A founder can spend a weekend mapping competitors, ask a handful of friendly people “would you use this?”, skim a few TAM numbers, then decide the go-to-market plan is LinkedIn outbound.
That feels like market research. Usually, it is not.
It does not prove the target customer has an urgent problem. It does not prove they have tried to solve it. It does not prove they have budget. It does not prove you can reach them.
Research is not thinking harder. It is finding out.
TL;DR
The right type of market research depends on the decision you need to make next.
Use secondary research to understand the market, competitors, substitutes, categories, pricing, and regulation. Use primary research to learn what your specific customers do, buy, avoid, and complain about. Use qualitative research to understand why and how. Use quantitative research to understand how many, how often, and how much.
For startups, the useful order is often:
Idea stage: landscape research, competitor and substitute mapping, early customer interviews.
Problem validation: behavior-based interviews, workflow research, existing workaround research.
MVP stage: usability tests, sales conversations, pricing signals, activation and drop-off analysis.
Growth stage: channel tests, keyword research, surveys, cohort analysis, funnel diagnostics.
Design-partner readiness: evidence that a reachable segment has a painful problem and is willing to engage beyond praise.
For a broader foundation, read this guide to market research for startups. If you are already interviewing customers, use these customer research questions to avoid wish-based conversations.
What are the types of market research?
The main types of market research are primary research, secondary research, qualitative research, and quantitative research. Primary research is data you collect yourself. Secondary research is data someone else already collected. Qualitative research explains why people behave the way they do. Quantitative research measures how many people behave that way, how often, or how much.
These are not separate boxes. Useful startup research usually combines them.
Research type | Plain meaning | Startup use |
|---|---|---|
Primary research | Data you collect yourself | Customer interviews, sales calls, surveys, usability tests, pilots |
Secondary research | Data someone else collected | Industry reports, public datasets, competitor pages, app reviews |
Qualitative research | Explains why and how | Workflows, motivations, objections, triggers, buying context |
Quantitative research | Measures scale or frequency | Search volume, conversion rates, retention, survey results, pricing response |
For example, if you are exploring a B2B SaaS idea, you might use secondary research to map existing tools, pricing pages, job posts, and category language. Then you use qualitative primary research to interview people about the last time they solved the problem. Later, you use quantitative research to measure activation, conversion, retention, or willingness to pay.
The weak question is, “Should I do interviews or surveys?”
The better question is, “What decision am I trying to make, and what evidence would change my mind?”
Practical framework: market research methods by startup stage
Use this as the practical map.
Startup stage | Decision | Best-fit methods | Useful output |
|---|---|---|---|
Idea stage | Is this market worth exploring? | Secondary research, competitor mapping, substitute research, early interviews | Target segment, known alternatives, early pain patterns |
Problem validation | Is the problem painful and repeated? | Behavior-based interviews, workflow mapping, workaround research | Repeated pain, current spend, urgency, manual workarounds |
Segment selection | Which customer group can we reach? | Reachable-segment research, channel research, geography validation | Segment with pain plus a believable access path |
MVP stage | Does the product solve the job? | Usability tests, pilot calls, activation analysis, pricing conversations | Product fixes, usage signals, commitment signals |
Go-to-market stage | Which message and channel can create demand? | Channel tests, keyword research, landing page tests, win/loss calls | Repeatable acquisition signals and tactical diagnostics |
Design-partner readiness | Do we have enough evidence for deeper collaboration? | Interview synthesis, pilot data, buyer conversations, commitment tracking | A specific design-partner ask tied to a painful use case |
This is the difference between doing research and making research useful. A founder does not need a beautiful report. A founder needs decision cues.
Idea stage: start with the landscape
At the idea stage, secondary research is often underrated because founders want to jump straight into building or interviewing.
Before you ask people anything, learn the current landscape:
What do customers already use?
What categories already exist?
What language do competitors use?
What are customers complaining about in reviews, forums, and communities?
What substitutes do people rely on?
What changed recently that might make the problem more urgent?
Good secondary sources include competitor websites, pricing pages, review sites, job descriptions, public datasets, app store reviews, niche communities, and government data. The U.S. Small Business Administration guide to market research and competitive analysis is a useful starting point for market research and competitive analysis basics. For demographic and economic context in the U.S., Census data can help you understand population and business context, but it cannot prove demand for your specific product.
