I run a pipeline that reads public complaints — Reddit, Hacker News, GitHub, Stack Exchange, Trustpilot, app store reviews, product forums, X — and scores each recurring one for commercial intent. Two LLM passes, deduplication across distinct authors so one loud thread does not look like a market, then an editorial gate. 942 have cleared it.
Here are the ten highest-scoring right now, with the raw numbers. Commercial intent is a 0-100 estimate of willingness to pay; confidence is how sure the scoring is.
1. Automate 40% of property manager calls, cut after-hours costs
intent 100/100 · confidence 90/100 · Real Estate / Local Operations · surfaced from reddit
Read the problem and the evidence
2. Stop Stripe Connect fraud before platform shutdown
intent 100/100 · confidence 90/100 · Security / Compliance · surfaced from reddit
Read the problem and the evidence
3. AWS Marketplace usage fraud detection and remediation
intent 100/100 · confidence 80/100 · Security / Compliance · surfaced from reddit
Read the problem and the evidence
4. 60-second mobile-first e-signatures, no account needed
intent 90/100 · confidence 95/100 · Legal / Practice · surfaced from twitter
Read the problem and the evidence
5. Podcast download validation that filters bots reliably
intent 80/100 · confidence 80/100 · Creator Operations · surfaced from reddit
Read the problem and the evidence
6. 24/7 M365 phishing response for BYOD education
intent 80/100 · confidence 80/100 · Security / Compliance · surfaced from reddit
Read the problem and the evidence
7. Recover failed Stripe payments for Laravel apps
intent 80/100 · confidence 80/100 · Finance / Accounting · surfaced from reddit
Read the problem and the evidence
8. Stop metadata drift across Glue and other data catalogs
intent 80/100 · confidence 80/100 · Data / Analytics · surfaced from reddit
Read the problem and the evidence
9. Secure occupied home showings for property managers
intent 80/100 · confidence 90/100 · Real Estate / Local Operations · surfaced from reddit
Read the problem and the evidence
10. Stripe dunning that counts all retry attempts
intent 80/100 · confidence 90/100 · Finance / Accounting · surfaced from reddit
Read the problem and the evidence
What I notice looking at this list
Security and compliance is over-represented at the top. Three of the ten are fraud or abuse response. That was not true of the corpus as a whole — across all 942, security is 7th by volume. It is a this-week artifact, and I mention it because it is exactly the kind of pattern that looks like a trend if you only read the top of a list.
The good ones are boring. Automating after-hours property manager calls scores 100. Nobody is going to tweet about building that.
My corpus has a bias I cannot fully remove. Dev tools are 44% of everything I hold, and my biggest single source is GitHub (563 items) ahead of Reddit (299). Not because developer problems are worth more — because developers complain in public, in writing, on platforms with APIs. Anyone mining public text inherits this.
Querying this from your editor
There is a free MCP server, so your coding agent can search the corpus without leaving the editor:
claude mcp add --transport http ddmarketer https://www.ddmarketer.com/api/mcp
Three tools need no key and no account: search_gaps, get_top_gaps, and validate_idea. That last one scores an idea you already have, and it returns a weak verdict when it only finds loosely related complaints rather than pretending a near-miss is validation.
A dependency-free reference client is on GitHub if you want to see exactly what goes over the wire — about 50 lines of plain fetch, no SDK.
The caveat that matters
Complaint volume is not demand. Somebody being annoyed on Reddit does not mean they will pay to fix it. This narrows where to look. You still have to talk to people before you build.
Top comments (0)