10 Customer Retention Metrics to Track (Formulas and Examples)
- The standard retention rate formula subtracts customers acquired during the period. Skip that subtraction and you are measuring growth, not retention, and a strong acquisition month will hide real customer loss.
- Retention rate and churn rate are the same number from opposite ends: over the same period and the same starting base they add up to 100 percent. Repeat purchase rate is not part of that pair, because its denominator is buyers in the period, not customers at the start.
- Cohort measurement beats snapshot measurement. A snapshot blends every acquisition wave into one average, so a shift in your acquisition mix looks identical to a shift in retention.
- Rate metrics tell you how many customers stayed; value metrics (CLV, AOV, time between purchases) tell you whether the ones who stayed are worth more; experience metrics (NPS, CSAT) tell you why. You need one from each group.
- Every retention metric needs a stated definition of an active customer and a stated measurement period. Without those two, the number cannot be compared to last quarter or to anyone else.
Customer retention metrics measure whether customers keep buying from you after the first purchase. Ten of them cover the ground: customer retention rate, churn rate, repeat purchase rate, loyal customer rate, customer lifetime value, average order value, time between purchases, Net Promoter Score, customer satisfaction score, and days sales outstanding. Each has a formula, and each answers a different question.
Three of these are confused with each other constantly. Retention rate and churn rate are the same number seen from opposite ends. Repeat purchase rate looks like a third view of the same thing but uses a different denominator, so it is not. This guide gives every metric its formula and a worked example, puts them in one table, and is explicit about the definitions that get muddled.
What are customer retention metrics, and what does each group do?
Customer retention itself is the ability to keep customers buying over a defined period. Keeping an existing customer is generally cheaper than winning a new one, because you have already paid the acquisition cost and the customer already knows the product. The metrics below are how you find out whether that is actually happening in your business.
Two definitions have to be settled before any of these formulas produce a comparable number:
- What counts as an active customer. In subscriptions this is easy: an active subscription. In eCommerce there is no cancellation event, so you have to pick a window, for example "purchased in the last 180 days". The window should cover roughly two of your typical purchase cycles.
- What the measurement period is. Monthly, quarterly, annual. A metric calculated monthly cannot be compared with the same metric calculated quarterly, and a period shorter than your purchase cycle will report churn that has not happened.
Write both down and put them next to every number you report. Most retention arguments inside a company are actually arguments about these two definitions.
Which rate metrics show how many customers stayed?
1. Customer Retention Rate (CRR)
Customer Retention Rate is the percentage of the customers you started a period with who were still customers at the end of it. It is the headline retention number.
Worked example. A brand starts the quarter with 1,000 customers, acquires 200 new ones, and ends with 1,050.
CRR = ((1,050 − 200) ÷ 1,000) × 100 = (850 ÷ 1,000) × 100 = 85 percent.
The subtraction of new customers is the whole point of the formula, and it is the step most often dropped. Without it, the calculation returns 105 percent and reports a retention success in a quarter where 150 existing customers actually left. Retention rate has to measure the customers you already had; growth is a separate question.
2. Churn Rate
Churn Rate is the percentage of customers who stopped buying during the period. It is retention rate read from the other side.
Worked example. A subscription service starts the month with 1,000 customers and loses 50.
Churn Rate = (50 ÷ 1,000) × 100 = 5 percent.
Measured over the same period and on the same starting base, churn rate and retention rate add to 100 percent. In the CRR example above, 150 of the original 1,000 customers were lost, so churn was 15 percent against a retention rate of 85 percent. Report the one that fits the conversation: retention rate when you are talking about relationships holding, churn rate when you are sizing the hole acquisition has to fill. In eCommerce, churn is silent, which is why the definition of an active customer matters so much here. There is more on the causes and the fixes in our guide to reducing customer churn in eCommerce.
3. Repeat Purchase Rate (RPR)
Repeat Purchase Rate is the share of your customers who bought more than once in the period. It measures whether a first purchase turns into a second.
Worked example. Of 1,000 customers who bought in the last 12 months, 400 placed two or more orders.
RPR = (400 ÷ 1,000) × 100 = 40 percent.
Note the denominator: customers who bought during the period, not customers you had at the start. That is what makes this a different metric from retention rate rather than another view of it. RPR is also period-sensitive in a way the rate metrics are not. Lengthen the window and the number rises mechanically, because customers have had more chances to buy again.
4. Loyal Customer Rate
Loyal Customer Rate is repeat purchase rate with a higher bar. Instead of "more than one purchase", you set a threshold that means something in your category, such as three or more orders in a year.
