As of 13 August 2026, AI can only partly research customer needs for your new product.

What to actually do

1. Hand it to a person

   The route this page recommends

   A person who owns the outcome does this end to end, worth it when the failure is dear.

2. Use a tool built for this

   Second choice

3. Do it yourself

   The distant third

   A chat interface, power-user skill, and roughly 1 hour until you can act on the result.

   How to actually do it

   1. Open a working document and record the product description, proposed price, target customer, intended problem, business decision and research deadline.
   2. Gather existing evidence such as customer interviews, survey responses, support messages, sales notes, product analytics and competitor observations, removing personal data that is not needed.
   3. Paste the product context, constraints and evidence into the prompt, and ask the model to separate direct evidence from interpretation and hypothesis.
   4. Use the drafted recruitment criteria and neutral interview script to contact a varied set of relevant customers or potential customers, asking about current behaviour and alternatives before showing the product.
   5. Paste anonymised interview notes and survey results back into the same chat or working document, asking the model to code recurring needs, differences between segments and contradictory evidence without treating repeated opinions as proof of demand.
   6. Compare every proposed finding with the original quotations, response counts and product analytics, then ask a colleague who understands the customers to challenge the sampling and conclusions.
   7. Write the final report with evidence, confidence levels, unresolved questions and next actions, and use it to choose a small test such as a prototype, landing page or customer trial rather than treating the AI report as validation.

   Prompt

   Act as a careful customer-research analyst for a new product. Do not claim that a need is widespread unless the evidence supports that conclusion, and do not invent customers, quotes, market figures or competitor facts.
   
   Product: [describe the product, current stage and proposed price]
   Target customers: [describe the people or organisations you think may buy it]
   Problem it is intended to solve: [describe the problem]
   Business decision this research must inform: [state the decision and deadline]
   Evidence available: [paste customer interviews, survey responses, support messages, sales notes, usage data or links to public sources]
   Constraints: [state budget, access to customers, geography, sector and any legal or privacy limits]
   
   Produce the following:
   1. A clear restatement of the research question and the assumptions that must be tested.
   2. A separation of direct evidence, reasonable interpretation and unsupported hypothesis.
   3. A prioritised research plan using the available access to customers, including who to recruit, how to avoid a biased sample, and the questions to ask in interviews or surveys. Ask about current behaviour, alternatives, frequency, consequences and willingness to change before asking about the proposed product.
   4. A short interview script and a survey draft with neutral wording.
   5. A framework for analysing the results, including how to compare customer segments and identify contradictory evidence.
   6. A report template with findings, evidence quotations, confidence levels, unresolved questions and recommended next actions.
   7. A list of claims that must not be made until more evidence is collected.
   
   If the evidence is missing or weak, say exactly what is missing. Finish with the smallest set of customer conversations or data checks needed to test the most important assumption. Do not recommend building the product solely because interviewees say they like the idea.

   Open it prefilled in ChatGPT or Claude, or copy it into Gemini, which takes no prefill link.