Founder conducting a customer research interview by video call

Last Updated: Jun 22, 2026

Your website is live. You post occasionally. Perhaps you have tried outreach or a small advertising campaign. Yet people still ask, “What exactly do you do?”

The usual reaction is to change the headline again, add more features, or buy another marketing tool. But the problem may be simpler: your message is based on how you describe the business, not how customers describe the problem.

You do not need a large research budget to improve this. Five well-run conversations will not prove what an entire market wants, but they can reveal useful patterns, objections, and exact words. This guide shows you how to collect that evidence and turn it into a clearer website, sales message, and content plan.Once your message is clear, you can use it to build a simple founder marketing stack around real customer evidence.

Concise answer

Customer research for a startup means gathering first-hand evidence about why people look for a solution, what they have tried, what makes the problem costly, and how they choose. A practical starting point is to interview five recent customers or qualified prospects, capture their exact words in a spreadsheet, group repeated patterns, create a one-page Message Evidence Map, and test one message for two to four weeks.

Key definitions

Customer research is the systematic collection of evidence about customers’ situations, behavior, needs, choices, and language.

Voice of the customer (VoC) is the exact language customers use to describe their problems, desired outcomes, objections, and decision criteria.

Marketing message is the clear explanation of who you help, which problem you solve, what useful outcome you support, and why someone should believe you.

Evidence is something observed or said by a real person in a relevant situation. A founder’s belief may be a valuable hypothesis, but it is not customer evidence yet.

Why founders struggle with their message

Founders spend every day inside their product or service. Customers do not.

You may think in terms of capabilities, technology, processes, and category labels. A customer is more likely to think about a missed deadline, an awkward manual task, an uncertain decision, or a result they cannot achieve.

This difference creates familiar problems:

  • the homepage describes features but not the customer’s situation;
  • sales calls require long explanations;
  • content covers broad industry topics but not urgent questions;
  • SEO targets volume without enough business relevance;
  • AI-generated content sounds polished but interchangeable;
  • advertising amplifies a message that was never validated.

Your website is usually the most important place to apply this language because it acts as the foundation for the rest of your marketing. The answer is not to ask people whether they “like” your headline. People are polite, and opinions about hypothetical purchases are weak evidence. It is better to ask about a real event that already happened.

The 5-Conversation Message Sprint

The sprint has six steps:

  1. Choose one decision.
  2. Recruit five relevant people.
  3. Ask about recent behavior.
  4. Capture exact phrases.
  5. Build a Message Evidence Map.
  6. Publish one test and review the signal.

You can complete the work in a week with video calls, a spreadsheet, and a few focused hours.

Step 1: choose one decision

Do not begin with “I want to understand my customers.” That is too broad.

Choose a decision you need to make, such as:

  • Which problem should lead the homepage?
  • Why do suitable prospects delay buying?
  • What proof do prospects need before booking a call?
  • Which customer question should become our next guide?
  • Which outcome should we emphasize in sales outreach?

Write the decision at the top of your research sheet. It will keep the interviews useful.

Good sprint question: “What message should lead our bookkeeping-service homepage for owner-led businesses?”

Too broad: “What do small businesses think about finance?”

Step 2: recruit five relevant people

Choose people who have recent experience with the problem. Recent behavior is easier to remember and more useful than general opinion.

A practical mix is:

  • two recent customers who chose you;
  • one customer who considered alternatives;
  • one qualified prospect who did not buy or delayed;
  • one person who matches the audience and recently solved the problem another way.

If you have no customers yet, speak with people who recently experienced the problem and made some form of decision. Do not interview only friends who want to encourage you.

Keep the invitation simple:

“I am improving how we explain and support [problem]. Could I ask you about the last time you dealt with it? This is research, not a sales call. It will take 20 minutes.”

Do not promise anonymity unless you can provide it. Explain how notes or recordings will be used, ask permission before recording, and handle personal data appropriately. If you use an AI transcription or analysis tool, remove unnecessary personal information and check the tool’s data terms first.

Step 3: ask about recent behavior

Use the same core questions in each 20-minute conversation. Follow interesting answers, but avoid teaching, correcting, or selling.