Worked example. A brand has 1,000 customers, and 250 of them placed at least three orders in the last 12 months.
Loyal Customer Rate = (250 ÷ 1,000) × 100 = 25 percent.
This metric is only as good as the threshold you choose, so choose it from your own data rather than copying one. A reasonable method: look at the order-count distribution of your customers and set the bar where the drop-off flattens, which is usually where genuinely habitual buyers separate from the occasional ones. Because the threshold is yours, this number is for tracking against yourself over time, never for comparing against another company.
Rate metrics by cohort and by RFM segment, calculated from your order history rather than a spreadsheet.
Learn more about Nexus by Omniconvert →Which value metrics show what a retained customer is worth?
5. Customer Lifetime Value (CLV)
Customer Lifetime Value estimates the total revenue a customer generates across the whole relationship. It is the metric that makes retention a financial argument rather than a marketing one.
Worked example. Average purchase value $50, purchase frequency 5 orders per year, average customer lifespan 3 years.
CLV = 50 × 5 × 3 = $750 per customer.
That is revenue CLV. Multiply by gross margin to get the number you can actually compare with customer acquisition cost, which is the comparison that decides whether a channel is worth spending on. Note the direct link to churn: average lifespan is 1 ÷ churn rate, so a fall in churn lengthens lifespan and lifts CLV without anyone spending more on marketing. Our full breakdown of the calculation and its variants is in the customer lifetime value guide.
6. Average Order Value (AOV)
Average Order Value is how much a customer spends per transaction. On its own it is a commerce metric, not a retention metric, but it is one of the three inputs to CLV and it shows whether retained customers are buying more deeply over time.
Worked example. $10,000 in revenue from 200 orders.
AOV = 10,000 ÷ 200 = $50.
The retention question is not what your blended AOV is but whether AOV rises with order number. If a customer's third order is worth more than their first, the relationship is deepening. If it is worth less, you are probably retaining people with discounts. Split AOV by order sequence before drawing conclusions, and see how to increase average order value for the levers.
7. Time Between Purchases
Time Between Purchases is the average gap between one customer's consecutive orders. It sets the clock for everything else: it tells you what measurement period to use, when a customer is genuinely late, and when a win-back email is early rather than annoying.
Worked example. A customer placed 4 orders across 12 months. There are 3 gaps between those 4 orders, so the average gap is 12 ÷ 3 = 4 months. Average that figure across all repeat customers to get the brand-level purchase cycle.
Subtracting one matters: four orders create three intervals, not four. Once you know the typical gap, a customer who has passed roughly twice that gap without ordering is your practical churn threshold, and a lengthening gap for a previously frequent customer is one of the earliest churn signals you can get.
Which experience metrics explain why customers stay or leave?
8. Net Promoter Score (NPS)
Net Promoter Score asks one question, usually "how likely are you to recommend us to a friend or colleague?" on a 0 to 10 scale. Answers of 9 and 10 are Promoters, 7 and 8 are Passives, and 0 to 6 are Detractors.
Worked example. Of 200 respondents, 120 are Promoters, 50 are Passives and 30 are Detractors. Promoters = (120 ÷ 200) × 100 = 60 percent. Detractors = (30 ÷ 200) × 100 = 15 percent.
NPS = 60 − 15 = 45. The result is a score between −100 and +100, not a percentage, and Passives count only by diluting the two percentages.
NPS is a relationship measure, so send it on a schedule rather than after every transaction, and always ask the follow-up "why". The score tells you the direction; the comments tell you what to fix. See our comparison of NPS, CSAT and CES for when to use each one.
9. Customer Satisfaction Score (CSAT)
Customer Satisfaction Score measures satisfaction with one specific interaction: a delivery, a support ticket, a return. It is transactional where NPS is relational.
Worked example. 200 customers respond to a post-support survey on a 1 to 5 scale, and 150 of them rate the interaction 4 or 5.
CSAT = (150 ÷ 200) × 100 = 75 percent.
The denominator is respondents, not everyone you surveyed, so report your response rate alongside the score. And state your definition of "satisfied" (typically the top two boxes on a 5-point scale), because a CSAT calculated on top-one-box is a different and much harsher metric.
What is the tenth, operational retention metric?
10. Days Sales Outstanding (DSO)
Worked example. $100,000 in accounts receivable, $500,000 in credit sales, over a 30-day period.
DSO = (100,000 ÷ 500,000) × 30 = 6 days.
If you sell direct to consumers and take payment at checkout, DSO is not your metric and you should ignore it rather than force it into a report. Where it does apply, read it per account rather than in aggregate: a customer who used to pay in 20 days and now takes 55 is signaling either a dispute you have not heard about or a decision they have not told you about yet.
