Customer research is the practice of gathering real evidence about what customers need, want, and struggle with, instead of guessing. For a beginner or a small business, this does not require an expensive panel or research agency. Three accessible methods, short interviews, reading existing reviews and support conversations, and studying real search questions, can reveal most of what you need to refine an offer or write better marketing.
Designing neutral interview questions
A neutral question doesn't hint at the answer you want. 'Would you like a faster version of this product?' is leading, almost everyone says yes to a hypothetical improvement. 'Walk me through the last time you tried to solve this problem, what happened?' is neutral and produces a real story instead of a polite guess.
- Ask about past behavior, not hypothetical future behavior: 'Tell me about the last time you...' rather than 'Would you ever...'
- Ask open questions that can't be answered yes/no: 'What made you choose that option?' rather than 'Did you like it?'
- Avoid naming your own product or solution until late in the conversation, so answers aren't shaped by trying to be polite to you.
- Ask about workarounds: 'What do you currently do instead?' often reveals the real competitor, which is sometimes 'nothing' or 'a spreadsheet.'
Ten interview prompts (deliverable, part 1)
- Tell me about the last time you dealt with [the problem area]. What happened, step by step?
- What have you tried so far to solve this? What worked, what didn't?
- What's the most frustrating part of that process for you?
- If you could wave a wand and fix one part of this, what would it be?
- How much time or money does this problem currently cost you, roughly?
- Who else is involved when you make a decision like this?
- What would make you trust a new solution enough to try it?
- What's stopped you from solving this already?
- Where do you usually look when you're trying to solve a problem like this?
- Is there anything about this I haven't asked that I should have?
Reading reviews and support conversations
Existing reviews, whether of your own product, a competitor's, or an adjacent category, contain real customer language you couldn't invent yourself. Read reviews specifically looking for: repeated complaints (signals of unmet needs), the exact phrases people use to describe their problem (useful for ad copy and headlines), and what people compare the product to. Support conversations, emails, chat logs, call notes, if you have an existing product, are even richer, because they capture real friction in the customer's own words at the moment it happened.
Stated preferences vs observed behavior
People are not always accurate predictors of their own future behavior. Someone might say in an interview that price is their biggest concern, while review and behavior data shows people in that segment actually churn due to poor onboarding, not price. Whenever possible, weigh what people say against what they actually do (returns, cancellations, usage data, abandoned carts). When the two conflict, behavior usually tells you more about priorities, though stated reasons can still reveal language and framing worth using in your marketing.
| Source | Shows you | Watch out for |
|---|---|---|
| Interview answers | How people explain and frame their experience | Politeness bias, forgetting details, guessing about the future |
| Reviews / support logs | Real language and friction points, often at the moment of frustration | Only the most extreme (very happy or very angry) customers tend to write reviews |
| Behavior data (if available) | What people actually did, regardless of what they said | Doesn't explain why, needs interviews to interpret |
Using search questions responsibly
Looking at what people actually search for, via keyword research tools, 'People also ask' style results, or community forums, can reveal real demand at scale. Use this as a supplement to direct customer contact, not a replacement for it; search volume shows that a question exists, not that your specific offer answers it well, and not every searcher is a realistic customer for your business.
Research-synthesis table (deliverable, part 2)
| Theme | Evidence source(s) | How many mentions / strength | Priority |
|---|---|---|---|
| e.g. Pricing clarity confusion | 3 interviews, 6 reviews | 9 mentions | High |
| e.g. Slow customer support response | Support logs | 12 tickets in 30 days | High |
| e.g. Desire for a mobile app | 2 interviews only | 2 mentions | Low (watch, don't act yet) |
Consent and handling customer information sensibly
When interviewing real customers, tell them plainly that you're doing research to improve the product or service, ask permission before recording, and keep notes limited to what's relevant for the research purpose. This is basic good practice, not a substitute for your own legal or compliance obligations, which vary by location and should be checked separately if you handle sensitive personal data at scale.
Common mistakes
- Asking leading questions that confirm what you already believed.
- Interviewing only happy existing customers and skipping people who churned or never bought.
- Treating five enthusiastic interview answers as proof of large-scale market demand.
- Ignoring the exact words customers use and replacing them with internal jargon in marketing copy.
- Doing one research round and never repeating it as the business or market changes.
When informal research isn't enough
For decisions with high financial stakes, a major product pivot, a significant price change, entering a new market, five to ten informal interviews are a useful starting signal but not sufficient evidence alone; consider larger-scale surveys, usability testing, or professional market research before committing significant budget. For very fast-moving, low-cost experiments, like testing a new headline or hook, informal research plus a small live test is usually proportionate.
- Neutral, open-ended interview questions reveal more than leading or yes/no questions.
- Reviews and support conversations are a free, often underused source of customer language and pain points.
- Stated preferences (what people say) and observed behavior (what people actually do) can differ; weigh both.
- Search questions show real, large-scale intent, but should be treated as one data source among several.
- Synthesize findings into a prioritized list of needs, not just a pile of quotes.



