The math that looks obvious and usually isn’t
Say you have 200 customers paying $30 a month. That’s $6,000 in MRR. Raise the price to $35 and, on paper, you’re now looking at $7,000 a month for doing nothing except changing a number in a settings panel. That’s the pitch you see in every “just raise your prices” post.
What that math skips is the part where some of those 200 customers leave. A minority, usually, though enough that the real number you land on is somewhere between $6,000 and $7,000, and you don’t know where until it happens. I’ve run price increases on small subscription products and the gap between “what the spreadsheet promised” and “what actually landed in the account” has been real every single time. Sometimes small, sometimes not.
This isn’t a reason to never raise prices. It’s a reason to measure the thing you’re actually changing, which is churn, not just the sticker price.
Churn is the tax on every price increase
Churn, in plain terms, is the percentage of customers who cancel in a given period. If you have 200 customers and 6 cancel in a month, your monthly churn is 3%. It’s the simplest number in your business and also the easiest to misread, because a single month of churn tells you almost nothing about why people left.
A price increase adds a new reason to churn that wasn’t there before. Some customers who were on the fence about your product anyway will use the price bump as the excuse to finally cancel. Others who were genuinely happy will stay because the extra $5 a month isn’t worth the hassle of switching. The mistake is assuming you know in advance which group is bigger. You don’t. That’s what cohort tracking is for.
Watch the cohort, not the aggregate
A cohort is just a group of customers who share a starting point, usually the month they signed up. Cohort analysis means you track that group over time instead of blending everyone together into one churn number.
When you raise prices, the useful thing to build is a before-and-after cohort comparison: take the customers who were already subscribed before the price change, and watch their monthly retention curve starting the month the new price took effect. Compare that curve to what the same tenure group looked like in the months before you touched pricing. If churn in month one after the increase is meaningfully higher than churn in month one for equivalent cohorts before the change, that’s the price increase talking. If it’s roughly flat, the increase probably wasn’t the trigger for whatever churn you do see.
The trap is looking at total cancellations that month and calling it a verdict. Total cancellations include people who were always going to leave for unrelated reasons: they stopped needing the product, they had a bad support experience, their business closed. Blending that in with price-driven churn tells you nothing you can act on.
Grandfathering as a built-in experiment
One thing that’s saved me from guessing is grandfathering existing customers at their old price while charging new signups the higher one. It’s a defensible move on its own, since surprise-billing someone is a poor way to treat a customer, and it also hands you a cleaner read on the question you actually care about, which is: does the new price change who signs up and how long they stick around, without any noise from existing customers reacting to a change they didn’t ask for.
If you want to know what the increase itself does to your current base, you eventually have to test it on some of them, not just on new signups. A common way to do that without exposing your whole customer list to it is a holdout: raise the price for a segment (new signups, or a subset of an existing plan tier) and leave a comparable group at the old price, then compare churn and revenue per customer across the two. This is the same logic as an A/B test, just applied to pricing instead of a landing page. The honest caveat: pricing tests take longer to read than a button color test, because churn shows up over months, not days. Don’t call a result from three weeks of data.
CLV tells you if the trade was worth it, eventually
Customer lifetime value, CLV, is a rough estimate of how much revenue you’ll collect from a customer over the whole time they stay subscribed. The simplest version is average monthly revenue per customer divided by monthly churn rate. If a customer pays $30 a month and churns at 3% a month, the back-of-envelope CLV is $30 divided by 0.03, which is $1,000.
Raise the price to $35 and drop nothing else, and CLV goes up proportionally, to about $1,167. But if churn rises from 3% to 4% because of the price change, CLV actually falls, to $875. Same price increase, opposite direction, depending entirely on what happens to churn. This is why “we raised prices” is not itself a result. The result is whatever CLV and net revenue retention do after, and you can’t know that on day one.
Net revenue retention (how much revenue you keep from an existing cohort of customers over a year, including the effect of price changes, upgrades, downgrades, and cancellations) is the number that eventually tells you whether a price increase paid off. It takes longer to read than churn does, because you need enough time for the churn effect to fully show up, not just the first wave of people who cancel immediately.
What the numbers can’t tell you
None of this replaces judgment. A cohort chart can show you that churn ticked up after a price change, but it can’t tell you whether that’s because the new price was too high, because the increase was communicated badly, because a competitor dropped their price the same month, or because of something in the product itself that had nothing to do with pricing. Correlation in a cohort curve is a strong hint, not a verdict.
Exit surveys, the short “why are you cancelling” prompt on your cancellation flow, are one of the few ways to get at the actual reason instead of inferring it from a chart. They’re biased (people don’t always tell the truth, and plenty skip the survey entirely), but a spike in “too expensive” answers right after a price change is a much more direct signal than churn percentage alone.
I’d also be careful about small sample sizes. If you have 40 customers total, a price increase that costs you 2 of them looks like a 5% churn spike, and that’s a genuinely different conversation than the same 5% spike measured across 2,000 customers, where you have enough volume to trust the pattern rather than the noise.
What I actually do before touching a price
Before I raise a price on anything I run, I pull the churn rate for the trailing three months as a baseline, I decide in advance whether existing customers get grandfathered or not, and I set a date to actually look at the cohort data again, usually 60 to 90 days out, rather than checking obsessively in the first week when the numbers are too thin to mean anything. I also don’t treat the first uptick in cancellations as proof the increase failed. Some of that is always noise. The cohort comparison over a couple of months is what separates a real reaction from normal month-to-month churn wobble.
Raising prices is a legitimate lever. It’s just not a free one, and the only way to know what it cost you is to measure the churn it caused against the revenue it added, honestly, with real data from your own customers rather than a spreadsheet assumption.
If you want more of this kind of plain-English breakdown of the metrics that actually run a small business, you can find the rest of what we cover here.
Get new guides and videos first — join the Telegram channel.