A discount does not only lower the price on the deal in front of you; it changes who says yes. The buyers already willing to pay full price sit at one end of a demand curve, and every step down the price ladder recruits from further along it — people who valued the product less to begin with. That group churns faster, files more support tickets per dollar of revenue, and expands less, because they bought on price and will leave on price. So average revenue per customer falls for two compounding reasons at once: everyone pays less, and the marginal customer was worth less before a single cancellation is counted.

Consider a subscription software company that cut its entry price by a third to hit a growth number. Signups rose immediately and the growth chart looked exactly as promised. Within three quarters the support queue had grown faster than headcount, the share of accounts cancelling inside their first ninety days had climbed, and the payback period on acquisition spend had stretched past the point where the average account survived. The revenue line was still rising while the business underneath it got worse, and no single monthly report put both facts on the same page. Restoring the old price did not undo it, because the cheap cohort was now the majority of the install base and had begun to set expectations for the roadmap.

The trap is that volume and value get measured by different teams on different clocks. Signups and revenue report weekly and point up; churn, cost to serve, and payback report quarterly or not at all, and point down. A company can run for a year on the fast metric before the slow one arrives to contradict it. The discipline that prevents this is not refusing to discount — it is refusing to count a customer as won until that cohort's payback period is shorter than the time the cohort actually stays.