At this stage, do not obsess over TAM slides. A huge market does not mean your wedge is real. You are looking for the shape of the market.
A useful idea-stage output might look like this:
Target user: “RevOps managers at mid-market B2B SaaS companies.”
Pain hypothesis: “Manual reporting across CRM, product analytics, and finance tools.”
Current alternatives: “Spreadsheets, BI dashboards, internal analyst, agency support.”
Suspected trigger: “Board reporting, sales forecast misses, new VP Sales, messy CRM migration.”
Open risk: “Do they feel this as a budget-worthy problem or just annoying admin?”
That is already stronger than “AI reporting for revenue teams.”
Customer discovery: ask about behavior, not wishes
An easy way to ruin customer discovery is to ask people to predict themselves.
Weak question: “Would you use a tool that saves time on reporting?”
Most polite people will say yes. It costs them nothing. It gives you almost no evidence.
Better questions:
“Walk me through the last time you built this report.”
“What tools did you use?”
“How long did it take?”
“Who asked for it?”
“What happened when it was late or wrong?”
“Have you paid for anything to solve this?”
“What did you try before?”
“Why did that not work?”
The quality of customer interviews depends on asking about past behavior instead of hypothetical intent. Use this guide to customer research questions if you want a sharper question bank.
Weak note: “They said they would probably use it if it saved time.”
Useful signal: “In a hypothetical research note, a buyer says she regularly uses spreadsheets and automation tools to prepare a recurring leadership report.”
Weak note: “A few people liked the demo.”
Useful signal: “Someone asked for access, invited a teammate, shared sample data, and agreed to test it on the next reporting cycle.”
The second version gives you something to act on.
Competitor research should include substitutes
“No direct competitor does this exact thing” is not proof of demand.
Sometimes it means you found a gap. Sometimes it means the problem is not painful enough. Sometimes it means customers solve it with substitutes that do not look like software.
Substitutes can include:
Spreadsheets
Agencies
Freelancers
Internal ops people
Generic tools
Manual processes
Doing nothing
A consumer app may compete with entertainment, habit, boredom, or status. A B2B tool may compete with an internal analyst, a weekly meeting, or a spreadsheet nobody likes but everyone understands.
Competitive research should answer:
What does the customer use now?
What budget does that alternative come from?
What does the current solution do well?
What pain remains unsolved?
What would make switching worth it?
What would make switching too risky?
Founders often study direct competitors because they are easy to find. Buyers compare the full set of ways they can solve or ignore the problem.
Market selection: pain only matters if you can reach the people who feel it
A market is not just a group of people with a problem. It is a group of people with a problem you can reach.
This matters when founders define markets too broadly.
“Parents” is too broad. “SMBs” is too broad. “Europe” is usually too broad.
A founder might interview a small group in the UK, run one ad test in Germany, see weak results, and say, “Europe did not work.” That is not a clean market signal. France, Spain, Germany, the UK, and the U.S. can behave differently across language, pricing, trust, regulation, purchasing habits, and channels.
You do not need a massive international research program. You do need to avoid pretending one mixed signal explains five markets.
For market selection, research should answer:
Who has the pain?
Who can you reach?
Through what channel?
At what cost?
With what message?
In which geography or niche?
With what buying trigger?
The sharp question is not only, “What hurts?”
It is, “What hurts among the people we can actually access?”
MVP stage: use research to diagnose
Once you have an MVP, research changes shape. You are no longer only asking whether the problem exists. You are learning whether your product, message, pricing, onboarding, and channel are working together.
Useful MVP-stage methods include:
Usability testing
Sales and pilot conversations
Activation analysis
Drop-off diagnostics
Pricing conversations
Support-ticket review
Session recordings or product analytics
Design-partner feedback
If users sign up but do not activate, the answer is not automatically “no demand.” It could be unclear onboarding, wrong user, weak promise, missing integration, bad timing, poor data import, or a product that solves only half the workflow.
Usability testing helps here because it shows where users get stuck while trying to complete a real task. Nielsen Norman Group’s Usability Testing 101 is a useful primer if you need the basics.
