I run a small data pipeline that reads public software complaints: GitHub issues, Reddit threads, Stack Overflow questions, Hacker News comments, Trustpilot reviews, app store reviews, public forums, and posts on X. It clusters them into validated gaps and scores each one for commercial intent. The corpus just passed 1,000 underlying complaints, so I published the aggregate stats and spent an evening staring at them. Some of what came out surprised me.
Current state: 988 validated gaps distilled from 1,012 underlying complaints, recomputed hourly. Everything below is quoted from the live stats pages as of September 5, 2026.
The most complained-about products
Ranked by how many validated complaints name the product:
| # | Product | Named complaints | Avg emotion (1-5) |
|---|---|---|---|
| 1 | MySQL | 51 | 3.4 |
| 2 | Airbnb | 38 | 4.1 |
| 3 | Shopify | 36 | 3.9 |
| 4 | Elasticsearch | 34 | 3.5 |
| 5 | QuickBooks | 24 | 4.1 |
| 6 | Stripe | 20 | 4.5 |
| 7 | Excel | 18 | 3.4 |
| 8 | Claude | 18 | 3.9 |
| 9 | GitHub Actions | 16 | 3.6 |
| 10 | Zapier | 15 | 4.2 |
Full table with method notes: Most complained-about SaaS tools
Two things stand out to me. First, MySQL at 51 named complaints is a volume story: a huge user base repeatedly hitting the same sharp edges, at moderate emotion (3.4/5). Second, anger is decoupled from volume. Stripe is only sixth by count, but its complaints average 4.5/5 in emotional intensity, the highest in the top ten. People are not mildly annoyed about Stripe; they are describing frozen payouts and account shutdowns. QuickBooks (4.1) and Zapier (4.2) follow the same pattern: money-adjacent tools produce money-adjacent rage.
Obvious caveat, printed on the page too: these counts reflect where complaints were collected, not market share, and products with fewer than 5 mentions are not ranked.
Where complaints actually live
| Platform | Gaps | Share |
|---|---|---|
| GitHub | 302 | 31% |
| 290 | 29% | |
| Stack Overflow | 208 | 21% |
| Hacker News | 76 | 8% |
| Trustpilot | 59 | 6% |
| X (Twitter) | 28 | 3% |
| App Store | 14 | 1% |
| Public forums | 11 | 1% |
Full breakdown: Where SaaS complaints live
If your idea validation process is "search Reddit", you are reading 29% of the complaints and missing the other 71%. And the split is not uniform across markets. E-commerce complaints are 82% Reddit. Data and analytics complaints are 70% Stack Overflow. Dev tools complaints are 58% GitHub. Real estate complaints are 94% Reddit. The platform you read determines the market you think exists.
This is also a quiet argument against single-source research tooling in general: any pipeline built on one platform's API inherits that platform's coverage bias and its policy risk.
Anger and willingness to pay
Every complaint is scored 1 to 5 for emotional charge. The most emotionally intense categories (categories with fewer than 10 gaps excluded):
- Finance / Accounting: 4.1/5
- Real Estate / Local Operations: 4.1/5
- Security / Compliance: 4.0/5
- E-commerce: 3.8/5
- Dev Tools / SaaS Infrastructure: 3.6/5 (the biggest category at 387 gaps is also one of the calmest)
Corpus-wide, 58% of complaints score 4 or 5 out of 5. The full distribution is 1/5: 1 complaint, 2/5: 48, 3/5: 370, 4/5: 468, 5/5: 101. Details: Emotional intensity by category
Willingness to pay is the more interesting number. A complaint gets flagged when the person signals they would pay for a fix. Across the whole corpus, that flag is on for 93%. By category:
- HR / Recruiting: 97%
- Real Estate / Local Operations: 96%
- Security / Compliance: 96%
- Marketing Operations: 96%
- No-Code / Automation: 90% (the lowest ranked)
Details: Willingness to pay by category
Honest note on that 93%: there is a selection effect. The corpus is validated gaps, not random internet grumbling, so a high overall rate is partly by construction. What I find informative is the spread and which categories sit on top: recruiting, real estate, and security complaints tend to be people describing problems that cost them money this week.
How the numbers get made
Short version, because the method matters more than the charts. The pipeline reads public posts across the 8 platforms, an LLM classifier extracts structured complaints, dedup collapses repeats across distinct authors so one viral thread does not look like a market, and a second pass clusters complaints into gaps. Each gap gets a commercial intent score (0 to 100), an emotional intensity read (1 to 5), and the willingness-to-pay flag. Product mentions are matched on word boundaries. Categories with fewer than 10 gaps are excluded from rankings because a percentage over two or three rows is noise, not signal. Every stat page recomputes hourly from the live corpus.
Two biases I cannot remove. One: developers complain in public, in writing, on platforms with APIs, so dev tools are over-represented (387 of 988 gaps). Two: collection coverage is not market share, anywhere in this data. I would rather publish with those caveats on the page than pretend the corpus is a random sample of all software pain.
Full writeup: methodology
What I would build from this
Three gaps from the corpus that line up with where the stats point:
1. Airbnb host protection from fraudulent cleanliness claims (intent 80/100). Airbnb is the number 2 most complained-about product at 4.1/5 emotion, and real estate is the most emotionally intense category at 96% willingness to pay. The gap: hosts getting hit with fake cleanliness claims from guests fishing for refunds, with no good evidence trail on their side. Read the full gap dossier
2. SMB IT disaster recovery for non-technical management (intent 80/100). Security and compliance sits at 4.0/5 emotion with 96% willingness to pay, and these complaints come from office managers, not sysadmins: the person who got handed "the IT" and just realized there are no tested backups. Read the full gap dossier
3. WooCommerce plugin bloat killing checkout conversion (intent 80/100). E-commerce complaints are 82% Reddit-sourced and 92% willing to pay, and this one is the classic: 40 plugins deep, checkout takes six seconds, the owner knows it is costing sales but cannot tell which plugin does it. Read the full gap dossier
Each dossier includes the underlying complaints, an MVP scope, suggested pricing, competitors, and risks.
The caveat that matters
Complaint volume is not demand. Someone raging about QuickBooks does not mean they will pay you to fix it. Data like this narrows where to look; it does not replace talking to people. If you want to query the corpus yourself, the search tier is free, and the stats pages above are free to cite with attribution (they recompute hourly, so cite the access date).
Top comments (0)