The B2C Churn and Retention Playbook > Chapter 1: Measure

Measure: churn rate, retention rate and what good looks like

The churn rate most growing brands report is flattered by their own growth. New customers pour into the denominator, the percentage drifts down, and a leak that’s getting worse looks like one that’s getting better. We’ve watched retention teams celebrate a falling churn rate in the same quarter they lost more customers than ever.

The formula itself is simple. Churn rate is the number of customers you lost during a period divided by the number you had at the start of it, times 100. The honest version of that number, for any brand with real acquisition volume, is a cohort curve: group customers by when they joined and follow each group over time. The curve answers the one retention question a single percentage can’t, which is where on the timeline the leak actually happens.

This chapter covers the formulas, the trap in each of them, benchmarks by B2C industry so you know roughly where you stand, and a cohort analysis you can rebuild in a spreadsheet by lunch.

How to calculate churn rate

Start with the standard version, the one Amplitude’s churn guide and most finance teams use:

Churn rate = (customers lost during the period / customers at the start of the period) × 100

A subscription brand starts October with 500,000 active subscribers. During the month, 20,000 cancel or fail to renew. Churn is 20,000 / 500,000 = 4.0% for October.

Now the trap. That same month, acquisition brought in 60,000 new subscribers. If someone divides the 20,000 lost by the 560,000 the brand had at any point during the month, churn reads 3.57%. Same leak, better-looking number, and next month it gets “better” again as long as growth continues. Paddle’s churn rate guide shows exactly this effect, with identical customer behavior producing 6.25% one month and 5.87% the next purely because new customers slid into the denominator.

Keep new customers out of the denominator. If you want one number that accounts for them, use the netted variant that Shopify recommends for ecommerce: (customers at start minus customers at end plus new customers acquired) / customers at start. With our numbers that’s (500,000 − 540,000 + 60,000) / 500,000 = 20,000 / 500,000 = 4.0%, which is the number you started with, as it should be.

Two more decisions belong in the same spreadsheet.

First, customer churn versus revenue churn. Losing 4% of customers can mean losing 2% of revenue if the leavers are low spenders, or 7% if your heaviest users are walking. Track both. A rising gap between them tells you which end of the base is leaking.

Second, the period: monthly or annual. It changes everything, and people compare across periods without noticing. 5% monthly churn compounds to roughly 46% of the base gone in a year (0.95 to the twelfth power leaves you 54% of what you started with). 5% annual churn is a different business entirely. State the period every time you state the number.

Keep new customers out of the denominator.

Retention rate: the formula and the trap

Customer retention rate is churn’s mirror image, with one extra step that trips people up. HubSpot’s version is the standard:

Retention rate = ((customers at end of period − new customers acquired during period) / customers at start of period) × 100

Same October: (540,000 − 60,000) / 500,000 = 480,000 / 500,000 = 96%. Retention plus churn equals 100%, as long as both subtract the new customers the same way. The trap is leaving the subtraction out. 540,000 / 500,000 is 108% retention, a number we’ve seen in real board decks, and it means nothing.

The period matters even more here, because “90% retention” sounds excellent and often isn’t. 90% annual retention means you keep nine of every ten customers across a full year, which is strong for any subscription, telecom or insurance business. 90% monthly retention means you keep about 28% of a cohort after twelve months (0.9 to the twelfth). For a bank, that’s a crisis. For a casual mobile game, it might be a good quarter. Which brings us to benchmarks.

What good looks like: churn and retention benchmarks by B2C industry

Benchmarks are a sanity check, and only that. They tell you whether you’re in the normal range for your business model, and nothing about whether your own cohorts are improving, which is the number that matters. Here’s the current range, from sources that publish their own data.

IndustryBenchmarkSource
Subscription, all industries3.27% average monthly churn, of which 2.34% voluntary and 1.25% involuntary (failed payments)Recurly churn rate benchmarks, July 2026
Ecommerce subscriptions4.25% monthly churn (2.87% voluntary, 1.38% involuntary); churn falls as price rises, from 4.29% at $10 to $25 per customer to 2.87% at $100 to $250Recurly, same report
Digital media and entertainment subscriptions4.14% monthly churnRecurly, same report
Premium streaming (SVOD)4.6% weighted average monthly churn across the category in 2025Antenna, Premium SVOD 2025 year in review
Mobile gamesMedian day-1 retention around 22%, with the top 10% of games near 40%; median day-7 just under 4%; median day-30 around 0.7% to 0.8% (16,262 games, 2025 data)GameAnalytics 2026 mobile and PC benchmarks
Fintech appsTraditional banks’ day-30 app retention runs 1.5 to 2 times higher than neobanks’ (UK)AppsFlyer finance app growth release, June 2026
Insurance29% of insurance customers switched companies last year; only 51% of high-value customers say they will definitely renewJ.D. Power 2026 outlook, via Repairer Driven News
Telecom (postpaid mobile)0.92% monthly postpaid phone churn at AT&T, Q3 2025AT&T Q3 2025 results, via Benton Institute

A few things jump out of that table. Involuntary churn is more than a third of all subscription churn in Recurly’s data, and it responds to payment logic (retry schedules, card updaters, a reminder before the card expires), so split it out before you judge your retention program. Price protects: the more a subscriber pays, the less likely they are to leave, which says something about who you should be acquiring. And the gaming numbers show why “is 5% churn good” has no answer without an industry attached. A game that keeps 4% of players at day seven can be a hit. A bank that keeps 4% of new accounts at day seven is closing.