The same rule applies to early channel tests.
If a cold email campaign gets very low opens, that does not prove the market hates the product. It may be a list-quality problem, a subject-line problem, a sender reputation problem, or a segment problem.
Weak early channel data is a diagnostic before it is a verdict.
Growth stage: quantitative research becomes more useful
In the early stage, qualitative research often carries more weight because you are still learning what game you are playing.
As you grow, quantitative research becomes more useful because you have more behavior to measure.
Growth-stage market research methods include:
Funnel analysis
Cohort and retention analysis
Win/loss interviews
Customer surveys
Keyword research
Channel testing
Pricing tests
Segment-level conversion analysis
Review mining
Sales objection analysis
Keyword research can be useful because it shows existing demand language. If people search for a problem, category, or job to be done, that can shape SEO, paid search, positioning, and product pages.
The specific tool matters less than the method. Compare intent, volume, difficulty, conversion potential, and fit with your segment.
Surveys can also help at this stage, but they are not magic. A survey is useful when you know who you are asking, why you are asking, and what decision the answer will affect.
A survey sent to a vague audience with vague questions produces vague confidence.
Weak notes vs. useful signals
Use this as a quick quality check on your research notes.
Weak note | Why it is weak | Useful signal |
|---|---|---|
“They liked the idea.” | Praise is cheap. | They asked for access, shared data, invited a teammate, or agreed to a pilot. |
“They said they would pay.” | Hypothetical money is not money. | They already pay for a workaround or accepted a paid pilot conversation. |
“No competitor does this.” | You may be missing substitutes. | Customers spend time or money on spreadsheets, agencies, internal work, or adjacent tools. |
“The market is huge.” | Size does not prove reachable demand. | A narrow segment has frequent pain and a channel you can access. |
“The ad test failed.” | One weak test has many possible causes. | You diagnosed audience, message, offer, landing page, channel, and economics separately. |
“We validated Europe.” | Too broad to be useful. | You tested specific countries or segments separately and know what changed by market. |
Research should make your next move clearer. If your notes only make the pitch deck sound better, they are not doing enough work.
How to choose the right market research method
Use this decision rule:
Name the decision.
Name the riskiest assumption.
Choose the method that tests that assumption with the least theater.
Define what evidence would change your mind.
Turn the result into a decision, not a report.
Examples:
Should we build this? If the risky assumption is that the problem is urgent and repeated, use behavior-based interviews and workaround research.
Which segment first? If the risky assumption is that you can reach this buyer efficiently, use reachable-segment and channel research.
Which market or geography? If the risky assumption is that unit economics and behavior are similar across markets, use geography-specific validation.
What should the MVP include? If the risky assumption is that users need certain features to solve the core job, use workflow interviews and usability tests.
Is onboarding broken? If the risky assumption is that users understand the value but fail to activate, use product analytics and usability sessions.
Which channel should we scale? If the risky assumption is that early traction is repeatable, use channel tests, funnel analysis, and win/loss calls.
Are we ready for design partners? If the risky assumption is that the pain is specific enough for collaboration, use pilot conversations, commitment tracking, and evidence synthesis.
Notice what is missing: “Do a survey because surveys are research.”
Surveys are useful when the question needs a broader pattern. They are weak when you are still discovering the right question.
When secondary research is enough
Secondary research is enough when you are trying to understand context:
Market categories
Competitor positioning
Pricing ranges
Public demand signals
Regulations
Demographics
Search behavior
Existing alternatives
Macro trends
It is not enough when you need to know whether your specific reachable segment cares.
A report can tell you a category is large or growing. It cannot tell you whether a small fintech startup will trust your new tool, switch from its current workflow, and prioritize implementation this quarter.
Use secondary research to avoid ignorance. Use primary research to test your specific bet.
When qualitative research is enough
Qualitative research is enough when you are still trying to understand:
The workflow
The language customers use
Why current tools fail
What triggers urgency
Who is involved in the decision
What “good enough” looks like
What people have already tried
You usually do not need statistical confidence to identify early patterns, but you should not treat a small qualitative sample as proof of market frequency. If several buyers describe the same painful manual process, workaround, and deadline pressure, you have a stronger problem signal than a pile of polite opinions.
