How IdeaChecker scores an idea
Most AI “validators” say yes to everything. Ours is built to find the reasons an idea fails first, score before writing, cite or mark unverified, and let the numbers decide the verdict.
Eight dimensions
| Dimension | Weight | What a 10 means |
|---|---|---|
| Problem severity | 15% | a frequent, painful, expensive problem people already pay or hack around to solve |
| Market size and demand | 15% | clear search and spending demand, a reachable segment, growing |
| Competition and saturation | 15% | few direct rivals, weak incumbents, a credible wedge; 0 = saturated with strong, cheap rivals |
| Timing | 10% | regulation, technology or behaviour shifted recently in favour of the idea |
| Feasibility | 15% | can be launched by this founder with this budget, time, skills and licences |
| Monetization and unit economics | 15% | customers pay, margins cover acquisition, pricing has precedent |
| Founder fit | 5% | skills, network, location and time match what the idea needs |
| Fatal risks | 10% | no single dependency, legal issue or platform rule that can end it; 0 = one obvious killer |
Score and verdict
Each dimension is scored 0 to 10 and multiplied by its weight; the weights add up to 100, so the score is out of 100. Build starts at 70, Pivot at 45, below that is Kill. The thresholds are fixed and the verdict is computed by the server from the dimension scores: the text of the report cannot change it.
Rules the checker follows
- Score every dimension before writing any prose, from evidence, not from how exciting the pitch sounds.
- Red flags come before strengths and must be specific to the idea, the market and the location.
- Never invent numbers, competitors, prices or statistics. A dimension without a source is printed as unverified.
- The founder’s claims about demand are claims to test, not facts.
- Local ideas are judged against the named local market; online ideas against the global one.
Where the evidence comes from
The Quick Check is a first pass from the model’s general knowledge, with no research; treat it as a sceptical friend, not a study. The Full and Deep Checks run web searches, read the sources, and add real listings from our Local business directories for the idea’s market, so “competition” is counted rather than guessed.
How verdicts fall
0% Build · 12% Pivot · 88% Kill across 43 checks
If Build ever exceeds a third of all checks we recalibrate the rubric. A checker that approves most ideas is not checking anything.
What the check is not
It is an opinion built from public sources and a fixed rubric. It is not legal, financial or investment advice, and it cannot know your customers better than ten conversations with them will. Use the kill criteria and the 30-day plan to find out.