The Churn Rate Formula, and Why Monthly Times Twelve Is Wrong

A subscription brand starts March with 2,000 customers, signs up 300 more and loses 90. Ask three people on the team for the March churn rate and you can get 4.5 percent, 4.07 percent and 3.25 percent. Each of them divided correctly. Then the finance lead multiplies 4.5 by twelve, puts 54 percent annual churn in the plan, and the real figure is 42.5 percent. The brand has just budgeted for a base that retains a fifth fewer customers than it will.

The churn rate formula is one division, and the argument is always about what goes into it. This guide covers the formula, the five denominators people use for the same month, the conversion from monthly to annual churn, and why a blended churn rate is the wrong number to judge an ad channel on.

The formula, and the version to use by default

Churn rate is the customers you lost in a period divided by the customers you had at the start of it. The churn rate calculator returns it with retention and average lifetime next to it. By hand, it looks like this:

Churn rate = Customers lost in period / Customers at start of period
Churn rate = 90 / 2,000 = 4.5%

Retention rate = 1 − Churn rate = 1 − 0.045 = 95.5%

Revenue churn = Recurring revenue lost / Recurring revenue at start

Average lifetime = 1 / Churn rate = 1 / 0.045 = 22.2 months

Revenue churn runs on the same shape but weights each customer by what they pay, so losing ten small plans moves it less than losing one large one. Without a subscription, the word "lost" needs a rule you write down first. A common one for a store is no order in a window set from your normal gap between purchases, say 180 days. Whatever the rule is, it has to stay fixed, or the trend you read is the rule changing.

The last line, lifetime as one over churn, is a shortcut with real limits. They are covered in the customer lifetime value formula, so treat 22.2 months here as the arithmetic and not as a forecast.

Five denominators for the same month

Go back to March. The business began with 2,000 customers and added 300. It lost 90, but 25 of those had joined during March, so 65 came from the starting base. It ended the month with 2,210. From those five facts you can write five churn rates, and all of them appear in real reports.

The spread runs from 3.25 to 4.50 percent on the same 2,000 customers, and the top figure is 38 percent higher than the bottom one. Dividing by ending customers flatters a growing business, because it spreads the losses over 300 people who were not around for most of the month. The 4.50 percent line has the opposite flaw. Its numerator includes 25 losses from customers who are not in its denominator, so it measures neither the base nor the newcomers cleanly.

Starting customers only is the cleanest read of how an established base behaves, and it is the one to report by default. The last row is a different question, and a more useful one for anyone buying customers. Newcomers left at 8.33 percent, more than twice the rate of the base. Blend them together and that difference disappears into a single figure.

Monthly times twelve overstates annual churn

Churn applies to whoever is left, so it compounds. The customers you lose in month two come out of a base already thinned by month one. To annualise, raise retention to the twelfth power:

Annual retention = (1 − Monthly churn) ^ 12
Annual retention = 0.955 ^ 12 = 57.55%
Annual churn = 1 − 0.5755 = 42.45%

Monthly x 12 = 4.5% x 12 = 54.0%

Reverse: Monthly churn = 1 − (1 − Annual churn) ^ (1 / 12)
Monthly churn for 40% annual = 1 − 0.60 ^ (1 / 12) = 4.17%

Out of 1,000 customers, the compounded figure leaves 575 after a year and the multiplied one leaves 460. The multiplied version is wrong in a predictable direction, and the error grows with the churn rate. At 3.25 percent a month it reads 39.0 percent against a true 32.7. The reverse direction trips people too. Dividing a 40 percent annual target by twelve gives 3.33 percent a month, which would compound to 33.4 percent, not 40.

For a media buyer this matters because annual churn is what a lifetime value figure leans on, and lifetime value sets the bid. A plan built on 54 percent assumes 460 survivors where the business will keep 575, so it undervalues every customer and caps spend lower than the account can bear. Constant churn is itself a simplification, since churn is often heaviest in the first months. Use the exponent to convert, and use cohort data to check whether the rate holds.

Blended churn cannot rank your channels

Split March's 300 new customers by where they came from. Channel A brought 150 and lost 6, which is 4.00 percent in the first month. Channel B brought 150 and lost 19, which is 12.67 percent. Together that is the same 25 losses and the same 8.33 percent as before, and the blended view tells you nothing about which channel to fund.

Now add price. Say a customer pays $43 a month at 42 percent contribution, so each paid month returns $18.06. Channel A costs $38 per new customer and channel B costs $30, which makes B 21 percent cheaper. Hold each channel at its first-month churn for six months, an assumption that probably overstates B, and the expected paid months are 5.43 for A and 4.39 for B. Six-month contribution comes to $98.08 against $79.32.

Channel A: $38 / $98.08 = $0.39 of CPA per $1 of six-month contribution
Channel B: $30 / $79.32 = $0.38 of CPA per $1 of six-month contribution

The 21 percent price advantage shrinks to about 2 percent once churn is priced in, and a slightly worse retention curve would flip it. A buyer looking only at CPA would shift budget to B with confidence. The customer acquisition cost calculator and the CAC payback period are where the churn gap turns into a spending decision.

What to do with this

Take last month's lost customers, tag each one with the channel that acquired it, and divide by that channel's new customers. That gives you first-month churn by channel. Put those rates, not the blended one, into the LTV calculator and compare the result with each channel's CPA before you move any budget.

Figures here are illustrative. They come from one worked example built so that every intermediate reproduces the printed result, not from a survey, and the channel comparison assumes churn stays constant for six months. Real subscription and repeat-purchase businesses differ widely, so benchmark against your own cohorts, not against someone else's headline number.

Related reading: LTV to CAC ratio and CAC vs CPA.