Complaints are cheap research, usually buried

Someone has done detailed unpaid analysis of your weakest process and handed it to you.

A complaint is a customer spending their own time to tell you precisely where your process failed, with a worked example attached. Commissioned research would cost a great deal and be vaguer. Yet complaints are typically routed to a function whose job is to close them, measured on speed and volume, and never read as a body of evidence. The individual case gets an answer. The pattern behind two hundred similar cases goes unexamined, because nobody's role is to look at the pile rather than the item.

Categories make this worse. Complaints get sorted into buckets defined years ago for reporting convenience, and the bucket labels hide the very thing you need to know. Half of everything lands in a general category, the detail lives in a free text field that no report touches, and the resulting summary tells leadership that the largest cause of dissatisfaction is other. Reading a hundred raw cases end to end will teach you more than any dashboard built on that classification.

Treat the complaint file as your cheapest source of truth. Read it directly and regularly, in the customer's words rather than in the summary. Count causes rather than cases. Then ask the uncomfortable question: which of these did we already know about, and what has stopped us fixing it. The answer is often not ignorance. It is that the fix belongs to a team that never sees the complaints.