Customer research for a new business: questions that change what you build

Start with a situation you can recognise

'Small businesses' is usually too broad to guide a useful interview. A better starting group might be independent caterers who quote for corporate lunches each week. The situation gives you a place to look, a workflow to discuss and a purchase trigger to investigate.

Recruit people beyond friends and existing supporters. Include someone who recently bought an alternative, someone who decided against buying and someone still handling the problem manually. A small initial round can reveal language and contradictions, but it cannot establish market size or statistically prove demand.

Use the last-time interview

Ask for permission to take notes and explain how you will use them. Begin with the last time the problem occurred. Keep your product pitch until after the account of what happened.

• What happened the last time you needed to do this?
• What triggered the work, and what happened next?
• What did you use, and where did that approach fall short?
• What did the problem cost in time, money or missed work?
• Who chose the supplier or approved the purchase?
• What would make changing the current approach difficult?

Separate the quote from the conclusion

Use a simple record: customer situation, direct observation, interpretation and open question. 'I retyped 40 order lines yesterday' is an observation from the interview. 'They need an AI order platform' is your interpretation. Keeping those apart prevents the proposed solution from swallowing the evidence.

For a fictional caterer, five interviews might reveal three different issues: menu revisions, delivery timing and invoice follow-up. Do not blend all three into one large product. Investigate which issue recurs in a reachable segment and already creates a reason to spend.

Move from conversation to a small commitment

After the interviews, offer a clearly described pilot. Name its scope, duration, price, what you need from the customer and how both sides will judge completion. A booking, paid test or introduction to the decision-maker gives different information from a compliment. Record which commitment occurred instead of labelling all positive responses as validation.

Keep the commitment proportionate to your ability to deliver. If you cannot yet provide the service, describe the test honestly. Do not present a waitlist as a finished product or count a research participant as a customer.

A one-week research plan

Draft one segment hypothesis, recruit an initial set of participants, conduct the interviews and group the observations by recurring situation. At the end of the week, decide whether to narrow the group, change the problem or test an offer. Keep the unanswered questions beside the decision.

Strategyzer's customer jobs, pains and gains provide a useful organising lens. The interview sequence here is dotSuper's practical synthesis. Neither a completed canvas nor a positive interview removes the need to test a purchase and delivery.