Seven practical interview questions

  1. Take me back to the moment you realized this needed attention. What happened?
  2. What was difficult, slow, risky, or frustrating about the situation?
  3. What did you try before looking for another solution?
  4. Where did you look for information or help? What words did you search or ask?
  5. Which options did you consider, including doing nothing?
  6. What made you trust one option and hesitate about another?
  7. What changed after the problem was solved—or what do you hope will change?

Useful follow-ups are short:

  • “Can you give me an example?”
  • “What happened next?”
  • “Why did that matter?”
  • “What do you mean by that word?”
  • “How were you handling it before?”

Avoid leading questions such as “Would an automated dashboard have helped?” Ask “What would have made the situation easier?”

Also avoid asking people to design your product. Their experience is evidence; your job is to decide how to respond.

Step 4: capture exact phrases in a free evidence sheet

A simple spreadsheet is enough. If you need a broader research plan, the SBA’s market-research guidance explains how direct customer research can complement existing market information. HubSpot also provides a more extensive market-research guide and templates. Create these columns: Interview, Trigger, Problem in their words, Practical or emotional impact, Previous workaround, Alternatives, Decision criteria, Objection, Desired outcome, Exact quote

Use Google Sheets, Microsoft Excel, LibreOffice Calc, or another tool you already have. One row can represent one useful observation.

Keep interpretation separate from quotation. “Needs visibility” is your summary. “I could not tell what I could safely spend this month” is the customer’s language. Save both, but do not confuse them.

After each interview, spend ten minutes adding the strongest evidence. This is easier than analysing five messy transcripts at the end.

AI can help cluster anonymized notes into themes, but it should not decide what customers meant. Check every suggested pattern against the original words. Never upload confidential customer information to a public AI tool without an appropriate legal and privacy basis.

Step 5: build a one-page Message Evidence Map

Review the sheet after five conversations. Highlight repeated ideas, strong contrasts, and surprising phrases.

Then complete this map:

1. Audience and situation
Who repeatedly experiences the problem, and in what moment does it become urgent?

2. Trigger
What event starts the search for help?

3. Core problem
How do customers describe the problem in plain language?

4. Cost of the problem
What time, money, risk, uncertainty, or emotional burden does it create?

5. Desired progress
What useful change are they trying to make?

6. Alternatives and objections
What else do they try? Why might they do nothing? What makes them hesitate?

7. Decision criteria and proof
What must they understand or trust before taking the next step?

8. Search and question language
Which phrases did they search, ask colleagues, or use repeatedly?

Now draft a message:

We help [specific audience in a recognizable situation] solve [problem in customer language] so they can [desired progress]. Our approach [credible distinction], with [relevant proof].

Treat the sentence as a draft, not a magic formula. Clarity and evidence matter more than filling every bracket.

Step 6: turn one insight into a small test

Do not rewrite everything at once. Choose one repeated pattern and test it where it matters most.

You could:

  • replace the homepage headline and supporting paragraph;
  • add an FAQ answering the strongest objection;
  • write one article around a repeated “how do I…” question;
  • revise the opening of a sales email;
  • add the proof people said they needed;
  • change a form call to action from a vague “Contact us” to a clearer next step.

Run the test for two to four weeks, or until you have enough relevant conversations to judge it. For a low-traffic business, qualitative signals may arrive before reliable conversion statistics.

Track a few signals:

  • Do prospects understand the offer faster?
  • Do sales calls begin with more relevant questions?
  • Do replies use the same problem language?
  • Does the updated page generate more suitable enquiries?
  • Does Search Console show impressions for relevant problem-based queries?

Free or low-cost tools can help: Google Search Console for organic queries, Google Analytics for on-site actions where consent and configuration are appropriate, and Microsoft Clarity for aggregated behavior patterns where privacy requirements are met. A simple CRM or spreadsheet can keep interview contacts and leads organized. If you later automate the follow-up process, automate only after the message is clear.

Use customer language for SEO and GEO (carefully)

Customer language is a strong starting point for search research because it reveals real questions and problems. It is not a substitute for checking search results or demand. This is also the principle behind Pain Point SEO: start with the buyer’s problem and intent rather than search volume alone.

