How this instrument works
Customer retention rate measures what share of the customers a business already had are still customers when the period closes. The formula does that by subtracting new customers from the ending count before dividing by the starting count — a small step that matters more than it looks. Without it, a business that lost a third of its original base but backfilled the gap with new signups could read a healthy-looking ending headcount and mistake growth for loyalty. Stripping the new accounts out first forces the arithmetic to answer one question only: of the people who were already paying, how many stayed?
A revenue-operations lead or a CFO reaches for this figure when the raw data on hand is three headcounts — customers at the start of a period, customers at the end, and how many of those end-period customers signed on fresh — rather than a dedicated log of who canceled. Many CRM exports and monthly dashboards carry exactly those three numbers without a separate 'customers lost' field, so this formula backs the loss out algebraically instead of requiring a tally that may not exist. Boards and investor updates also favor the retention framing over a churn framing for the same underlying arithmetic, since a stated 95% reads differently than a stated 5% even though the two numbers are mirror images of each other.
The result is a headcount ratio, not a revenue ratio — a customer who downgraded to the cheapest plan but never canceled counts exactly the same as one who kept paying full price, and a customer worth ten times the average counts the same as one at the entry tier. It also assumes every account entering as 'new' truly is new for that period; a canceled customer who rejoins and gets logged as new rather than as a returning original customer will quietly inflate the reading, since the formula has no way to tell a fresh signup from a win-back unless the underlying data distinguishes them.
- Enter Customers at start of period — the headcount on day one, before any of that period's activity.
- Enter Customers at end of period — everyone still active or newly won by the close, added together.
- Enter New customers added during period — the count to strip out so only original customers remain.
- Read Retention rate, % — the instrument computes (end minus new) divided by start, times 100.
- Recompute each period on the same length so one reading compares cleanly against the last.
Worked example — 560 end, 85 new, 500 at start
A subscription business opens the quarter with 500 customers and closes it with 560, having won 85 brand-new accounts along the way. Enter 560 as Customers at end of period, 85 as New customers added during period, and 500 as Customers at start of period, and the instrument returns (560 − 85) ÷ 500 × 100 = 95%.
Read naively as end divided by start, 560 ÷ 500 comes out to 112% — a figure that looks like growth in the original base and hides the fact that 25 of the original 500 customers actually left. Subtracting the 85 new customers first exposes that reality: only 475 of the original 500 were still customers at the close, a 95% retention rate that new signups were quietly covering up in the raw headcount.
Questions
Why subtract new customers before dividing by the starting count?
Because ending headcount alone conflates two separate things: customers who stayed and customers who just arrived. A period can lose a third of its original base and still show ending growth if enough new signups landed in the same window, so dividing end by start directly can overstate loyalty or even exceed 100%. Subtracting new customers first isolates the original base and answers only how many of them remained.
Is this the same thing as 100% minus churn rate?
Algebraically, yes, when both figures use the same starting count and period length — retention and churn are two readings of the same underlying loss. The difference is what data each formula expects: churn rate needs a direct tally of customers lost, while this one works from three counts — start, end, and new — that many dashboards and CRM exports already carry without a separate loss field.
How does this differ from Net Revenue Retention?
This is a logo count, not a dollar figure — it treats a customer who downgraded but didn't cancel the same as one still paying full price, and it ignores expansion revenue from upsells entirely. Net Revenue Retention weights by contract value and folds in both expansion and contraction, which lets it climb past 100%; this formula cannot, since retained customers can never outnumber the customers a period started with.
What counts as a new customer instead of a retained one?
A new customer is anyone who signed on for the first time inside the period being measured. Anyone present at the start who is still active at the close counts toward the retained numerator regardless of plan changes, pauses, or downgrades along the way — the formula only checks whether they are still a customer, not whether the terms of that relationship changed.
What retention rate counts as healthy for a subscription business?
Benchmarks vary by segment: gross logo retention in the 90% to 95% range annually is commonly cited as healthy for mid-market and enterprise SaaS, while businesses selling to small and very small customers often run meaningfully lower because those accounts churn more readily. Compare a reading against your own trailing periods and your segment's typical range rather than a single universal figure.
Can the result come out above 100% or below zero?
Not from valid data — retained customers are a subset of the starting count, so the formula is bounded between 0% and 100% whenever end, new, and start are recorded consistently. A result outside that range signals a data problem, most often a win-back customer logged as new when they should count as an original customer who returned, or a new-customer count that was pulled from the wrong period.
References
- U.S. Small Business Administration — Marketing and sales guide
- U.S. Small Business Administration — Grow Your Business guide
Read this first: This instrument shows arithmetic, not advice. Real offers add fees, taxes and terms that vary by lender and place — verify the figures against your actual paperwork before deciding anything.