# Retention Rate

> Calculate customer or logo retention for one fixed period, then decide whether the result is eligible to interpret.

- Type: Calculator: Keep the starting cohort
- Tags: Metrics, Retention
- Growth levers: Retention (primary)
- ~1377 words

---

**Retention Rate Calculator.** Share of customers who stick around across a period. Inputs: Customers at end of period, Accounts that entered after start, Customers at start of period. Outputs: Retention rate.

Customer retention rate is the percentage of a fixed starting customer or account cohort that remains active at the end of a fixed period. The number is only trustworthy when you can match that starting cohort to the ending ledger by stable account ID.

That sounds like bookkeeping, but it changes the answer. A total customer count can rise while part of the original base disappears. Retention asks one narrower question: of the accounts that existed at the opening boundary, how many are still active at the closing boundary? Use this calculator for that question, not for a blended total or a revenue number.

## Set the cohort boundaries before you calculate

Choose one unit first: a paying customer, billing account, workspace, or user. Do not call a user series logo retention, and do not switch from billing accounts at the start to workspaces at the end. Then freeze the population rules. The same paid status, product, geography, plan, and account-ID rules need to apply at both boundaries.

| Field | What to count | Boundary rule |
| --- | --- | --- |
| Accounts at start | Each active account in the chosen population at the opening instant | Count each stable account once. Exclude prospects, duplicate records, and accounts outside the population. |
| Accounts that entered after start | Every account active at end that was absent at start | Include an eligible reactivation that was absent at start. Do not add a start-cohort account that briefly churned and reactivated before end. |
| Accounts at end | Each active account in the same population at the closing instant | Keep the same product, geography, status, and ID rules used at the opening boundary. |
*A three-count retention calculation only works when the same stable-ID population can be reconciled at both boundaries.*

The middle field is where most retention spreadsheets go wrong. It does not mean first-time signups. It means every end-active account that was not in the start cohort. If account C was absent at the start, returns during the period, and is active at the end, C entered after start. If account A belonged to the start cohort, briefly churned, then returned, A still belongs to the original cohort. Without stable IDs, you cannot make that call safely.

## How customer retention is calculated

> **Formula:** Customer retention rate = ((accounts at end - accounts that entered after start) / accounts at start) x 100. The subtraction isolates the start cohort that remains active.

Take an annual logo-retention period. You start with 800 accounts, 90 accounts that are active at the end were absent at the start, and you finish with 760 active accounts. The retained start cohort is 760 - 90 = 670 accounts. So ((760 - 90) / 800) x 100 = 83.75%, which the calculator displays as **83.8% retention**. The scenario assumes stable IDs, one consistent population, and no unresolved merger or migration.

There is a useful sensitivity check hidden in that example. With 800 starting accounts, one misclassified end-active account changes retention by 100 / 800 = 0.125 percentage points. Ten misclassified accounts change it by 1.25 points. Small cohorts make identity mistakes show up faster, which is why the calculator asks you to confirm that the account ledger can be reconciled before it prints a result.

## When the calculator should refuse a verdict

A raw percentage is not automatically a retention result. For the three-count shortcut to be coherent, accounts at end minus accounts that entered after start must sit between zero and the starting account count. If 80 accounts are active at end and 90 entered after start, the implied retained start cohort is negative. If 1,000 are active at end, 90 entered after start, and 800 existed at the start, the implied retained start cohort is larger than the cohort itself. Both are data-definition problems, not retention results of a negative value or 113.8%.

The second failure mode is identity, even when the arithmetic looks sensible. A migration can replace IDs. A merger can turn two billing accounts into one. A split can do the reverse. Unknown reactivation history can make a returning account look newly acquired. Leave the stable-ID checkbox unticked when any of those cases remains unresolved. The calculator will show a review message rather than a metric, verdict, or action recommendation. Go back to a direct ledger of the start cohort first.

That distinction matters: invalid or identity-incompatible inputs get no retention result at all. Coherent counts with stable IDs get a result, but they still may not get an external comparison. Those are different kinds of uncertainty, and the page keeps them separate.

## Customer retention, GRR, and NRR are different metrics

| Metric | What stays in the numerator | Question it answers |
| --- | --- | --- |
| Customer or logo retention | Original accounts that remain active | Did the starting account cohort stay? |
| Gross revenue retention (GRR) | Revenue from the start cohort after churn and contraction, excluding expansion | How much starting revenue is still protected before upsells? |
| Net revenue retention (NRR) | Revenue from the start cohort after churn, contraction, and expansion | Is the starting revenue base shrinking or expanding? |
*The units and expansion treatment differ, so one metric cannot benchmark another.*

A company can keep most logos while losing a large account's revenue. It can also lose some logos while its remaining accounts expand. That is why customer retention cannot exceed 100%, while [net revenue retention](https://www.productgrowth.blog/calculators/net-revenue-retention) can. If revenue health is the decision in front of you, calculate GRR and NRR from the start cohort's revenue instead of stretching this account-count result past its job.

