We've been sitting on this data for a while, trying to make sure we were interpreting it correctly before putting it out into the world. The signal is clear enough now that we're confident sharing it.
Across more than 120,000 customer visits logged in SafiTrack over the past eighteen months, we found a consistent pattern: accounts that churned almost always showed a measurable drop in visit frequency before any other leading indicator moved. The average lead time was 47 days.
What the data actually shows
We segmented churned accounts and compared their visit frequency in the 90 days before cancellation against their historical baseline. Two patterns emerged clearly.
First, the gradual drift: accounts where visit frequency slowly declined over two or three months before churn. These customers weren't vocal. They didn't submit support tickets or call to complain. They just became gradually less visited. Often because the rep had deprioritised them as shinier opportunities took up more calendar space.
Second, the sudden drop: accounts where a previously healthy visit cadence fell off a cliff, usually within a four-week window. These often correlated with rep turnover, territory reorganisations, or periods where the rep was pulled into a large new account pursuit.
Why this metric matters more than NPS or support tickets
Customer health scores typically rely on product usage data, NPS surveys, and support interaction volume. These are useful but they have a lag problem: by the time a customer is submitting complaints or showing poor NPS scores, they're often already mentally out the door.
Visit frequency is different because it's a measure of your behaviour, not theirs. You control it. And because you control it, a decline is an early signal that something has shifted (a de-prioritisation decision, a capacity constraint, a relationship that's been taken for granted) before the customer has had the chance to respond to that neglect.
How to act on this signal
The practical implication is straightforward. For any customer relationship with a defined visit cadence, you need a system that alerts you when that cadence is breaking down. Before it's been broken for two months.
In SafiTrack, this is what the visit health alert does: it compares actual visit frequency against each account's target cadence and flags accounts that are falling behind. The idea is to surface the "quiet drift" accounts (the ones that are easy to neglect precisely because they're not shouting at you) before the drift becomes a problem.
The counterintuitive finding on account size
We expected the pattern to be stronger for smaller accounts, which are often more dependent on the relationship quality and less locked in by contractual complexity. The data said the opposite. The 47-day early warning was actually more pronounced in mid-market accounts. The ones large enough to have multiple internal stakeholders but not large enough to have a dedicated customer success manager.
Our theory: with enterprise accounts, relationship health is tracked obsessively because the stakes are visible. With SMBs, the churn decision often happens quickly with limited warning. It's the mid-market accounts, caught in the middle, where the gradual drift pattern is most predictable and most preventable.
A caveat on causality
We should be clear: visit frequency decline is a correlate of churn, not necessarily a cause. It's possible (perhaps likely) that in some cases, both the declining visits and the eventual churn are downstream effects of some third factor: a shift in the customer's business priorities, a budget cycle, a new decision-maker who doesn't value the relationship the same way.
That doesn't make the metric less useful. Even if declining visits are a symptom rather than a cause, they're a symptom you can observe and respond to, which puts you in a far better position than waiting for the cancellation notice.