After classifying 24,216 raw items and deduplicating down to 6,672 complaints, four things only show up at that scale — they are invisible if you read a few dozen posts.
1. The top problem does not belong to the operator
Across the whole corpus the leaders are missing mobile features (1,162 mentions) and price and licensing (1,153) — both are problems for the software vendor. Sorted by mean severity, the order flips entirely: sensitive data and compliance 3.91, internet dependency 3.75, broken process 3.70. The corpus mean is 3.31 with a standard deviation of 0.89, and the lowest-severity label is tool sprawl at 2.53. Volume measures how widespread a problem is, severity measures how much it hurts — at small scale the two get read as one number.
2. The source decides the ranking
In store reviews, missing mobile features are 26.5% and pricing 24.5%, while manual re-keying is 0.5%. On Reddit and Hacker News (1,924 items), manual re-keying is 16.8%, missing mobile features drop to 1.4% and pricing to 5.1%. Same question, two different industries. Read only one source type and the conclusion is reliably about half wrong.
3. Most workarounds buy nothing at all
Among the Reddit and Hacker News items, 4,739 never say how the writer copes. Of those that do: no workaround 203 (the largest group), building a private tracker 124, using AI 122, copy-pasting across files 108, emailing for the data 74. A trillion-dollar software category exists side by side with people choosing to do the work by hand — not for lack of tools, but because the tools do not connect to the system already running.
4. Method note — the easiest thing to misread
Percentages only mean something within a source. Google Play supplies most of the volume behind both leading groups (1,134 of the 1,162 mobile-gap mentions and 1,049 of the 1,153 pricing mentions), so the whole-corpus ranking mostly reflects how store users grade apps. Item counts are not people counts: 473 rows carry no severity because they are title-only entries, and the workaround field is present on only 29% of rows — so every workaround conclusion must state that it rests on a much smaller set.
All the tables are in the 6,672-complaint ranking; how the data was collected is in collecting 24,216 items.