Where the ideas come from
Every report on this site is free to read, so it is only fair to be equally open about how it was put together. This page explains the research behind the 431 ideas in the database, what the score means, and where it falls short.
The signals
Ideas start from evidence that someone already has a problem, not from a brainstorm. The research looks for three kinds of signal, and an opportunity usually needs at least two of them before it becomes a report.
People complaining in public
Forum and community threads where the same frustration comes up repeatedly, with enough detail to tell a real workflow from a passing annoyance. Reddit is by far the most cited source in the catalogue.
Search demand
Keyword and trend data showing whether interest in a problem is growing, seasonal or already dying. This is what separates a genuine shift from a news cycle.
Something that just changed
A platform launching a feature, a regulation coming into force, a price moving. Most reports include a “why now” section anchored to a dated announcement, which is why so many citations point at company newsrooms and government sites.
Most cited sources in the reports
Counted from the links inside the reports themselves, 310 of the 431 ideas carry at least one. This table is generated from the database, not written by hand.
| Source | Reports citing it |
|---|---|
| trends.google.com | 80 |
| reddit.com | 58 |
| nhs.uk | 32 |
| newsroom.spotify.com | 21 |
| newsroom.tiktok.com | 20 |
| blog.youtube | 19 |
| about.fb.com | 14 |
| nuffieldhealth.com | 13 |
| producthunt.com | 12 |
| who.int | 12 |
| theportugalnews.com | 11 |
| news.ycombinator.com | 11 |
How the score works
Each idea is rated 0 to 10 on twelve dimensions with equal weight. The twelve add up to 120, and the total is divided by 1.2 to land on a 0 to 100 score. Nothing is weighted higher than anything else, which is deliberate: the score is a comparison tool between ideas, not a verdict on any single one.
| Problem | How much the pain actually hurts, from nice-to-have to hair on fire |
| Market Size | How big the reachable market is, not the headline TAM |
| Profit Margin | What is left after delivery costs |
| Recurrence | One-off sale, repeat purchase or predictable MRR |
| Average Ticket | Value per transaction, since a higher ticket means fewer customers to validate |
| Scalability | Whether growth costs the same as the first customer |
| Accessibility | Whether one person can start it without a team or capital |
| Distribution | How hard the first hundred customers are to reach, roughly the inverse of CAC |
| Defensibility | How long a copy would take, and what stops it |
| Time-to-Revenue | How long until the first euro |
| Regulation | Inverted: heavy licensing scores low, no red tape scores high |
| Trend | Whether the market is growing, flat or shrinking |
Scores are frozen once an idea is published. Re-running the import updates the text of a report but never rewrites its scores, so a number you saw last month is the same number today.
The idea of the day
One idea is featured at a time, picked as the highest scoring one that has not been featured yet. When the catalogue runs out, the rotation starts over. Being featured says nothing extra about an idea beyond its score and its turn in the queue, and every other idea in the database is just as readable.
Portugal Market Insights
The sector dashboards are a different kind of data: financial aggregates published by Banco de Portugal through the Central de Balanços, covering roughly 550,000 companies grouped by CAE Rev.3 sector. Those figures are pulled directly from the BPstat public API and are not scored or interpreted by us. When a number there disagrees with a number in a report, trust the dashboard.
What this is not
The research is assembled with AI and reviewed before publishing, which makes it fast and wide, but also fallible. A report can cite a source that has since changed, estimate a market with the wrong assumption, or miss a competitor that already solved the problem well.
Treat a high score as a reason to spend an afternoon validating, not as permission to skip validation. Nothing here is investment advice, and no idea has been tested in the market by us. Market sizing figures are estimates, and where a report says TAM or SAM, read it as an order of magnitude.
If you find something wrong, tell us. Corrections are the cheapest way to make this better: get in touch.
431 ideas, every report open, no account needed.
Browse the database