Churn and revenue retention

Churn is the loss of a customer or of the money that customer was paying, and revenue retention is the share of an earlier period's recurring revenue that the same customers still produce today.

Retention curves answered a question about people. A cohort of a thousand signups falls to a floor of 22 per cent, and the floor says that roughly a fifth of every intake keeps opening the product. Every figure on that curve is a headcount, and a headcount treats a customer paying ninety pounds a month and a customer paying nine thousand as one person each.

The vocabulary for counting the money came from subscription software finance and reached product teams through the board pack. A lender, an investor and a finance director all reach for gross and net revenue retention early, because the two figures decide how much growth the company has to buy each year to stand still.

A product manager meets the arithmetic the first time a renewal forecast disagrees with a usage chart. Logins are healthy, the retention curve is flat, and the revenue plan is short by a quarter. Both pictures are correct, and the gap between them is the subject of this page.

The sections below separate usage retention from revenue retention, then set out churn counted in customers beside churn counted in money. The formulas for gross and net revenue retention follow, worked through on two companies whose figures point in opposite directions. The page then takes what net retention above one hundred per cent does to a growth plan, the revenue a signup cohort still produces after a year, and the definitions any quoted figure depends on.

Usage retention and revenue retention answer different questions

A usage retention curve asks whether somebody came back. A revenue retention figure asks whether the money came back. Four customers show how far apart the two answers can sit.

1. Active and paying the same. The ordinary case, retained on both measures.
2. Active and paying much less. A customer who logs in daily and cuts forty seats to eight is fully retained on the usage curve and has removed 80 per cent of what they paid.
3. Silent and still paying. An enterprise customer whose licence renews automatically while nobody opens the product is gone from the usage curve and present in full on the revenue one.
4. Gone from both. The case everybody measures, and the smallest of the four in most companies.

The second and third of those four are where the two views separate, and both are common in software sold to organisations. A seat reduction never appears on a usage curve, because the people who kept their seats keep logging in. A silent renewal never appears as revenue churn, because the invoice was paid.

Free products have only the first view available to them. There is no invoice, so there is no revenue to retain, which is the reason usage retention exists as a measure at all. A subscription product has both views and needs both, since either one alone describes half of what happened.

Churn counted in customers and churn counted in revenue

Counting the loss in customers is the simpler of the two and it goes by the name logo churn, from the practice of drawing customer logos on a slide. Logo churn is the share of the customers present at the start of a period who are gone by the end of it. Logo retention is the same figure the other way up.

Counting the loss in money needs three separate quantities. Revenue leaves by two routes and arrives back by a third.

1. Churned revenue. What the customers who left were paying.
2. Contraction. What the customers who stayed gave up, through fewer seats, a cheaper plan or a negotiated discount.
3. Expansion. What the customers who stayed added, through more seats, a higher plan, a new module or a price rise.

Those three quantities are what make the money count differ from the headcount in both directions. Ten customers leaving can matter less than one customer halving a licence, and a year in which nobody left at all can still lose a tenth of the revenue base to contraction.

Gross revenue retention and net revenue retention

The three quantities combine into two published figures, and the difference between them is one term.

Gross revenue retention takes the recurring revenue at the start of the period, subtracts what the leavers were paying and subtracts contraction, then divides by the starting figure. It measures leakage on its own, so it cannot pass 100 per cent whatever the company does.

Net revenue retention performs the same calculation with expansion added back before the division. It measures what the same set of customers is worth now against what they were worth a year ago, so it can land anywhere from nothing to well above 100 per cent.

Two properties of the pair are worth holding on to. Gross retention is the honest picture of loss, because no amount of upselling can hide inside it. Net retention is the picture of the business, because it says whether an existing customer base grows or shrinks on its own. A company quoting one figure and never the other has usually chosen the flattering one. Both belong in the same report.

Two cases where the counts point in opposite directions

The gap between the three measures is easiest to see on two companies that start the year identically. Each begins with 400 customers and £200,000 of monthly recurring revenue, which is an average of £500 a customer.

Every figure in the table is arithmetic anybody can repeat. Company one keeps 340 of its 400 customers, so logo retention is 85.0 per cent. Its gross revenue retention is £200,000 less £12,000 and less £6,000, all over £200,000, which is 91.0 per cent. Adding the £38,000 of expansion gives £220,000 over £200,000, or 110.0 per cent net.

Company two keeps 388 of its 400 customers, so logo retention is 97.0 per cent. Its gross revenue retention is £200,000 less £38,000 and less £4,000, all over £200,000, which is 79.0 per cent. Adding £10,000 of expansion gives £168,000 over £200,000, or 84.0 per cent net.

The two companies therefore rank in opposite orders depending on which measure is read. Company one lost 15 per cent of its customers and ended the year with more money from the same base. The sixty customers who left averaged £200 a month against a company average of £500, so they were the small ones, and the customers who stayed grew. Company two lost twelve customers, two of whom were among its largest accounts, and those twelve took £38,000 with them. Ninety seven per cent logo retention conceals a business that lost a fifth of its revenue base in a year.

