Market Research Best Practices

Market research works best when you combine talking to real people with a few simple, repeatable tests.

Start with customer conversations

Talk to real users before building. "Most people overcomplicate this and still get it wrong. Surveys and reports are nice, but real research is talking to potential customers and seeing if they will actually pay."

Bring a concrete prompt or prototype to get responses. "What works way better: bring a rough hypothesis or paper prototype... 'can I show you the rough version and get your honest reaction in 15 min?'"

Aim for 15–20 interviews for pricing and patterns. "15-20 interviews is better than 10-12 for pricing. patterns are clearer with more data."

Use small, focused surveys to validate

Use surveys to scale what interviews find, not replace them. "Interviews then survey to validate makes sense."

Choose appropriate methods by sample size (MaxDiff/Conjoint need larger n). "maxdiff is easier than conjoint if sample size is tight. conjoint needs bigger numbers."

Ask about real payments or past spending, not only hypothetical interest. "also ask what they pay competitors now, not just hypothetical willingness to pay. actual behavior is way more reliable."

Mine existing behavior and competitors

Look at bookings, reviews, and calendars to see real demand signals. "Start with proxies: look at similar listings near you, check occupancy, pricing, waitlists, seasonality. That gives you real behavior, not theory."

Read competitor reviews for exact customer language and pain points. "The real gold is in review mining (reading 50-100 reviews of your competitors...) Customers describe their exact problems in their own words."

Visit competitors and learn from people already succeeding in your space. "If your future competitors are pretty much booked out, then I would go and visit with them to get their advice. They are likely to have some great insights..."

Practical tests you can run quickly

Build one cheap prototype or landing page and measure clicks/signups. "write 3 core hypotheses, then run one tiny test per hypothesis before building anything big."

Offer to solve the problem manually for early customers to test willingness to pay. "The validation question isn't 'would you use this?' ... it's 'would you pay $X for this?' or even better 'can I charge you $X to solve this manually right now?'"

Recruit participants by embedding in communities or paying for their time. "post the idea in travel groups... if budget allows: pay for their time. UserInterviews or Respondent at $100-200/hr gets you motivated participants."

Use structured analysis and AI carefully

Structure data and create intermediate outputs before using AI for synthesis. "Then I apply those methods to create structured outputs... At this stage everything is still grounded in tables, charts, and quantified patterns - nothing AI."

Treat AI as a reasoning layer, not a replacement for raw analysis. "treating AI as a reasoning layer on top of solid quant/qual work is the right approach imo."

Spot-check AI summaries against raw data to avoid amplified bias. "The main risk I’d watch for is overconfidence in the intermediate outputs; if bias or framing creeps in earlier, AI will amplify it..."

Do a few interviews, run one small survey or landing-page test, and mine competitor behavior; those steps will give quick, actionable signals about demand and price before spending a lot on development.