What are all 10 customer retention metrics at a glance?
| Metric | Formula | The question it answers | What distorts it |
|---|---|---|---|
| Customer Retention Rate | ((End − New) ÷ Start) × 100 | How many of my existing customers stayed? | Forgetting to subtract new customers; an undefined "active customer" |
| Churn Rate | (Lost ÷ Start) × 100 | How big is the hole acquisition has to fill? | A churn window shorter than the purchase cycle counts slow buyers as lost |
| Repeat Purchase Rate | (Customers with 2+ orders ÷ Customers in period) × 100 | Does a first purchase turn into a second? | Rises automatically as you lengthen the period; different denominator from CRR |
| Loyal Customer Rate | (Customers above your threshold ÷ Total customers) × 100 | How many customers are genuinely habitual? | The threshold is self-set, so the number is not comparable across companies |
| Customer Lifetime Value | Avg purchase value × frequency × lifespan | What is a customer worth over the relationship? | Revenue CLV mistaken for margin CLV when compared against CAC |
| Average Order Value | Revenue ÷ Orders | How much is spent per transaction? | A blended figure hides whether AOV rises or falls with order number |
| Time Between Purchases | Period ÷ (Orders − 1) | How fast do customers come back? | Dividing by order count instead of the number of gaps; seasonal categories |
| Net Promoter Score | % Promoters − % Detractors | How strong is the relationship overall? | Low or skewed response rates; surveying only after a good experience |
| Customer Satisfaction Score | (Satisfied respondents ÷ Respondents) × 100 | Did this specific interaction go well? | An unstated definition of "satisfied"; the denominator is respondents, not recipients |
| Days Sales Outstanding | (Receivables ÷ Credit sales) × Days | Are credit customers still paying on time? | Irrelevant to businesses paid at checkout; aggregates hide the one account in trouble |
What is the difference between retention rate, churn rate and repeat purchase rate?
The clearest way to hold them apart is to look at what each one is dividing by:
- Retention rate divides by the customers you had at the start of the period, and counts how many survived to the end. It is about the base you already owned.
- Churn rate divides by the same starting base and counts the ones who did not survive. Same denominator, opposite numerator, so the two are complements.
- Repeat purchase rate divides by the customers who bought during the period. Someone who was a customer at the start and bought nothing is in the retention denominator but not in the RPR denominator at all.
The practical consequence: a brand with a nine-month purchase cycle can report a strong annual retention rate and a weak quarterly repeat purchase rate at the same time, and neither number is wrong. They are answering different questions over a window that suits one and not the other. If you want the full comparison with worked numbers, see repeat purchase rate vs retention rate vs churn rate.
One more pair worth separating: customer churn and revenue churn. Customer churn counts heads. Revenue churn counts money. Losing 5 percent of customers who happen to be your highest spenders is a far worse quarter than the customer churn number will suggest, which is why high-CLV segments deserve their own churn line.
Cohort vs snapshot: how should you measure retention?
Here is how a snapshot misleads. A brand doubles paid acquisition in Q1. First-time buyers churn faster than established customers in almost every business, so the following quarter's blended retention rate falls. Nothing about the customer experience changed; the mix did. Read as a snapshot, that looks like a retention problem and sends the team off to fix something that is not broken.
The cohort view separates the two. Track the January cohort, the February cohort and so on, and compare each at the same age, month 3 against month 3. If the March cohort's month-3 retention is below January's, retention genuinely moved. If every cohort's curve is unchanged and only the blend shifted, the story is acquisition, not retention.
Segment the cohorts as well as dating them. Customers acquired on a first-order discount, customers from a specific channel and customers who bought a specific first product all retain differently, and the differences are usually larger than any month-over-month movement in the average. RFM segmentation is the standard way to cut the same data by behavior rather than by acquisition date, and the two views answer different halves of the question.
How do you set up retention measurement?
-
Define an active customerPick the recency window that separates a live customer from a lapsed one, based on your typical time between purchases. Roughly two purchase cycles is a sound default. Write the definition down. Every metric below depends on it.
-
Choose a measurement period that matches your purchase cycleMonthly reporting suits subscriptions and fast consumables. Quarterly suits most eCommerce. If your customers buy twice a year, a monthly retention rate reports noise. The period has to be long enough for a loyal customer to have had a real chance to buy again.
-
Pick one metric from each groupOne rate metric, one value metric, one experience metric. Retention rate, CLV and NPS is a good starting trio. Reporting all ten from day one produces a dashboard nobody reads and no decisions.
