Same question — what do office workers actually suffer from — three source types, three different lists. The source column in the 6,672-complaint corpus tells you what kind of voice you are reading, not just how much of it there is.
Three kinds of voice
- Google Play and App Store reviews: product anger. Missing mobile features account for 26.4% (Play) and 26.8% (App Store) of complaints; pricing and licensing 25.0% and 22.7%.
- Reddit and Hacker News: process narrative and workarounds. Manual re-keying is 28.7% on Hacker News alone and 16.8% across Reddit plus Hacker News, while barely appearing in store reviews.
- Stack Exchange: a specific technical bug. Broken formulas are 31.2%, access rights 16.9%, manual re-keying 12.6%.
The direct consequence: decide what to build from store reviews and you conclude the problem is missing mobile features. Decide from Reddit and Hacker News and you conclude it is manual re-keying. Both are right, but they are different problems — one belongs to the software vendor, one to the operator.
Pick the source by the question
For what is wrong with my product, read store reviews: high volume, easy to collect over time, but it only answers product questions. For where the process breaks inside a company, read Reddit and Hacker News, because that is the only place people describe their workarounds and the extra tools they bought. For the specific behaviour of one formula or macro, Stack Exchange. The results from the first two source types: 6,672 complaints and the workarounds.
Actual access status from this server
- Reddit on the original domain: HTTP 000 — blocked at the ISP, DNS returns ::1. Collected through the arctic-shift mirror.
- Microsoft Answers (Excel): 200 after a single 301, readable through CloakBrowser CDP. Microsoft Tech Community: 403 on a direct request, needs rendering.
- Google Docs Editors Community 200, Atlassian Community 200, Spiceworks Community 200 — direct read.
- G2, GetApp, TrustRadius, Trustpilot: 403 on direct requests, but the content is still readable through an extractor.
- Google Play 200 (via google-play-scraper), App Store reviews via RSS 200, Hacker News via the Algolia API 200 with no auth, Stack Exchange API 200 with a quota.
Method note: every item carries the same label set and the same 1–5 severity scale, so comparisons across sources use one yardstick — but percentages only mean something within a source, and must not be added across sources collected in different ways.