Customer success metrics that actually matter

Gross versus net retention

Gross revenue retention asks a simple question: of the recurring revenue you started the period with, how much did you keep? It counts churn and downgrades, and it stops at one hundred percent. Nothing you upsell can hide a loss.

Net revenue retention asks a different question: after churn, downgrades, expansion and price increases, how much of that starting revenue do you now have? It can go well above one hundred percent, which is why it appears on every board slide.

The two are not competing versions of the truth. Gross retention tells you how well you hold onto what you sold. Net retention tells you how the book of business is growing. Teams get into trouble when they report only the second one.

Reading the two together

A healthy pattern is strong gross retention with net retention comfortably above it: you are keeping customers and they are buying more. A warning pattern is net retention above one hundred percent while gross retention slides, because a few large expansions are covering a steady leak of smaller accounts. That looks like growth for two or three quarters and then stops looking like anything good at all.

Read both next to logo retention, which counts customers rather than revenue. If you are keeping revenue but losing logos, you are becoming dependent on a shrinking number of accounts, and your renewal risk is concentrating even as the revenue chart flatters you.

One practical habit: report gross retention, net retention and logo retention in the same place every month, always in that order. It removes the temptation to lead with whichever one looks best.

Churn signals that arrive early

Cancellations are rarely a surprise to anyone paying attention. They are the end of a pattern that started months earlier. The signals our panelists trust most are behavioural rather than sentimental:

• The champion goes quiet. Not hostile, just slower to reply and no longer forwarding your emails internally.
• Usage narrows. Total logins hold up but the number of distinct active users or teams drops. One person keeping the account alive is not adoption.
• Meetings get delegated downward. The person who signed stops attending reviews and sends someone with no budget authority.
• Nobody can restate the goal. When you ask what success looks like and get a different answer than you did at the handoff, the outcome has drifted.
• Support tickets shift from how-to to it-does-not-work. Frustration replaces curiosity.

A high satisfaction score sits comfortably alongside every one of those. That is why sentiment on its own is a poor early warning system.

Health scores worth acting on

A health score earns its place when a red account triggers a specific action by a named person within a defined time. If red just means a coloured cell in a spreadsheet, the score is reporting, not management.

Three rules keep scores useful. Use few inputs, because a score built from fifteen weighted factors cannot be explained to the person expected to act on it. Weight behaviour above opinion, so adoption breadth and progress toward the goal count for more than a survey response. And write down what each colour obliges someone to do.

It is also worth reviewing the score against reality every quarter. Pull the accounts that churned and check what colour they were ninety days out. If they were green, the inputs are wrong, and no amount of dashboard work will fix that.

The one measure per account

Underneath all of this sits the measure most teams never define: progress toward the outcome the customer actually bought. It is specific to the account, agreed with the customer, and stated in their language rather than yours.

Capture it at the handoff, review it out loud in every business review, and use it to settle arguments about whether an account is doing well. When that measure is moving, the retention numbers usually follow. When nobody can name it, every other metric on this page is guesswork with a decimal point.