what a kept customer is worth. |
| Net revenue retention (NRR) | Recurring revenue now from customers who existed 12 months ago ÷ their recurring revenue then × 100 | Recurring revenue per customer on both dates, new customers left out. | Subscriptions, paid memberships and B2B. |
| Gross revenue retention (GRR) | The NRR formula with expansion removed, so it tops out at 100% | Lost and reduced revenue per customer. | Separating churn from upsell. |
| Active member rate | Members with a qualifying action in the window ÷ all enrolled members × 100 | The qualifying action, the window and the enrolled count. | Loyalty programs, where enrolled lists only grow. |
| Cohort retention | Members of a start cohort active in period n ÷ cohort size × 100 | Each customer’s start date and activity by period. | Comparing customers who joined at different times. |
Two choices decide most of these numbers: the action that counts as active, and the window. Write both down before the first report and keep them fixed. Our customer lifetime value guide works through that formula with numbers.
Revenue retention needs the same care. Datadog's 2025 annual report spells out its method. It takes recurring revenue from every customer it had 12 months earlier, then measures what those same customers pay now. Upgrades count, and cuts and lost customers come off. Revenue from new customers is left out. On that basis Datadog reported a trailing 12-month rate of about 120% at the end of 2025. The customers it already had were paying about a fifth more than a year earlier.
How do you measure customer retention when customers never cancel?
Subscription businesses see churn on a date. Retailers, restaurants and most loyalty programs never do. Fader, Hardie and Shang describe this noncontractual setting in Marketing Science. The firm never sees the moment a customer leaves, and a long gap since the last purchase is the only clue. In retail, retention is a definition you choose.
Public companies show the choice in their filings. In its January 2025 results, Starbucks counted U.S. Rewards members active in the past 90 days. Chewy counts a customer as active after at least one order in the preceding 364 days. Chewy notes that the change in that count mixes new customers coming in with lapsed ones going out. An active count on its own is a size, and a retention rate needs the same customers on both dates.
Set the window from your own purchase rhythm. Pull the gaps between purchases for repeat buyers, then pick a window that covers most of them. A coffee chain and an airline will land far apart, and both are right. Then hold the window fixed. A window that changes mid-year breaks every comparison with the year before.
How do you measure customer retention step by step?
- Pick the unit. Decide whether you count a person, a household or an account. A program that pools points by household has to settle this first.
- Define active. Name the action, such as a purchase, and the window, such as 90 days.
- Freeze the starting list. Save the IDs of everyone active on the first day of the period.
- Count who is still there. At the end, count the starting IDs that meet the active rule again. Divide by the size of the starting list.
- Split the total. Repeat the count for each join cohort and each acquisition channel. The total hides where the change came from.
- Add the value view. For the customers who stayed, report frequency and margin. A flat count with falling spend shows up here first.
Which customer retention metrics should you track?
Choose by how the relationship ends.
- Subscriptions and contracts. Track churn, GRR and NRR. The date a customer cancels is on record, so churn is a count. GRR shows the revenue you keep, and NRR adds what existing customers grow into.
- Paid memberships. Track the renewal rate on memberships that have come due. Costco's fiscal 2025 annual report shows a precise method. It measures renewals over the period 7 to 18 months before the reporting date. Memberships with an expiration date in the last six months stay out, because most late renewals arrive within six months. At the end of fiscal 2025 its rates were 92.3% in the U.S. and Canada and 89.8% worldwide. Our paid loyalty programs guide covers the design side.
- Retail, restaurants and travel. Track the active member rate and cohort retention, with repeat purchase rate and frequency beside them. Churn here is an inference from silence. Watch cohorts at the same age.
- Loyalty programs. Lead with the active member rate, because the enrolled count only rises. Pair it with the program metrics in the loyalty program KPIs blog.
Then tie the count to money. Lifetime value turns a retention gain into dollars. A two-point gain in a high-margin segment is worth more than the same gain in a low-margin one, and only the value view shows it. Segmentation turns that into separate targets per segment.
What mistakes distort customer retention metrics?
Using the end-of-period shortcut
The common formula is ((E minus N) ÷ S) × 100. E is customers at the end, N is new customers and S is customers at the start. It undercounts when new customers leave inside the same period. They sit in N but are missing from E, so they are taken off twice.
For illustration, a business starts the year with 1,000 customers. Of those, 850 stay. It adds 200 new ones, and 50 of those leave before the year ends. E is 1,000, so the shortcut reads 80%. The real retention of the starting customers is 85%. Keep the list of starting IDs and count how many are still active at the end.
