Understand an idea check

"Check my idea" runs the same research pipeline as a normal Discovery run — but the subject is your own idea. The system reads your pitch, derives the market it competes in and the buyer it serves, researches that market's real public discussion, and then puts your idea through the same evaluation every generated idea gets: calibrated scoring, an adversarial review, and a live web probe for competitors that already ship it.

How to read the verdict

The check answers three separate questions, because they deserve separate answers:

1. Is the problem real? Answered from linked discussion — real quotes from real threads, with the breadth stated as posts, distinct accounts, communities, and months. Thin or missing evidence is stated plainly, and an idea whose problem no thread names is graded as a hypothesis, not as evidence-backed.
2. Is the space occupied? The direct-equivalent probe looks for products shipping the same core mechanism. Named competitors are listed with what they ship, a snippet-derived price band, and their known gap. The adversarial review searches more broadly for material overlap and objections. If the direct probe does not return a result but that review finds a real concern, the verdict says Concerns found and sends you to “What would kill it.” It does not relabel an adjacent tool as a verified direct competitor.
3. Is there demand? Not measured in this check — deliberately. Searching for a solution is a behavior, one tier stronger than complaining about a problem, and measuring it (real search volume, market size, pricing) is what Deep Research adds.

The evidence-confidence line rates the analysis, not your idea: how much linked discussion backs the read, stated separately from the verdict itself.

Where the other ideas come from

To grade your idea, the run generates and scores every other approach the same evidence supports. That pool is the benchmark your idea was ranked against.

Your idea is pinned at the top for comparison, but it is not automatically ranked first. The number beside it is its score rank. A submitted idea shown first with a #4 marker ranked fourth against that pool. Pinning changes presentation, not scoring, and nothing replaces your idea unless you choose a different candidate yourself.

The detail view can show Product shape and Delivered as separately. Product shape describes the business archetype. Delivered as records the primary surface in your pitch, such as a browser extension, API, mobile app, report, or service. The check preserves that delivery identity while it develops the brief for scoring.

If a competitor caps your idea, the run drafts one revision that attacks the competitor's known gap. It appears only when it scores better than your original, always labeled as a revision of your idea, with the reason it was drafted. When the revision doesn't beat your original, the check says so instead of showing it.

What a desk check cannot tell you

• We read what people said unprompted. We could not ask them the follow-up question.
• Nothing here is commitment evidence. No one has paid, pre-ordered, or given up anything for this idea.
• Buying-signal strength is relative, never a price prediction.

That's why the check ends with the cheapest next tests — the ones only you can run, each with a published kill number: five problem interviews, a landing page with ~50 clicks (kill below 1% click-to-email), three half-price pre-sales (kill below 3), a concierge weekend. Refusing to commit is itself an answer.

If your idea is ruled out

The check keeps your idea as the subject even when the screen rules it out — the problem can be real while this particular product isn't the way in. The page states why, what evidence would change the verdict, and offers the re-run escape: refine your idea in your own words and check it again. Deep Research is deliberately unavailable for a ruled-out idea — we won't sell a deep dive on an idea our own screen ruled out.