METHODOLOGY
Evidence over hype.
Every published opportunity is a source-linked research brief, not an AI-generated guess.
How a signal becomes an opportunity
- Source responsibly. We use permitted public APIs and public pages, honor access rules and rate limits, keep a minimal excerpt, and link to the original.
- Normalize and deduplicate. Repeated complaints are grouped by workflow and buyer. A unique source cannot be published twice.
- Score the evidence. A reviewer applies the published rubric below and records a plain-English rationale. The rationale matters more than the number.
- Review before publication. A person inspects the source, excerpt, buyer claim, alternatives, stated gap, narrow MVP, price hypothesis, and proposed first test.
The 100-point evidence rubric
| Factor | Points | What earns the score |
|---|---|---|
| Intent and specificity | 0–25 | An explicit problem, search, workaround, feature request, or buying signal tied to a concrete job. |
| Buyer clarity | 0–15 | A reachable role or segment is identifiable from the source and workflow. |
| Urgency and frequency | 0–15 | The pain is costly, risky, time-sensitive, or repeats often enough to justify action. |
| Repetition | 0–15 | Independent reactions, comments, or corroborating sources show the signal is not isolated. |
| Reachability | 0–10 | Prospective users can be found through a focused community, profession, or installed product. |
| Existing alternatives | 0–10 | Users already spend money, time, or effort on a workaround and can describe its gap. |
| Recency and source quality | 0–10 | The source is current, inspectable, attributable, and specific enough to verify. |
Scores below 70 are normally kept in review. Scores from 70–84 are promising signals that still need customer discovery. Scores of 85 or more indicate unusually strong observable evidence—not guaranteed demand.
What a score means
A score reflects the strength and usefulness of observable evidence. It is not a market-size estimate, valuation, revenue forecast, investment recommendation, or guarantee that a product will succeed. A single post can be useful evidence, but never proves a market.
How AI may be used
AI may help normalize, classify, or summarize research. It may not invent a quote, source fact, buyer, demand signal, market number, or willingness to pay. Human review remains required before publication.
Corrections and removals
Every source subject or visitor can submit a correction or removal request. We record the request for human review and can archive a card without deleting the original request history.
Methodology last reviewed: August 22, 2026.