District Research

What people complain about in software they already pay for

We read every one-to-three-star review we could reach across 24 categories of paid software, and counted what actually recurs in the 64,871 of them long enough to diagnose. Not opinions about the market. Counts.

64,871diagnosable complaints
24categories measured
5worth building
12ruled out, with reasons

Why a review is better evidence than a survey

A one-to-three-star review from a paying subscriber is three proofs at once: the problem is real, the wallet is already open, and the person is actively looking for a way out. Nobody writes one to be helpful. That makes it the cheapest honest signal available about software — and almost nobody counts them systematically.

The two findings below came out of doing so, and both of them contradicted what we expected before we started.

Finding

Most software complaints are commercially worthless — and you can tell which in one number. Where the employer buys the software and the employee uses it, complaint volume stays high and purchasing power goes to zero. We measured 0.00 on five separate products.

Finding

How we killed our own best idea with more data — a thesis drawn from 29 complaints across 6 apps looked strong, survived a second look, and collapsed at 58 apps. The pattern that kills a false thesis is specific and easy to test for.

The 24 studies

Each one carries its own numbers, its own apps and its own verdict — including the negative ones, which are the more useful half. How this is measured.