Margin Drift
Gross margin held at 72% for four quarters, then 64% by the fifth. Nobody could name when it moved — a hundred small concessions, none visible alone.
At one company, gross margin held at 72% for four straight quarters, then dropped to 64% in the fifth. The leadership team noticed during a budget review and could not say when the slide had started. There had been no quarter of collapse. There had been four quarters of small movement, none big enough to trip a flag.
The 8-point drop was a hundred small concessions. A drift toward larger customers carried more support overhead. A new payment processor ran twenty basis points higher on fees. Customer-specific feature work got booked into CS instead of as services with their own margin profile. The discount field widened by an average of three points. The hosting bill grew twenty percent a year against revenue growing thirty.
None of these moved the headline number enough to trigger a look. All of them pushed the same direction. Four quarters in, the aggregate was the 8-point drop the team finally saw.
This is how margin moves at growth-stage companies. Not in events. In drift. It stays invisible at the aggregate until enough time passes that the new level is the baseline. By then the company has been running against a margin assumption that is wrong, the cash-flow projections are built on it, and the runway is shorter than the dashboard claims. The unit-economics model nobody opens is where that wrong assumption hides.
What catches it is decomposition. Margin reported as one number is margin reported in the format most likely to hide drift. Break it into four to six components: direct COGS, support cost per customer, discount rate, contract mix, payment processing, hosting. Each gets a named owner graded on the trend, not the level. That is the same visibility that turns a standing discount back into a decision instead of a default.
Most companies skip this because the decomposition is work and the headline looks fine until it doesn't. The visibility is the whole point. A discount rate creeping half a point a quarter is correctable at half a point. The same rate compounded for two years is the price of the business. And when COGS drifts, the trap is assuming it moved with volume when most of it was fixed cost all along.
Decomposed margin lets the company correct early. Aggregate margin lets the company learn, a year late, what the drift already cost.