Healthy Customers Are Successful

Every Customer Success organization I have led has had a health score, and I have built the same one three times, for companies across the scale and service spectrum, because the thing that makes it work is not the data. It is a rule. In 2018, running Customer Success for a hospitality-software company, I wrote the rule down for the team and taught it in an hour, and that presentation and that lesson plan are now on the shelf as The Customer Health Score: a manual, and a workbook that is the instrument.

Four questions that can each fail

A health score that is a weighted number is a score that can be argued with. Three good signals average away one bad one, and the bad one is the churn. So the instrument asks four questions — is the customer using what they bought, are they getting the results they bought it for, what is it like to be served by us, what are they saying about us — from four sources: the product, the CSM, the support queue and the invoice, the survey. Each answer is Green, Yellow or Red. And the rule is deliberately unforgiving: any component in a fail state makes the account Red, and names the play that answers it. A customer who has stopped logging in is a fact about the renewal on its own, however the other three look. The instrument says so without arithmetic.

There is a number beside the color — a resemblance score, zero to a hundred — and it exists for trend and ranking only. It never overrides the color. A Yellow drifting from 80 to 55 is visible before it fails; that is all the number is for.

Resemble whom?

Not all customers are measured equally. A customer at each stage of the journey is modeling the behavior that leads to a renewal, an expansion, or a churn, and that behavior can be scored — against the right thing. The right thing is not an industry benchmark, which describes someone else's customers. A company generally knows what a good customer from its own pool looks like. Take the accounts that renewed and grew, read their numbers at each stage, and those are the thresholds. The question the score asks is then always the same: how much does this customer, now, resemble a Green customer at this stage?

The stages matter because of where churn originates. Most of it leaves the path during onboarding and in the ninety days after first value, which is why health is most predictive there, why the Hypercare profile in the workbook is the tightest, and why a Hypercare account that is not Green gets its play this week and not this quarter. Hypercare is a touch, not a tier: whatever the account pays, a new customer gets a human while the behavior that predicts its renewal is being formed.

What we do with the answer

An instrument measures and then says what to do, or it is a report. Each failed component names one of the Operating Model's four plays — not engaged, an adoption workshop; cases aging or an invoice unpaid, a success initiative; a detractor, the feedback cycle; a review overdue, the business review — and the scorecard writes the play on the row. A CSM's week is the play-due column sorted by color. A manager's Monday is the Book tab: Reds first, then Hypercare, then the plays due. The executive team's forecast is the same tab, and nobody has to ask the account owner what's happening.

When I first scored a whole book this way and read the Reds one by one, three causes accounted for nearly all of them: inconsistent adoption, operations that had slipped, and a play that the score had named and nobody had run. All three were controllable. That is the most hopeful thing a health score ever told me, and it is the reason the manual ends with the lesson plan: a score the team does not understand is a column they ignore.

Fourteen pages, free to use with attribution, with the scorecard workbook beside it — example thresholds in blue, an invented book of twelve restaurants so you can see the rule work, and a Book tab for Monday. The company and its platform are gone from both; the rule is intact.