Validate Your Startup Idea Before You Build It

Most startup ideas do not fail because of code. They fail because founders spend months building before validating demand. Startup MRI exists to help you identify the biggest risks before you invest weeks or months into a product. You fill in five fields — the problem, target audience, monetization model, acquisition channel, and your technical background. A deterministic rule engine runs 34 weighted checks across seven dimensions: validation, distribution, market demand, monetization, competition, founder skills, and build ease. Those seven are then combined into an opportunity score and an overall startup score. The full report generates in under 60 seconds. By default, every word in the report comes from the rule engine. When LLM access is configured, an optional second pass polishes the prose — but never the numbers, and never the recommendations. That deterministic layer is what makes two identical inputs produce identical reports: you can rerun the same idea next week and verify the analysis is reproducible.

No, and it doesn't try to. Three things make startup outcomes un-predictable in principle: market timing shifts, team execution variance, and exogenous shocks. Any tool that claims to predict them is selling a feeling, not a forecast. Startup MRI scores the inputs you actually control — clarity of the problem, strength of the distribution channel, realism of the monetization, and the gap between your background and what the idea needs. It's designed to reduce uncertainty, not eliminate it. A high score means your inputs are aligned with patterns seen in successful indie launches. A low score means they aren't. Either way, the decision — and the outcome — are still yours.

Three things. First, the numbers are deterministic. Submit the same inputs tomorrow and you get the same score — every time, byte-for-byte. That consistency makes comparisons and iteration easier: when you change one input and rerun, you know exactly which score moved and why. ChatGPT gives you a different answer each time and tends to lean positive, because language models are trained to be helpful and encouraging. Second, every number traces back to a specific rule that fired because of a specific input. If a score feels wrong, you can almost always point to the input that triggered it and the rule that fired. Third, the report includes a 4-week first-customer plan generated from the same rule engine that produced your score. ChatGPT gives you a generic 30-day plan; Startup MRI's plan reflects your chosen channel, audience, and monetization. The AI layer in Startup MRI polishes prose. The analysis itself comes from the rule engine.

Accuracy tracks input quality. A vague idea plus guessed answers will produce a vague report, and the score will reflect the vagueness — the report doesn't know your idea better than you described it. The rule engine is intentionally opinionated. Each of the 34 checks encodes a pattern drawn from the customer-development and lean-startup literature listed on our Sources page: weak distribution, untested pricing, founder-skill mismatch, and over-scoped MVPs. The score expresses how clearly your inputs address those patterns. It is not a prediction of traction. Use the score as a starting point, not a verdict. The seven sub-scores and the evidence section are where the actual signal lives.

Yes. The full evaluation — all 7 dimensions, the 4-week first-customer plan, and a shareable report URL — is free during Phase 1, which runs through the public launch. No signup, no email gate, no credit card. We may add optional premium features later (deeper competitive analysis, custom benchmarks, team workspaces), but the structured report itself will stay free. The part that helps you decide whether to spend six months of your life on an idea should not be the part we charge for.

Three steps, in order. First, read the critical-assumption section. The engine picks the single assumption that, if false, would invalidate the rest of the plan. If you can find evidence for it before you build, do. If you can't, the rest of the report is secondary until you do. Second, run the 4-week first-customer plan as a checklist. The plan is generated from your inputs — it specifies the actions that would put your idea in front of potential customers within 30 days. Treat it as the actual project, not a follow-up. Third, ignore the overall score after the first read. It's a summary number, not a target. If three sub-scores are weak, you have three specific things to address — not a vague "the idea is risky." Do these three things before writing code, and you'll know within a month whether the idea is worth the next six months. Skip them, and the report is a coin flip.