Once you know what to look for, quantitative research helps you measure how common it is.
When quantitative research is worth doing
Quantitative research is worth doing when you have a clear question and enough relevant data.
Good uses:
Measuring activation by segment
Comparing conversion by channel
Testing pricing sensitivity
Measuring retention by cohort
Ranking problems among a known audience
Estimating search demand
Comparing usage frequency
Finding drop-off points in onboarding
Weak uses:
Asking strangers if they like an idea
Surveying a generic audience because interviews feel awkward
Producing fake precision for a messy early decision
Treating a small, biased sample as proof
Quantitative research can make weak thinking look scientific. Keep the decision close to the data.
Before design partners: what evidence should you have?
Formal design partners are not just people who like your idea. They should have a reason to spend time with you because your product might solve a real problem for them.
Before approaching them, your research should give you:
A specific target segment
A painful, repeated problem
Evidence of current behavior or workaround
A clear trigger that makes the problem urgent
Proof you can reach this segment
A narrow use case for the first product version
A reason this partner would benefit from early involvement
Some commitment beyond praise
That commitment might be data access, a pilot meeting, internal stakeholder introduction, workflow walkthrough, paid pilot discussion, or a concrete timeline.
The bridge from research to design partners is simple: good research gives you a specific ask.
Weak ask: “We’re building an AI tool for sales teams and would love your feedback.”
Stronger ask: “In our research, RevOps teams described spending several hours each week reconciling CRM and finance data for recurring reporting. We’re building a narrow MVP for that workflow. Could we map your current process and test whether this removes one reporting cycle?”
The second ask is better because it is grounded in a real job, a real workflow, and a reason to care. Treat the time estimate as illustrative unless it comes from your own interviews.
If you are turning this into a customer discovery kit, include the segment, pain hypothesis, interview notes, current workaround evidence, open risks, and the design-partner ask. Do not bury the decision in a long research document.
Common mistakes when choosing market research methods
Treating method labels as the work. “We did qualitative research” does not mean much if you asked hypothetical questions to the wrong people.
Looking only for confirmation. If your research could not change your mind, it was not research. It was rehearsal.
Skipping substitutes. Customers do not care whether a competitor fits your category definition. They care how else they can solve or ignore the problem.
Treating broad markets as one market. “Europe,” “SMBs,” and “parents” hide too much variation to guide an early startup.
Overreacting to one channel test. Early channel data matters, but it needs diagnosis. Bad results can come from targeting, creative, timing, offer, landing page, pricing, or the channel itself.
Turning notes into bloated reports. Your notes are for decisions. They should be cues you can use.
FAQ
What are the main types of market research?
Primary research, secondary research, qualitative research, and quantitative research. Primary means you collect the data yourself. Secondary means someone else collected it. Qualitative explains why and how. Quantitative measures how many, how often, or how much.
Which market research method should a startup use first?
Start with the method that answers your next decision. At the idea stage, that usually means secondary landscape research plus behavior-based interviews. With an MVP, look at usage data, usability tests, sales conversations, and drop-off points.
When should startups use surveys instead of interviews?
Use interviews when you are still learning the problem, language, workflow, and buying context. Use surveys when you already know the right questions and need to measure patterns across a relevant audience.
What is the difference between market research and customer research?
Market research looks at the broader market: customers, competitors, substitutes, categories, pricing, demand, and market structure. Customer research focuses on customer behavior, pain, workflows, motivations, and buying decisions.
How do I know if my research is strong enough?
Strong research shows real behavior: current workarounds, repeated pain, existing spend, urgency, switching triggers, usage data, or commitments. Weak research sounds like praise, opinions, hypotheticals, or vague interest.
Do I need market research before building an MVP?
Usually, yes, but not a huge research project. You need enough evidence to know who the MVP is for, what painful problem it solves, what customers do today, and what would count as a better outcome.
Can one failed channel test invalidate my startup idea?
Usually not by itself. A weak channel test may mean the audience, message, offer, landing page, list, creative, timing, or channel was wrong. Treat early channel results as diagnostic data before treating them as a verdict on demand.
What research do I need before approaching design partners?
You need a specific segment, a painful use case, evidence of current behavior, a believable access path, and a clear reason the partner should spend time with you.
