Churn rate
Churn rate explained: how it is calculated, and why a dealership never sees it happen
Churn rate is the percentage of customers who stop doing business with you over a defined period, calculated as customers lost during the period divided by customers at the start of it. It is the mirror image of retention, and the version that gets attention because a loss reads more urgently than a hold.
Also called attrition rate, customer churn, or lapse rate. Where money rather than headcount is the unit, it becomes revenue churn, and the two rarely agree.
How is churn rate calculated?
Churn rate = customers lost during the period ÷ customers at the start of the period × 100
The arithmetic is trivial. The definitions are not:
- What counts as a customer. A dealership has no subscription list, so you have to invent one. The usual construction is anyone with at least one paying visit in the previous twelve months, which means your customer base changes every time you change the window.
- What counts as lost. Nobody cancels. They simply do not come back, so a customer is only classified as churned once enough time has passed, and that judgement is always retrospective.
- Whether new customers are in the denominator. Adding customers acquired during the period dilutes the rate and makes a growing business look loyal. Measure churn on the opening base only.
Churn rate: a worked example
A dealership starts the year with 4,200 active service customers, defined as at least one paying visit in the prior twelve months. Over the year, 780 of them never return.
Customer churn is 780 ÷ 4,200, or 18.6 percent. Retention is 81.4 percent. Same fact, two framings, and most boards react differently to the first one.
Now weight it by money. Say the 780 who left averaged 720 dollars of annual spend while the 3,420 who stayed averaged 480. The opening base was worth 2.2 million dollars a year; the leavers took 561,600 of it. Revenue churn is 25.5 percent against a customer churn of 18.6, because the customers who left were the ones with the older cars and the bigger bills. Report the headline number alone and you understate the hole by seven points.
Gross churn and net churn
Gross churn counts what left. Net churn subtracts the growth that came from customers who stayed and spent more, which in a workshop is real: an ageing parc means the same customer generates more each year.
A store can run 18 percent gross churn and near-zero net revenue churn while its customer base quietly shrinks by a fifth a year. The revenue line looks stable right up until the base is too small to hold it. This is the single most common way churn gets hidden in fixed operations.
How churn rate gets misread
- No usage segment. A customer driving 4,000 miles a year is not due yet. Classify them as churned and you will chase people who never left while missing the ones who did.
- Recall and warranty visits counted as activity. A manufacturer campaign drags in customers who had already gone elsewhere for everything they pay for. The churn number improves; nothing else does.
- Snapshot instead of cohort. A single annual figure buries the shape of the loss. Churn in year one behaves nothing like churn in year four, when the warranty ends and the independent becomes a live option.
- Confused with defection. Churn says a customer stopped coming to you. Defection says where they went. Urban Science, working from registration data, reported an average dealership defection rate of 20 percent in 2025 and found that 74 percent of dealers lack full visibility into where those customers land.
- Read without the value of what left. Churn and customer lifetime value only mean something together. Losing 200 low-value customers and losing 200 high-value ones produce the same rate and completely different years.
- Targeted without a window. "Get churn under 15 percent" is unactionable until someone writes down the measurement window, and it is trivially achievable by lengthening it.
What churn tells you that retention does not
Mechanically, nothing. They sum to 100. The difference is what each one invites you to look at.
Service retention points at the customers you kept and tends to produce campaigns aimed at people who were coming anyway. Churn points at a specific list of names who did not come back, which is a harder document to read and a far more useful one. Pull fifty of them, look at the last repair order each one had, and the pattern is usually visible inside an hour.
What you tend to find is that churn is lagging by design. It confirms a loss that was decided a year earlier, at a write-up where a customer did not understand what was being recommended. CSI and declined work move first, and they tell you what churn is going to say next year.
