Logo vs. revenue churn. The two churns, and what each tells you.

Two numbers, two stories

The reason to track both is that they fail in opposite directions. You can lose many small accounts and still hold onto your revenue — logo churn spikes, revenue churn barely moves. Or you can lose a few big accounts — the salmon, the tuna, the whales — while your logo churn looks perfectly healthy, because the customer count hardly budged. One headline number can't show you both, which is exactly why you keep two.

For us, revenue churn was the more important of the two, and the logic is almost tautological once you say it out loud: MRR dollars become, and are, revenue. Losing ten minnows and losing one whale can read identically on a logo-churn chart and mean completely different things to the business. The dollar number doesn't let you fool yourself about which one happened.

Reading the combinations

The value is in holding the two side by side and reading the gap. Each combination points to a different diagnosis and a different action.

Logo churn bad, revenue churn fine. You're shedding customers but keeping dollars — which means the customers leaving are your small ones. That's a signal to go review your ICP. Which customers are actually leaving, and why? How do you tighten your go-to-market to focus on the most valuable ones? The question we were always asking, in our own shorthand: how do we land fewer minnows and trout, and catch more salmon, tuna, and whales?

Revenue churn bad, logo churn fine. The inverse, and the dangerous one: you're keeping your customer count up while losing your high-value accounts. For us this was almost never the case — and that fact is itself the lesson. Our smaller customers, the lower-MRR minnows and trout, always churned at a higher rate than our big accounts. That asymmetry is the whole reason revenue churn was the better metric for us: the dollars were stickier than the logos, because the valuable customers stayed and the marginal ones left.

When a whale moves the number

Both directions happened to us, and the revenue-churn spike is the one that gets your attention — a single big account contracting or cancelling can blow up the monthly number while logo churn shrugs. We didn't have a formula for this. When we saw a big spike, we unpacked it and talked it through as an issue: what happened, was it a contraction or a full cancellation, and could we save it? If the account was planning to leave or had already given notice, the save play kicked in.

That's the honest version of "managing" big-account volatility at SMB scale: not a model that smooths it out, but a weekly habit of looking a spike in the eye and deciding whether it's a problem to solve or a fact to accept. The number flags it; the conversation resolves it.

You want some churn

Here's the counterintuitive part, and it took us a while to internalize. Early on you assume both churn numbers should be driven as low as possible — zero is the dream. It isn't.

This is why a low revenue churn alongside a higher logo churn was a healthy signature for us, not an alarm. The two numbers together were telling us the base was self-selecting: bad-fit accounts leaving, good-fit accounts compounding. If we'd panicked at the logo number and chased those small accounts with retention effort, we'd have spent the team's energy holding onto exactly the customers who cost the most to serve and left the fastest.

The headlines of your customer base

The bigger payoff of looking at both numbers isn't either number. It's the question they force you to ask about your base. Once you can see that dollars and logos churn differently, the natural next move is to ask which customers stay and which leave — which industries, what revenue size, how much they're worth to you, where you sourced them from.

That's the through-line from a churn dashboard to a strategy. Logo and revenue churn are where the story first becomes visible; NRR is where it nets out into one verdict; and a cohort view of who stays, by segment, is where it turns into a go-to-market decision. The two churns are step one: stop reading a single blended number, and start reading the gap.