Skip to content
01 views

Startup Idea Validation Prompt for Claude

Turns a one-line startup idea into a structured critique problem severity, market size, competition, and a clear verdict modeled on the problem-solution-insight framework investors actually use.

Updated on October 6, 2026· By
Last tested October 6, 2026 against Claude
The Prompt
Act as a skeptical startup advisor evaluating a new idea not to encourage it, but to pressure-test it honestly. The idea is: "{startup_idea}". The target customer is: "{target_customer}". Evaluate the idea across these sections, in...
CategoryResearch
Perfect forStartup Idea Validation
Recommended models

Variables

Placeholder{Startup idea}
ExampleAn AI tool that automatically generates and sends invoice reminders for freelancers
What to change it toThe core idea being evaluated drives every section of the analysis.
Placeholder{Target customer}
ExampleFreelance consultants and agency owners with 5-20 active clients
What to change it toWho the idea is actually for shapes how urgency, competition, and market size get assessed.

Actual output, by model

Real output this exact prompt produced — not a description of what it might do.

Claude
[full 7-section evaluation as logged above, verbatim]

Which model runs this best

Claude delivered a genuinely skeptical, rule-compliant evaluation with no softening it correctly identified that the core idea solves an already-commoditized problem, named real competitors, and landed on an unsoftened "Weak signal" verdict with a specific named pivot, exactly as the structure requires. No edits needed.

Startup Idea Validation Prompt for Claude

Pressure-tests a startup idea across problem severity, competition, market timing, and a stated worst objection, ending in a direct verdict.

How To Use

  1. Replace {startup_idea} with a one- or two-sentence description of what you're building.
  2. Replace {target_customer} with who it's actually for.
  3. Paste into Claude.
  4. Treat the "Strongest Objection" section as the most useful part of the output, not the part to argue with.
  • 1.No AI model can validate willingness to pay treat a "Strong signal" verdict as permission to go talk to 10-15 real potential customers, not as the validation itself
  • 2.Run it again after any major pivot suggested in the verdict a changed idea deserves a fresh pressure-test, not an assumption that the fix obviously worked
  • 3.If the output still reads as encouraging, explicitly reply "be more critical, especially about the competitive landscape" it's easier to redirect than to trust a first pass that feels too easy

Example output

1. The Problem Late payment is real and well-documented in freelance/agency work but "invoice reminders" is a narrow slice of that problem. The actual pain...

FAQs

No, Nothing can predict that. It gives you a structured, honest read on the idea's weak points so you can decide what to test next, not a guarantee.

Where this prompt breaks

Known failure modes we hit while testing, so you can spot them before they waste a run.

A genuinely strong/plausible startup idea substituted in place of the weak example

Nothing broke, skepticism stayed calibrated rather than manufacturing false objections, landing on "Mixed signal" rather than forcing negativity

Missing target_customer, startup_idea provided

Silently assumes a generic target audience instead of flagging the missing input

Appended instruction conflicting with "Do not open with encouragement"

Nothing broke, held the original constraint and opened with the Problem section anyway, unlike a similar conflict test on a different prompt where the appended instruction won

No Perfect Match?

Generate a custom prompt built around your profession, AI model and goal.

Generate a Prompt →
Startup Idea Validation Prompt for Claude Description⌄

Ask most AI models to evaluate a startup idea and you’ll get encouragement dressed up as analysis “this has real potential” followed by a few soft caveats. That’s close to useless for the actual decision a founder needs to make, which is whether to spend the next year of their life on this. This prompt is built to resist that pull toward agreeableness by requiring the model to name the idea’s weakest point before it’s allowed to say anything positive.

It’s built for first-time founders, side-project builders, and indie hackers who need a real pressure-test before committing real time not a cheerleading session, and not a replacement for actually talking to potential customers.

What makes it hold up:

  • The structure follows the problem-solution-insight framework used in startup evaluation circles a startup idea is treated as a hypothesis with three parts to test, not a single thing to rate out of ten
  • A required “strongest objection” section forces a real critique before any encouragement, because an evaluation that leads with praise tends to stay soft even when it gets to the caveats
  • {startup idea} and {target customer} are separate variables, so the same framework produces a genuinely different read on a B2B SaaS tool for accountants versus a consumer app for parents the questions about urgency, willingness to pay, and competition land differently for each

One real limit worth knowing: no AI model can tell you whether real customers will actually pay see the Pro tips for the one step this prompt can’t replace.

Version history

  1. October 5, 2026Baseline entry first test of this prompt with startup idea="an AI tool that

Send me your AI wisdom

One email a week: new prompts worth using, and what's changed across the models.

Leave a reply

Your email address will not be published. Required fields are marked *