-
Calculate by cohort, not only in aggregateRebuild the same metrics for each acquisition month and compare cohorts at the same age. This is the step that turns a number you report into a number you can act on, because it tells you whether retention itself moved.
-
Give each metric an owner and a review rhythmA metric with no owner changes nobody's behavior. Attach each number to the team that can move it, review on the same cadence as the measurement period, and pair every reported figure with the definition and period used to produce it.
Once the definitions are settled, the calculation itself should not be manual work. Nexus by Omniconvert builds retention rate, churn, repeat purchase rate, CLV and cohort views directly from order data, and segments customers by RFM so you can see which groups are slipping rather than only that the average moved. Where the metrics point at a page or a flow that is losing people, Omniconvert Explore is where you test the fix, including the on-site NPS and CSAT surveys that feed the experience metrics above.
Measurement is the start, not the goal. What you do with the segments the metrics reveal is covered in our customer retention strategy guide.
FAQ: customer retention metrics
Customer retention metrics are the measures that show whether customers keep buying from you after their first purchase. They fall into three jobs: rate metrics that count who stayed and who left (customer retention rate, churn rate, repeat purchase rate, loyal customer rate), value metrics that show what a retained customer is worth (customer lifetime value, average order value, time between purchases), and experience metrics that explain why people stay or leave (NPS, CSAT). Days sales outstanding is added when you sell on credit terms.
Customer Retention Rate = ((Customers at End of Period - New Customers Acquired During the Period) / Customers at Start of Period) x 100. Subtracting new customers is the part people forget. Without it you are measuring growth, not retention, and a heavy acquisition month will hide the fact that your existing customers are leaving. Example: start with 1,000 customers, acquire 200, end with 1,050. Retention rate = ((1,050 - 200) / 1,000) x 100 = 85 percent.
They are two views of the same number. Retention rate counts the customers who stayed; churn rate counts the customers who left. Measured over the same period and on the same starting base, retention rate + churn rate = 100 percent. Use retention rate when you report on whether relationships are holding, and churn rate when you size the loss you have to replace or the revenue at risk.
No. They use different denominators and answer different questions. Retention rate uses the customers you had at the start of the period and asks how many were still customers at the end. Repeat purchase rate uses the customers who bought during the period and asks how many of them bought more than once. A brand can have a high retention rate and a low repeat purchase rate if its purchase cycle is longer than the period being measured.
Snapshot retention measures everyone who was a customer at the start of a period, mixing people acquired years apart into one number. Cohort retention follows one fixed group, usually customers acquired in the same month, and tracks what percentage of that group is still buying in month 1, 3, 6 and 12. Cohorts are the more honest view because they separate a real change in retention from a change in your acquisition mix.
No. Revenue measures the money a business generates over a period. Churn rate measures the percentage of customers who stopped buying in that period. The two can move in opposite directions: revenue can rise while churn rises, if new acquisition is covering the loss. That is exactly why you track them together. Revenue churn, which measures the money lost rather than the customers lost, is a separate metric again and is usually worse news, because higher-spending customers are more expensive to lose.
Product return rate is the percentage of items sold that customers send back, and it is a leading indicator of churn. A high rate points at a gap between what the product page promised and what arrived: sizing, quality, description accuracy or shipping damage. Returns also cost the customer time, which erodes trust even when you refund them promptly. Track return rate by product and by category rather than as one blended number, because the problem is almost always concentrated in a few SKUs.
There is no single good number, because retention rate depends on category, purchase cycle and how you define an active customer. A consumables brand and a furniture brand can both be healthy at retention rates that differ by 40 points. The useful benchmark is your own trend: compare this quarter's cohort against the same cohort age six and twelve months ago, and compare segments against each other. A retention rate that is improving cohort over cohort matters more than one that beats a published average.
Pick three metrics, not ten. Take one rate metric (customer retention rate, calculated the standard way, with new customers subtracted), one value metric (customer lifetime value, or average order value if your data is thin), and one experience metric (NPS or CSAT, whichever you can collect consistently). Write down your definition of an active customer and your measurement period before you calculate anything, because those two choices decide what the number means. Then rebuild the same three numbers by acquisition cohort. The blended average tells you where you are; the cohort view tells you whether it is getting better. Once those are stable and trusted, add the rest of the ten as questions come up.
Calculate your retention metrics from real order data
Nexus by Omniconvert turns your order history into retention rate, churn, repeat purchase rate, CLV and cohort views without a spreadsheet, and segments customers by RFM so you can see which groups are slipping. Built on 13 years of customer data across 7,000+ websites and 15+ industries.