How To Do A Market Study

The actual process, not the textbook version

A market study is just structured guesswork with receipts. You're trying to figure out whether a group of people will actually pay for something before you spend money building it. Most people overcomplicate this because they think they need a massive survey or a fancy report. You don't. What you need is a clear question, a source of real human behavior, and the discipline to not lie to yourself about the results. I spent three years doing this for early-stage startups, mostly in SaaS and DTC e-commerce. The mistake I see repeatedly is starting with the answer. Someone falls in love with their idea, then runs a "study" to confirm it. That's not a market study, that's confirmation bias with extra steps.

How To Do A Market Study Without Wasting Two Weeks

Start with the specific decision you need to make. Not "should we enter this market" — that's too broad to be useful. Pick something narrow like "will dentists in the Midwest pay $49/month for a scheduling tool that syncs with Open Dental?" Narrow questions force narrow research. Vague questions give you vague answers and a thick report nobody reads. From there, pick your data sources in this order: actual transactions first, stated intent second, demographics last. If you can find people already paying for a substitute solution, that's your strongest signal. Purchase history doesn't lie the way survey responses do. A competitor's pricing page, their G2 reviews, their Reddit threads, their customer support complaints — all of that tells you what people actually tolerate. Tolerance is a real metric. If people are complaining about price, you have room to undercut. If they're complaining about features, you have room to differentiate. If nobody's complaining and nobody's leaving reviews, that's either a peaceful market or a dead one and you need to dig harder. I remember working on a project for a client who wanted to launch a meal prep subscription for truckers. The obvious research path would have been a survey about healthy eating on the road. Instead, I pulled data from Amazon bestsellers in the long-haul nutrition category, scraped Truckers Report forum threads for anything mentioning food delivery or meal planning, and called three restaurant supply companies that sell to truck stops to ask what their bulk buyers actually order. Turns out the biggest demand wasn't for healthy meals, it was for shelf-stable protein options that didn't require refrigeration. The original idea had a market, just not the one my client imagined. That pivot saved them probably $80,000 in wasted inventory.

Tertiary and primary data, used correctly

Secondary data gets you to the starting line. Industry reports from Statista, IBISWorld, or even free Census data will give you TAM numbers and growth rates. These are useful for sizing and for sounding credible in a pitch deck. They're useless for deciding whether your specific offering will work. The problem with secondary data is that it's always looking backward. By the time a report says a market is growing at 12% annually, the growth might already be priced in or the conditions that drove it might have shifted. Primary data is where the actual work lives. Surveys, interviews, landing page tests, pre-sales. A well-designed landing page with a clear value proposition and an email capture or pre-order button can tell you more in 48 hours than a 200-question survey ever will. People clicking "sign up" when they think they're getting something for free is cheap and fast. People handing you actual credit card info is expensive and slow, but it's the kind of data that prevents disasters. There's a specific technique I use that most people skip: the fake door test. You build a simple page describing your product as if it exists, run $100 to $300 of ads to it, and measure the click-through rate to a "buy" or "sign up" button that leads to a "coming soon" message. A 3% to 5% click-through rate on the buy button is a strong signal. Below 1%, you probably have a messaging problem or a demand problem, and it's cheaper to find out now than after you've built the thing.

Defining your total addressable market properly

TAM, SAM, SOM. Everyone knows the acronyms. Very few people calculate them correctly. The standard mistake is using top-down numbers from a research firm and calling it your market. If Gartner says the global project management software market is $15 billion, that doesn't mean you can capture any meaningful slice of it. You need bottom-up math. Take your ideal customer profile and multiply it by the number of those customers in your reachable geography, then multiply by the annual revenue you expect per customer. That's your realistic upper bound. The number will be smaller than you want, which is the point. A market study that makes your idea look smaller than you hoped is doing its job correctly. I've seen founders treat SAM as their revenue target. It's not. SAM is the segment of TAM you can physically reach with your current distribution channels. If you're building a mobile app for pet owners in Brazil, your SAM is not the global pet tech market. It's the number of smartphone-owning pet owners in Brazil who pay for pet services, times your expected annual revenue per user. Do the arithmetic out loud. Write down every assumption. Anyone reviewing your study should be able to dispute a single assumption and see how the number changes.

Competitive analysis that isn't just a feature checklist

List every alternative your customer could choose instead of you. That includes direct competitors, but also the default option people use when they do nothing. The default option is almost always someone's biggest competitor and it's almost never on your radar. If you're building a meditation app, the default option is sleeping. If you're building a bookkeeping tool for freelancers, the default option is throwing receipts in a drawer until April. For each competitor, find their pricing, their review sentiment on G2 or Trustpilot, their marketing angles, and their customer churn signals. Churn signals are usually in the negative reviews. Look for patterns. If three different products in your space get complained about the same way, that's an underserved need. If every product gets the same complaints, that might be an unsolvable problem and the market is just hard. There's a counter-intuitive thing about competitive analysis that bears repeating: the fewer competitors you find, the more suspicious it usually is. A completely empty market is rarely an opportunity. It's usually a graveyard. Either the demand isn't there, or the economics don't work, or it's been tried and killed already. A crowded market with bad options and frustrated customers is often a better entry point than a blue ocean with zero buyers.

Customer segmentation beyond demographics

Age, location, and income are easy to measure but terrible for predicting behavior. Psychographic and behavioral segmentation cuts deeper. Group people by what they're trying to accomplish, not what they look like. A 28-year-old freelancer and a 54-year-old freelancer might have identical problems with your product even though their demographics are completely different. I once worked on a market study for a B2B scheduling tool and the initial segmentation was by company size. Small teams, medium teams, enterprise. The data was all over the place because company size had almost no correlation with actual scheduling needs. When we re-segmented by scheduling complexity — people who just need basic availability sharing versus people who need multi-resource coordination with approval workflows — the product-market fit signal became crystal clear. The complex schedulers were willing to pay 4x more and had 10x lower churn. Company size was noise.

What this approach cannot tell you

No market study can predict whether your execution will succeed. A strong market signal combined with weak product, poor distribution, or bad timing will still fail. A market study also cannot validate an idea that depends on network effects. If your product only becomes valuable when many people use it simultaneously, surveys and landing pages will lie to you because nobody wants to join a platform they believe will be empty. In those cases, you need to bootstrap the network separately, usually by targeting a very small niche first and expanding outward. Market studies are also vulnerable to recency bias in fast-moving categories. If you're researching AI tools in 2025, data from six months ago might already be obsolete. The market moved faster than the study. In those situations, treat your findings as directional rather than definitive and build in frequent re-evaluation points. Check your assumptions every quarter, not just at the beginning. If you're building something in a regulated industry like healthcare or fintech, a market study alone is not enough. You need compliance analysis baked in from the start. A product that fits the market perfectly but can't clear regulatory hurdles is not a market fit, it's a legal problem wearing a market fit costume.

The final step most people skip

Write a one-page summary of your findings before you do any further analysis. One page. Three sections: who the customer is, what they currently do instead of your solution, and what evidence suggests they would switch. If you can't fill out that page in under 30 minutes, you haven't done enough research yet. If you can, you have enough to make a go-or-no-go decision. Anything beyond that is usually just academic padding.