Why the absence of complaints is not evidence of satisfaction

A service reports eleven complaints across a quarter of forty thousand transactions and the board reads it as a quality result. In the same quarter, three per cent of regular customers stopped appearing. Nobody filed anything. The two facts sit in different systems and were never compared.

What everyone sees

A complaint log is treated as a sample of dissatisfaction, so a small log implies a small problem. It is not a sample. It is a record of the subset of dissatisfied people for whom the expected benefit of complaining exceeded the cost — a subset selected by temperament, time and confidence, not by severity.

What is actually happening

Hirschman’s framework separates two responses to decline: voice, which is costly and informative, and exit, which is cheap and silent. Where switching is easy, exit dominates, so the organisations with the most substitutable offering receive the least feedback about their failures. Voorhees, Brady and Horowitz surveyed dissatisfied customers directly and found the large majority never contacted the firm, with non-complainers showing lower repurchase intention than complainers whose issue was handled — the silent group being both larger and worse for the business. The consequence is a systematic inversion: the loudest signal comes from customers still invested enough to argue, and the absence of signal comes from the ones already gone.

Why it stays hidden

The gap hides because complaint volume is measurable and exit is a non-event. A complaint arrives with a timestamp, a category and an owner; a customer who simply never returns generates no row in any table. Reporting systems are built from things that happen, and the relevant thing here is a thing that stops happening.