Benchmark against your model, then stop benchmarking. Your October cohort against your July cohort is the comparison that pays.

5% to 46%. Five percent a month is nearly half your base gone in a year.

Cohort analysis for retention

Cohort analysis groups customers by a shared starting point, almost always the month they were acquired, and tracks what share of each group is still active one month later, two months later, and so on. The output is a table that reads like a staircase, and it’s the single most useful retention artifact a B2C team can own.

Here’s an illustrative one for a subscription brand, with numbers in the range we see for consumer subscriptions.

CohortStartedMonth 1Month 2Month 3Month 6Month 12
January100,00072,000 (72%)61,000 (61%)55,000 (55%)46,000 (46%)38,000 (38%)
February105,00075,000 (71%)63,000 (60%)57,000 (54%)47,000 (45%)
March110,00076,000 (69%)63,000 (57%)56,000 (51%)
April120,00080,000 (67%)66,000 (55%)58,000 (48%)

Read it in three passes.

First, the shape. The January cohort loses 28% in month one, another 11 points by month two, 6 more by month three, and then the slope flattens. Most of the leak happens in the first 90 days. After that, the customers who are still around behave like they intend to stay. That shape is common across subscription, ecommerce and gaming, and it tells you where retention spend earns the most: early.

Second, the trend down the rows. Acquisition is growing nicely, from 100,000 to 120,000 a month. Month-one retention is falling just as steadily, from 72% to 67%. Blend those two facts into a single churn rate and it looks flat or improving, because the growing cohorts dilute everything. The cohort table shows a brand that is getting better at buying customers and worse at keeping them, which is usually a sign that the newer acquisition channels or promotions are pulling in people who never meant to stay.

Third, the plateau. Where the curve flattens is your long-term retention floor, and it’s the number that drives lifetime value. If the January cohort settles at 38% after a year, that 38% is what every acquisition dollar is ultimately buying. Raise the floor by a few points and every cohort behind it inherits the gain.

Build the table by acquisition month first, then cut it by acquisition channel, by first promotion used, and by first product bought. One of those cuts will show you a cohort retaining ten points worse than the rest, and that’s your first retention project.

Most of the leak happens in the first 90 days. Measure there first.

Defining churn when nobody cancels

Every formula above assumes you know who left. Subscriptions make that easy: a cancellation is an event with a timestamp. Most B2C businesses don’t get one.

In ecommerce, customers go quiet. The honest definition is a purchase gap longer than the category’s normal rhythm. Shopify’s rule of thumb is to count a customer as churned when they haven’t reordered within two times the average repeat purchase timeline, so a coffee brand with a 30-day average cycle calls a customer churned at 60 days of silence, while a skincare brand on a 75-day cycle waits 150. Set the window per category, write it down, and don’t change it quarter to quarter or your trend lines will lie to you.

Gaming runs on a window of days. Most studios treat a player who hasn’t opened the game in 7, 14 or 30 days as churned, depending on the genre’s session rhythm, and measure retention at day 1, 7 and 30 rather than monthly.

Fintech accounts rarely close. The balance just drifts. The churn event worth measuring is the loss of primary status: direct deposit moves elsewhere, the card stops being used, the balance falls below a floor for 60 or 90 days. Define dormancy explicitly, or your retention rate will count thousands of empty accounts as loyal customers.

Insurance ties churn to the renewal date, which makes it countable but late. Measure non-renewals, and separately track the signals that precede them during the shopping window, because by the renewal date there is nothing left to retain.

Whatever the definition, it feeds everything downstream. The same label you use to count churn is the one a churn prediction model learns from, and the same cohorts are what you’ll compare against a holdout when you prove the lift. Get the measurement right once and the rest of the playbook inherits it.

Define churn once per business line, in writing, and never move the window.

Where we come in

Measurement is spreadsheet work, and you should own it. The step after it, scoring every customer’s churn risk weeks ahead so the measurement stops being a post-mortem, is where we come in, in one sentence: you bring your historical customer data and the churn definition you just wrote down, the Predictive AI Agent reads that data as it is, trains and validates the churn model, and we own the outcome, with scores and reasons live in the CRM, ESP or warehouse your retention team already uses. The churn prediction software page shows what those scores look like once they land.

See who will churn in the next 60 days:

Book a demo

FAQ

How often should we measure churn rate?

Monthly for the headline number, and by acquisition cohort every month as well, so you can see whether newer cohorts retain better or worse than older ones. Businesses with short cycles (gaming, some apps) measure daily or weekly retention at day 1, 7 and 30. Whatever the cadence, keep the definition of a churned customer fixed, or you lose the ability to compare periods.

What is a good monthly churn rate for a subscription business?

Recurly’s July 2026 network data puts the average at 3.27% a month across industries and 4.25% for ecommerce subscriptions, with about a third of it involuntary. Where you should sit inside that range depends on price point and category; subscribers who pay more churn less. Treat the benchmark as a sanity check and your own cohort trend as the target.

What’s the difference between customer churn and revenue churn?

Customer churn counts the people who left as a share of the people you had. Revenue churn counts the recurring revenue they took with them as a share of the revenue you had. The two diverge when leavers are unusually high or low spenders. Tracking both tells you which end of your customer base is leaking.