ChartMogul's [2023 SaaS Retention Report](https://chartmogul.com/reports/saas-retention-report/) separates customer retention, GRR, and NRR, and discusses its customer results by ARR and average revenue per account. SaaS Capital's [2023 survey](https://www.saas-capital.com/wp-content/uploads/2023/05/RB28WS1-2023-B2B-SaaS-Retention-Benchmarks.pdf) covers annual revenue retention by ACV for private B2B SaaS companies. Those are useful sources for understanding the boundary. They are not evidence for a universal customer-retention target in this calculator.

## Why the industry comparison is unavailable

The table below deliberately shows Unavailable in every cell. It is not a zero, a hidden failing grade, or a prompt to borrow a revenue-retention threshold. A defensible customer-retention comparison would need the same count-based measure, period, population, and category in every row. The sources reviewed for this page do not supply that matrix.

| Industry | Median | Good | Great |
| --- | --- | --- | --- |
| SaaS | Unavailable (unavailable) | Unavailable (unavailable) | Unavailable (unavailable) |
| Fintech | Unavailable (unavailable) | Unavailable (unavailable) | Unavailable (unavailable) |
| Dev Tools | Unavailable (unavailable) | Unavailable (unavailable) | Unavailable (unavailable) |
| AI/ML | Unavailable (unavailable) | Unavailable (unavailable) | Unavailable (unavailable) |
| E-commerce | Unavailable (unavailable) | Unavailable (unavailable) | Unavailable (unavailable) |
| Healthtech | Unavailable (unavailable) | Unavailable (unavailable) | Unavailable (unavailable) |
| Martech | Unavailable (unavailable) | Unavailable (unavailable) | Unavailable (unavailable) |
*Annual logo retention (%) · No defensible annual customer-retention industry comparator is currently used. Compare the same cohort and period with prior periods.*

That is a stronger answer than pretending 83.8% is good, bad, or average for every SaaS company. Compare this year's annual logo cohort with last year's annual logo cohort. Compare the same monthly cohort with the previous monthly cohort. Keep the period and inclusion rules fixed. Once a source reports a matching customer-count cohort for a specific category, it can earn a documented row. Until then, your own consistent history is the comparison that can guide a decision.

## What to do after you have a valid result

Start with the aggregate because it tells you whether the opening cohort held. Then make the cohort visible. Build a ledger or cohort table where every row carries a stable account ID, signup cohort, plan, acquisition channel, customer-size tier, renewal date, and active status at each boundary. Reproduce the same 800, 90, and 760 logic inside each slice before you interpret a difference.

Find the earliest interval where a cohort falls, then inspect the experience around that interval. If a plan or channel has a lower result, check the promise made at acquisition, the activation path, and the first event that signals real product value for that segment. Do not call any one stage the universal cause. The point of slicing is to turn one blended percentage into a small enough group that you can investigate with evidence.

For the same customer population and period, customer retention and [customer churn](https://www.productgrowth.blog/calculators/churn-rate) add to 100%. The 83.8% example therefore has 16.25% customer churn before rounding. That inverse is useful for reconciliation, but it does not replace the cohort slice. A churn percentage still cannot tell you which accounts left, when they left, or what they had in common.

#### What is a good customer retention rate?

There is no common industry band in this calculator because the reviewed public sources do not provide comparable customer-count thresholds for its seven categories. A valid result earns a comparison with your own prior period, using the same cohort definition and interval. A number from GRR, NRR, a different contract value, or a broad sector average is not a substitute.

#### What counts as an account that entered after start?

Count every account active at the end that was absent at the start. That includes an account that reactivated after being absent at the opening boundary. It excludes a start-cohort account that briefly churned and returned before the end. If account identity or reactivation history cannot establish the difference, use a direct start-cohort ledger instead of this three-count shortcut.

#### Can customer retention be above 100%?

No. Customer retention measures the portion of one fixed starting account cohort that remains active, so the retained portion cannot exceed the cohort. A value above 100% means the counts or cohort rules are incompatible, and this calculator should show a review message rather than a result. Net revenue retention can exceed 100% because expansion revenue is part of a different formula.

#### Can I convert monthly retention into annual retention?

Only as an illustrative survival calculation for the same fixed cohort and the same monthly survival rate. For example, 96% monthly retention repeated for 12 identical periods gives 0.96^12 = 61.3% annual retention. Do not annualize separately calculated blended monthly rates this way, because their populations and inclusion rules can change month to month.

**Next job: Build the stable-ID cohort slice.** Take the same starting population behind this result and break it out by signup cohort, plan, channel, customer-size tier, and renewal date. Find the first interval and segment where the retained start cohort falls before deciding what to change. Create the cohort ledger and reconcile its account IDs first.

---

All posts: https://www.productgrowth.blog/archive · Site: https://www.productgrowth.blog