Reading a rising blended rate as progress
Retention for a group of customers tends to rise with time, even when nobody changes. Fader, Hardie and colleagues explain why in the Journal of Interactive Marketing. Customers with a high churn risk drop out early, so the survivors lean loyal. A blended rate climbs as the base ages. Compare cohorts at the same age instead, such as month-six retention for this spring’s joiners against last spring’s.
Missing a shift in the mix
In fiscal 2025, Costco's renewal rate took in more memberships sold online, including some sold through digital promotions. Those members renew at a slightly lower rate on average, so they pulled the rate down. A new channel brings customers who behave differently. Split the rate by channel before you read the total.
Counting enrolled members as the base
A member count and an active base are different numbers. Costco’s paid member counts include memberships that expired in the prior 12 months without renewing. The filing says these are a small share. A loyalty program that treats every enrolled member as its base hides every lapse. Use active members as the starting group.
Multiplying monthly churn by twelve
Monthly churn compounds. A steady 5% monthly churn leaves 0.95 to the 12th power, or 54%, of customers after a year. Annual churn is 46%, well short of the 60% that simple multiplication gives.
Calling the member gap a retention lift
Members often start out ahead of non-members, because loyal customers tend to join first. That gap measures who joined, and our KPIs blog cites the research. To measure what a program or offer changed, hold out a random group and compare. Our control group testing entry and the loyalty program ROI guide cover the method.
How does cohort retention work in practice?
For illustration, take a restaurant brand that counts a member as active after at least one purchase in a calendar month. Members join at the till with a purchase, so month one is 100% by definition. It enrolls 10,000 members in January and 12,000 in April. Each cohort is tracked by its age in months, so the two compare fairly.
| Cohort | Size | Month 1 | Month 3 | Month 6 | Month 12 |
|---|---|---|---|---|---|
| January joiners | 10,000 | 10,000 (100%) | 4,200 (42%) | 3,100 (31%) | 2,400 (24%) |
| April joiners | 12,000 | 12,000 (100%) | 5,640 (47%) | 4,080 (34%) | Not yet reached |
April’s joiners hold up better at the same age: 47% against 42% at month three, and 34% against 31% at month six. Something changed between the two intakes. The next step is to split each cohort by acquisition channel and by welcome offer to find the cause.
The same January cohort gives the value metrics. Over 12 months, 5,300 of its members bought twice or more, a repeat purchase rate of 53%. They placed 38,000 orders in total, a purchase frequency of 3.8 per member.
Now the blended trap. Suppose the brand reports one June figure: members active in June ÷ all members enrolled since January. Every new intake counts as 100% active in its first month. A heavy enrollment month lifts the blend, even if no existing member bought more. The cohort view keeps the members apart from the marketing calendar.
How does GRAVTY report customer retention metrics?
GRAVTY is Loyalty Juggernaut's loyalty platform. GRAVTY Intelligence tracks how the program performs, with real-time analytics and pre-built KPIs. Member 360 keeps one view of each customer, with the household included. A retention rate is only as good as that customer count.
With Agentic AI Compass, a team can ask why a metric changed and compare any segment in plain language. When the fix is a win-back rule, the patented Visual Rules engine lets the loyalty team author it, simulate it and deploy it without an IT ticket. A holdout on that rule then shows whether it worked.
Frequently asked questions
What is a good customer retention rate?
A good customer retention rate is one that beats your own earlier cohorts at the same age. Cross-industry averages mix contract businesses with retailers, and they mix definitions of active. Fix your definition, then compare this year’s cohorts with last year’s.
What is the difference between retention rate and churn rate?
Churn rate is the mirror of retention rate: 100% minus retention, for the same customers and period. The pair only adds up when both start from one group. In retail, churn is inferred from a lapse window, so it carries that window’s assumptions.
How often should you measure customer retention?
Measure retention on the cycle your customers buy on. Monthly fits a coffee chain. Yearly fits a membership that renews once a year. Review cohorts each quarter whatever the cadence, because they take months to mature.
What is the difference between NRR and GRR?
Gross revenue retention counts only what existing customers keep paying, so it tops out at 100%. Net revenue retention adds expansion, such as upgrades and higher usage. At 90% GRR and 110% NRR, a business loses a tenth of its starting revenue and gains a fifth back from the customers who stay.
Is repeat purchase rate the same as customer retention rate?
No, they answer different questions. Repeat purchase rate asks whether a customer bought again inside one window. Retention rate asks whether a customer from the start of a period is still active at the end. Repeat rate is the better early signal for new customers.