Test a new service idea with real customers before you build it

Quick answer

Describe your idea in one sentence, ask AI for interview questions, the assumptions the idea depends on and a cheap two-week test, then talk to about five real potential customers and decide from what they say.

  • For: Freelancers and independent providers, Professionals, Small business owners and solopreneurs
  • Level: Explorer
  • Estimated setup time: 30 to 45 minutes to prepare, then about two weeks to test (an estimate; yours may differ)

This playbook has no recent review date. Check tool details on the official sites before you rely on it.

The problem

It is easy to spend time and money building a service nobody wants, and AI cannot tell you what your market will do.

Before you start

  • One sentence describing the service and who it is for
  • A way to reach about five potential customers
  • Willingness to hear that the idea needs changing

Tools you can use

  • A general-purpose AI assistant for questions and structure
  • A notebook or spreadsheet to record what people say
  • Alternative: write the questions yourself and use AI only to critique them

Step by step

AI can help you prepare to learn from customers. It cannot replace them. This playbook keeps the learning with real people.

Step 1: Describe the idea

Write one sentence: what the service is and who it is for.

Step 2: Prepare the test

Idea testing prompt
I am considering offering [service] to [customers]. List 10 questions I should ask 5 potential customers, the assumptions my idea depends on, and the cheapest test I could run in two weeks. Do not predict income; focus on what I can learn.

Step 3: Improve the questions

Remove leading questions and focus on what people do today and what it costs them, not on whether they would like your idea.

Step 4: Talk to real people

Speak to about five potential customers. Record what they say in their own words in one place.

Step 5: Decide

Decide whether to continue, change the idea or stop. Write down what you learned and the next small test.

What good looks like

  • Your decision rests on what real people said, not on AI output
  • You recorded answers consistently
  • You can state what you learned and what you will do next

Common mistakes

  • Asking leading questions
  • Talking only to friendly people
  • Treating AI estimates as market evidence
  • Building before you have heard from customers

Privacy and human review

Keep customer details out of prompts. Be honest with people about why you are asking. This playbook makes no promise about income or demand.

Did this work for you? Tell us what worked and what did not on the contact page. We use your feedback to improve the guides.