Reading the distribution, not the mean

Why the shape matters more than the middle

Health distribution is presented as a distribution deliberately. Two books with identical average health can require entirely different work: one is broadly fine with a tight cluster of accounts in serious trouble, the other is uniformly unremarkable with nothing acute. The first needs a small number of urgent, senior interventions. The second needs a change to how the whole book is being run. An average describes both identically and points at neither.

Reading the shape takes seconds once it is a habit. Where is the mass, how long is the tail, and is the tail thickening over time? Those three questions produce a diagnosis; "average health is 68" produces a slide.

Net retention decomposes into four levers

The same discipline applies to the number everyone reports. Net retention is not a primitive quantity. It is the net of those four movements, which is why the useful question is never "why is retention down" but "which of the four is doing it".

Each answer routes to entirely different work, and picking the wrong one costs you a quarter.

• Contraction dominant — customers staying and shrinking; a value problem in live accounts
• Cancellation dominant — customers leaving; the save motion started too late
• Expansion weak, contraction stable — a growth gap wearing a retention number's clothes

Weight everything by revenue

Account counts and revenue rarely agree, and the disagreement is where the risk hides. A book with ten small at-risk accounts and a book with one at-risk account carrying a fifth of the ARR can report the same at-risk percentage. Gross revenue retention is the headline figure precisely because it is weighted: it tells you how much of last year's ARR you kept, which is the question the business is actually asking.

Get in the habit of reading the count and the revenue side by side. When they point in the same direction, the picture is simple. When they diverge, the divergence is the finding: it usually means either a concentration risk sitting in one or two accounts, or a long tail of small accounts churning quietly enough that nobody has raised it.

A worked example: the quarter that looked flat

Net retention comes in at 99%. Reported as one number it invites a shrug: near enough to flat, no obvious action.

Broken into levers it stops being flat. Expansion was strong, contraction was minimal, and a single large cancellation pulled the whole figure down by several points on its own. The book is performing well and one account left. Those are completely different management conversations from "retention is flat", and only one of them is true.

Now weight it by revenue rather than logos and a second thing appears. The canceled account was small in count terms and large in revenue, and there are three more accounts in the same segment with the same profile. The finding is not the 99%. It is that a segment you were not worried about carries concentrated revenue and has just demonstrated it can churn.

None of that is visible in the headline, and all of it is visible in about ninety seconds of looking at the components. That ratio is why the habit is worth forming.