Two working conclusions follow. Logo churn is the measure to watch for a product sold at a similar price to everybody, such as a consumer subscription or a single plan sold to small firms. Revenue churn is the measure to watch as soon as the largest customer pays twenty times what the smallest pays, which is true of almost every product sold to organisations. Most products need both.

What net revenue retention above one hundred per cent changes

Net revenue retention above one hundred per cent changes the arithmetic of growth, because the existing base grows without anybody selling anything.

Company one starts at £200,000 a month with 110 per cent net retention. Left alone, that base is worth £220,000 a year later. Reaching a target of 30 per cent growth means £260,000, so new sales have to supply £40,000, which is 20 per cent of the opening base.

Company two starts at the same £200,000 with 84 per cent net retention. Left alone, that base is worth £168,000 a year later. Standing still at £200,000 already needs £32,000 of new sales, or 16 per cent of the opening base, and the same 30 per cent growth target needs £92,000, or 46 per cent. Company two has to sell more than twice as much as company one to reach the same place.

That difference compounds into everything else the company does. A sales team sized for 20 per cent costs a fraction of one sized for 46 per cent, so net retention sets the cost of growth long before anybody argues about pipeline.

The published benchmarks put those two companies at opposite ends of the distribution. SaaS Capital ran its fourteenth annual survey in the first quarter of 2025 and published the retention findings on 18 September 2025, drawing on more than a thousand private business software companies. The median net revenue retention was 101 per cent and the median gross revenue retention was 91 per cent. The median growth rate across the whole sample was 24 per cent, companies reporting net retention of at least 110 per cent grew faster than that median, and companies below 100 per cent grew more slowly. Companies in the band between 110 and 120 per cent reported a median growth rate nine percentage points higher than those in the band between 100 and 110.

Company one therefore sits at the top of the distribution on net retention while sitting at the median on gross retention. Company two sits at the median on logo retention and near the bottom on both revenue measures. A single number from any of the three would have ranked them wrongly.

The revenue a signup cohort still produces

Annual figures describe a whole company and hide the direction of travel, which is the same objection cohort analysis answered for usage. A revenue cohort curve repairs it by following one intake forward in money.

The method is the one already in use for retention curves, with the denominator changed. Group customers by the month they first paid, then plot what that group pays each month afterwards as a share of what it paid in its first month. One cohort of 120 customers at a workflow product, paying £24,000 between them in January, shows what the picture looks like when expansion is working.

The two lines separate at once and finish pointing in opposite directions. Forty per cent of the customers are gone by month twelve, and the cohort pays 15 per cent more than it did on the day it arrived. Average revenue per remaining customer moved from £200 to £383, because the customers who stayed added seats while the customers who left were the ones paying least.

A picture of this shape carries a warning worth stating. The rising line depends on a shrinking group of customers spending more, so it is fragile in a way the headcount curve is not. One large account leaving in month thirteen removes a share of the revenue that no number of small customers replaces, and the line falls further in one month than it climbed in twelve.

The definitions any quoted figure depends on

Comparing one company's revenue retention against another's depends on both using the same definition, and the definition has more choices inside it than the single number suggests. Four of them change the answer. None is usually published.

1. The population. SaaS Capital's formula takes the recurring revenue in December 2024 from customers who were already customers in December 2023, divided by total recurring revenue in December 2023. Customers who arrived during the year are excluded from both halves, since including them measures new sales.
2. The treatment of price rises. A company that raised list prices by five per cent has expansion of five per cent with no customer buying anything.
3. The capping rule. Gross retention is calculated by limiting each customer's ending revenue to what they started with, which is what stops one expanding account masking several shrinking ones.
4. The period. A monthly figure annualised and a figure measured across twelve months are different quantities, and the first is smoother.

A team reporting a figure records which of those four choices it made, in the same document as the measure itself, because the figure is meaningless to anybody who cannot reconstruct it. The strategy course treats the money side of this under product economics, where lifetime value and the cost of winning a customer set what a company can afford to spend. Revenue retention is the measurement underneath that ratio, since a customer's lifetime value is whatever the retention figure says it is.

Products sold without a long contract need one more decision. A monthly subscription churns on whatever date the customer picks, and a free product produces no churn event at all, so the team defines a lapse rule saying how long somebody must be absent before they count as gone. That rule is a choice with the same properties as a session timeout, meaning it changes every figure downstream of it and belongs in the tracking plan with a date.

None of these measures says why anybody left or why anybody grew. A gross revenue retention figure of 79 per cent reports a fifth of the base gone and names nobody, and a net figure of 110 per cent reports expansion without saying what the expanding customers had in common. The earliest place to look for either answer is what those customers did in their first days with the product, which is where the next page goes.