How to do market research for a startup (without spending money)

Traditional market research is designed for companies with budget, time, and an existing product to test. Surveys. Focus groups. Conjoint analysis.

None of it applies to an early-stage startup with an idea and three months of runway.

What you need is different: fast, cheap, directional signal about whether a problem exists, who has it, and whether they'd pay to solve it. Here's the research process that gives you that — before you spend anything on building.

Step 1: Define the problem you're researching (not the solution)

The most common market research mistake: researching demand for your solution instead of the existence of the problem.

If you search for "AI scheduling tool" you'll find competing products. You won't learn whether the underlying scheduling pain is real.

Search for the problem: "scheduling conflict," "double booked," "can't find a time to meet." These searches return evidence of pain, not product catalogs.

Define your research question before you start: "Do [specific type of person] have [specific problem] badly enough to change their behavior to solve it?"

Step 2: Reddit — the most honest market research tool that exists

Reddit is unfiltered. People describe their actual experience, not what they think a researcher wants to hear.

How to use Reddit for market research covers the full tactical process. The short version:

Find the communities. Search for your problem. List every subreddit where relevant conversations happen. Look for subreddits with 50k+ members and active recent posts.

Search for pain signals. Within each subreddit, search:

• "I wish there was a way to"
• "does anyone else have trouble with"
• "is there an app that"
• "I've been doing this manually and it's killing me"
• "frustrated with [existing tool]"

Measure the signal. Upvotes indicate how many people share the pain. Comments indicate how acutely they feel it. A thread with 400 upvotes and 80 comments about a specific workflow problem is better market research than any survey.

Collect verbatim quotes. Copy the exact language people use to describe the problem. This language becomes your landing page copy, your email subject lines, and your product positioning — because it's how your customer already thinks about the pain.

Step 3: GitHub — for technical and B2B markets

GitHub issues are public records of unmet demand in software markets.

How to read GitHub issues for product inspiration explains the full methodology. What you're looking for:

• Feature request issues with 20+ thumbs-up reactions (each reaction is someone voting for your product idea with their cursor)
• Recurring issues across multiple repositories in the same category
• README files that say "roadmap includes X" — unbuilt features with public demand
• Abandoned repositories with significant stars: 200 stars on an abandoned repo is evidence of demand that wasn't served

Step 4: Competitor research

Understanding your competitors is not about knowing they exist. It's about understanding exactly where they fail their customers.

Where to find this:

• G2 and Capterra reviews (filter by 1-3 stars)
• Reddit threads titled "[Competitor] alternatives" or "[Competitor] vs [Competitor]"
• Twitter/X searches for "[Competitor] sucks" or "[Competitor] problem"
• App Store reviews for mobile competitors

What to look for: Not vague complaints ("it's complicated") — specific ones. "It doesn't integrate with X." "The pricing changes every year and we can't budget for it." "Support takes 3 days to respond and we're a small team."

Each specific complaint is a product requirement — something your market tells you must be solved before they'll switch. How to analyze competitors before building your product covers this in full.

Step 5: Search volume — how big is the opportunity?

Once you've confirmed the problem exists, estimate its scale.

Google Trends (free): Shows relative search volume over time. More useful than the absolute number — you want to know if the problem is growing. An idea in a growing search category has momentum working for it. An idea in a declining category is swimming upstream.

Keywords Everywhere (cheap): Browser extension that overlays search volume on Google. Gives you enough directional data to know if you're looking at 100 searches/month or 100,000.

What the numbers mean for a startup:

• Under 1,000 searches/month: very niche. Possible, but distribution will be hard.
• 1,000-10,000: viable niche. Specific enough to build for, large enough to support a business.
• 10,000-100,000: competitive. Requires strong differentiation to capture any share.
• 100,000+: crowded. You need a wedge, not just a better product.

Step 6: Talk to people — but the right way

Online research tells you what people say in public. Conversations tell you what they actually do.

Your goal is not to pitch. Your goal is to understand the workflow.

Ask:

• "Walk me through how you handle [problem] today."
• "What have you tried? What worked, what didn't?"
• "What would have to be true for you to switch tools?"
• "What does this cost you — in time or money — right now?"

Do not ask: "Would you use something like this?" The answer is always yes and means nothing.

Book 10 conversations before you write a line of code. If you can't get 10 conversations, you don't know where your customers are — and that's information you need before building, not after.

What "enough research" looks like

You have enough signal when:

1. You can describe the exact person with this problem in specific, non-demographic terms
2. You have 10+ verbatim quotes from real people describing the same pain
3. You understand the existing alternatives and exactly why they fail
4. You know roughly how many people are searching for this category every month
5. At least one person has described a workaround — proof the problem is worth solving

If you can check all five: you're ready to validate your specific solution. How to validate demand in 24 hours is the fastest next step — a time-boxed sprint that answers whether people will act, not just agree.

If you can check fewer than three: keep researching. Every hour of research now saves a week of building the wrong thing later.

Get market signal on your idea in 60 seconds →