I want to validate this startup idea before committing meaningful time or money. Idea: [STARTUP IDEA] Target customer: [TARGET CUSTOMER] Problem evidence I have: [CURRENT EVIDENCE] Proposed solution and business model: [SOLUTION AND BUSINESS MODEL] Constraints: [CONSTRAINTS] Act as an experienced early-stage startup researcher. Do not treat enthusiasm, a large market, or competitor funding as proof of demand. First separate facts I supplied from assumptions that still need evidence. Then assess the idea across five areas: urgency and frequency of the customer problem; current alternatives and switching friction; a narrowly reachable initial customer segment; willingness and ability to pay; and feasibility of acquiring the first 10 customers.
Produce: 1. A one-paragraph problem and customer definition, rewritten to be falsifiable. 2. An assumption table with the assumption, why it matters, risk level, evidence currently available, and cheapest credible test. 3. A competitor and alternative map. Include doing nothing, spreadsheets, agencies, internal workarounds, and direct competitors where relevant; do not invent company names or claims. 4. Three validation experiments ranked by learning value per dollar and per week. For each, specify the audience, exact recruiting approach, test artifact, interview or conversion signal, pass/fail threshold, and likely bias to control. 5. A 30-day sequence showing what to do first, what evidence would change the plan, and when not to build. 6. A recommendation: proceed, narrow the segment, change the offer, or stop. State the reasoning and the single biggest uncertainty. Self-check before answering: flag every conclusion that relies on an unsupported market fact, and make pass/fail thresholds measurable. Ask up to 3 clarifying questions only if a required input is missing.
Fill in
| Placeholder | What to enter | Example |
|---|---|---|
| [STARTUP IDEA] | Describe the product idea and the problem it is intended to solve. | A service that turns weekly construction site photos into lender-ready progress reports. |
| [TARGET CUSTOMER] | Name the initial customer segment, job role, company type, and context. | Project managers at US commercial contractors with 20–150 employees managing renovation projects. |
| [CURRENT EVIDENCE] | List interviews, observations, pilot interest, sales data, or other evidence you already have. | Six managers said reporting takes 2–4 hours weekly; two offered to review a mock report; no paid commitments. |
| [SOLUTION AND BUSINESS MODEL] | Explain the proposed product, pricing approach, and how customers would buy it. | Upload photos and job notes; software drafts a report. Charge $149 per active project per month, sold by a short demo. |
| [CONSTRAINTS] | State your available time, budget, skills, network, and any regulated or technical limits. | One founder, $2,000 test budget, eight weeks, no construction-industry audience beyond the six interviews. |
How to use
- Replace every bracket with your actual idea, evidence, and constraints.
- Treat the assumption table as a research backlog, not a verdict from the model.
- Run the first experiment with real prospects and record the exact result against its threshold.
- Follow up with: “Revise the recommendation using these results: [RESULTS]. What should I test next?”
Variations
Interview guide
Use this when you need discovery interviews before choosing a solution.
Create a 30-minute customer-discovery interview guide for [CUSTOMER SEGMENT] about [PROBLEM AREA]. My current hypothesis is [HYPOTHESIS]. Write a warm opening, 10 ordered open questions, and two follow-ups per question. Focus on a recent real incident, current workflow, costs, approvals, alternatives, and buying process. Do not ask leading questions or pitch my idea. Add a note after each question explaining what evidence it seeks. End with a coding sheet for patterns across interviews. Ask up to 3 questions only if an input is missing.
Landing-page test
Use this when you need to test positioning and demand with a simple page.
Design a demand-test landing page for [OFFER] aimed at [AUDIENCE]. The customer problem is [PAIN], and the proposed call to action is [CTA]. Write the page copy: headline, subhead, three outcome-focused bullets, how it works, objections, FAQ, and CTA. Then define a two-week test plan with traffic source, budget, conversion event, minimum sample target, and interpretation rules. Do not claim results, testimonials, integrations, or guarantees I have not supplied. Flag claims that require proof.
Pilot proposal
Use this when a prospect is interested and you need a bounded paid pilot.
Draft a one-page pilot proposal for [PROSPECT TYPE] to test [PRODUCT]. Include the customer problem, pilot scope, what each side supplies, timeline, price or no-cost rationale, success metrics, data and security boundaries, review cadence, and a clear decision at the end of the pilot. Use these details: [PILOT DETAILS]. Keep commitments realistic and do not promise features, outcomes, or legal terms not provided. End with five questions to resolve before sending the proposal. Ask up to 3 clarifying questions only if needed.
Tips
- Interview people who recently handled the problem; opinions about hypothetical tools are weaker than accounts of a specific past workflow.
- Use a test that asks for a scarce commitment, such as an introduction, calendar time, a data sample, or payment, rather than a generic sign-up.
- Set thresholds before seeing results so one enthusiastic conversation does not outweigh repeated weak signals.
- Track disconfirming evidence separately, especially customers who have the problem but still will not change their current method.
FAQ
Can AI validate my startup idea without customer research?
No. It can expose assumptions and design tests, but actual customer behavior, access, and willingness to pay need outside evidence.
How many interviews should I conduct?
Start with 8–12 tightly defined prospects, then assess whether the same workflows, language, and objections repeat. Continue if your segment is still too broad or patterns conflict.
Should I build an MVP before validating?
Build only the smallest artifact needed for the next test, such as a clickable mockup, manual service, or landing page. Avoid building a full product before evidence supports the core demand.