For each repeated phrase:

  1. Search it and inspect the results.
  2. Note what the searcher appears to want: an explanation, steps, a comparison, a tool, or a provider.
  3. Check Google autocomplete and your own Search Console data for related wording.
  4. Group phrases that share the same intent into one useful page.
  5. Create the clearest answer you can, using real experience, examples, and limits.

If you want to explore search intent, keyword variations, and topic grouping in more detail, Ahrefs provides a comprehensive keyword-research guide.

For AI search and answer engines, make the page easy to understand and cite:

  • state who the guide is for;
  • define important terms;
  • give a concise answer near the top;
  • use descriptive headings and numbered steps;
  • include a realistic example;
  • distinguish evidence from assumptions;
  • add visible authorship, review dates, and sources;
  • keep important information in text, not only in images;
  • use structured data only when it matches visible content.

Google’s guidance recommends creating helpful, reliable, people-first content. Google also confirms that its normal SEO foundations continue to apply to AI Overviews and other AI search features. There is no special schema or formatting trick that guarantees inclusion.

A realistic example

Imagine a small bookkeeping consultancy serving founder-led creative businesses.

Its original homepage says:

Strategic financial support for ambitious entrepreneurs.

The sentence sounds professional, but it is difficult to picture.

In five conversations, three founders describe the same moment: they have money in the bank but do not know how much they can safely spend because tax, invoices, and upcoming costs are unclear. Two say they delay hiring or investing because they fear an unpleasant tax surprise. Several mention that monthly reports arrive too late to guide today’s decision.

The Message Evidence Map identifies:

  • Audience: owner-led creative firms without an internal finance person;
  • Trigger: a hiring, equipment, or tax decision;
  • Problem language: “What can I safely spend this month?”;
  • Desired progress: make decisions without waiting for year-end accounts;
  • Objection: fear that bookkeeping support will add meetings and jargon;
  • Proof needed: a clear monthly view and a predictable response process.

The business tests this homepage opening:

Know what your creative business can safely spend—before you make the next big decision.

It adds an FAQ about what the monthly view includes and publishes a guide called *How much can my small business safely spend this month? The message, sales conversation, and content now reflect one coherent customer problem.

This does not prove that every creative founder thinks the same way. It gives the consultancy a stronger, evidence-based hypothesis to test.

Common mistakes and how to avoid them

Asking for compliments
“Do you like our idea?” invites politeness. Ask about a real past situation instead.

Interviewing only happy customers
You will miss objections and alternatives. Include a lost, delayed, or non-buying prospect where possible.

Treating five conversations as market proof
Five interviews reveal language and hypotheses, not population percentages. Validate important decisions with more interviews, behavioral data, or quantitative research.

Summarizing too early
If every quote becomes “customers need efficiency,” the useful detail disappears. Keep exact phrases beside your interpretation.

Automating the thinking
AI can sort notes, but it can also flatten differences or invent confidence. Return to the source material before deciding.

Changing five things at once
You will not know which change helped. Test one meaningful message in one important place first.

Optimizing for traffic alone
A high-volume topic may have little connection to the problem you solve. Prioritize relevance, intent, and business usefulness alongside demand.

When to do it yourself and when to ask for help

Do the sprint yourself when the audience is accessible, the decision is narrow, and the consequences of a wrong conclusion are limited. Founders often benefit from hearing the language directly.

Professional support becomes useful when:

  • participants are difficult to recruit;
  • the decision involves a major investment, new market, or repositioning;
  • internal bias makes neutral interviewing difficult;
  • several customer segments give conflicting signals;
  • privacy, regulated data, or research ethics require specialist care;
  • you need statistically reliable estimates;
  • the findings must become a full strategy, website structure, brand system, or multi-channel campaign.

A good adviser should make the evidence clearer and the next decision easier—not create dependency or unnecessary process.

Conclusion

A clearer message rarely begins with a clever headline. It begins with careful listening.

Choose one decision, speak with five relevant people, record their exact words, build a one-page evidence map, and test one change. You will still have uncertainty, but it will be useful uncertainty: a grounded hypothesis that you can improve through real responses.

EMARINT supports startups with market research and marketing strategy, communication planning, branding, websites, search visibility, and sales enablement. If your customer evidence is scattered or difficult to turn into a practical message, EMARINT can help you bring the research and next steps into one clear system. You can also explore our practical startup marketing guides to continue building your marketing foundations.