# Market Research Methods for Pre-Seed Founders: Pick 3 URL: https://gotomarket-ai.com/journal/market-research-methods-pre-seed-founders Type: blog Locale: en Published: 2026-11-01 Updated: 2026-11-01 --- > Fifteen market research methods compete for your attention. Three are enough before Series A. Here is how to pick and sequence them. Most market research methods work. The trouble is that a pre-seed founder has two weeks, almost no budget, and a list of fifteen methods that all claim to be essential. Three of them are enough. Customer interviews, a behavior-based desk scan of your competitors, and one small survey will answer 90% of what you need before a seed round. The rest is sequencing. ## Which market research methods actually matter before Series A? CB Insights' [post-mortem analysis of failed startups](https://www.cbinsights.com/research/startup-failure-reasons-top/) puts "no market need" at the top, cited in 42% of cases. That is a research failure, not an execution failure. The founders built before they asked. The mistake is not skipping research. It is picking a method that cannot answer the question you have. Here is the map we use. Each method answers one type of question, costs a different amount, and fits a different stage. - **Customer interviews**: Question it answers: Why do they behave this way?; Cost: $0-300; Time: 2 weeks; Use it when: Always, first - **Competitor desk scan**: Question it answers: Who else owns this problem?; Cost: $0-100; Time: 2 days; Use it when: Before interviews - **Win/loss calls**: Question it answers: Why did buyers pick someone else?; Cost: $0; Time: 1 week; Use it when: You have 10+ lost deals - **Survey**: Question it answers: How many feel this way?; Cost: $50-500; Time: 1 week; Use it when: After 15+ interviews - **Community mining**: Question it answers: What words do they use?; Cost: $0; Time: 3 days; Use it when: Before writing copy - **Smoke test (landing page)**: Question it answers: Will they act, not just say?; Cost: $100-400; Time: 2 weeks; Use it when: You have a hypothesis - **Analyst reports**: Question it answers: How big is the segment?; Cost: $0-2,000; Time: 1 day; Use it when: Deck prep only The sequence matters more than the list. A founder who starts with a survey asks the wrong questions to 300 people. A founder who starts with interviews asks the right questions to 15. ## Why do customer interviews come first? Interviews are the only method that tells you what you did not know to ask. A survey can only confirm or reject the options you wrote down. An interview surfaces the option you never listed. Aim for 15 conversations with people who match your draft ICP, not friends and not investors. Most teams hit saturation, the point where you stop hearing new themes, between 12 and 20 interviews for a narrow segment. If every call sounds different at call 15, your segment is too broad. Split it and restart. ![Laptop on a table showing a customer discovery video call with a notebook beside it](https://fdzlnqpwsaniezitwiuw.supabase.co/storage/v1/object/public/cms-media/gotomarket-ai/2026-11/f7cef6-i2.webp) Three rules keep the data usable: - **Ask about the last time, not the next time.** "Walk me through the last time you tried to solve this" beats "Would you buy a tool that does X?" Every time. - **Never pitch in the first 20 minutes.** The first 15 minutes of the conversation are worth more than 10 slides of pitch. - **Record and transcribe.** Notes lose the exact phrasing, and the phrasing is the asset. It ends up in your landing page headline. Skip paid panels for this stage. A panel respondent is paid to finish, not to care. Recruit from your LinkedIn network's second degree, the Slack community where your buyer hangs out, or the people who left angry reviews on a competitor's page. ## How do you run a competitor desk scan in two days? This is the cheapest of all the market research methods, and the one founders skip because it feels too obvious. Do it before the first interview so you ask sharper questions. Build a table with five columns for each of 6 to 8 competitors: who they sell to, what they charge, what they claim on the homepage, what customers complain about in public reviews, and what they stopped saying in the last year. The last column is the useful one. Use the Wayback Machine to compare their homepage from 12 months ago against today. Two tools speed this up. A traffic and audience estimator shows where competitors get their visitors. An audience intelligence tool shows what your buyers read, watch and follow beyond your category. Worth the money if you sell into a segment with more than 10 serious competitors. Skip both if you have 3 competitors and a spreadsheet: a free week of manual scanning gives you the same answer. ## What do you do with win/loss calls when you have no sales team? Win/loss analysis sounds like an enterprise ritual, but it works with 10 data points. If you have lost deals, trial users who churned, or demos that went quiet, you already have the sample. Send a short, direct email: "We did not win your business. Could you give us 15 minutes to understand what we missed? No sales attached." Response rates are modest, so send 40 emails to land 10 calls. Offer a $25 gift card to push it higher. If you do have recorded sales calls, a conversation intelligence tool can surface the objections you hear most. It only pays off past roughly 50 recorded calls a month. Below that, listen to them yourself. ## When is a survey the right tool? After interviews, and only then. A survey answers "how many", so it needs hypotheses worth counting. Take the 3 or 4 themes that came up in at least a third of your interviews and turn each into one closed question. ![Flat-lay of printed survey charts, calculator, pen and index cards on a gray desk](https://fdzlnqpwsaniezitwiuw.supabase.co/storage/v1/object/public/cms-media/gotomarket-ai/2026-11/bae74c-i3.webp) Keep it under 10 questions and under 5 minutes. For a narrow B2B segment, 100 to 150 qualified responses beat 1,000 random ones. A founder survey sent to a community of 2,000 people at a 10% response rate gives you 200 answers for free. Qualify with a screening question first: job title, company size, and whether they own the budget. One test before you send: ask 3 colleagues to fill it in aloud. If anyone hesitates on a question, rewrite it. Hesitation means the question is ambiguous, and ambiguity ruins the data silently. ## Is a smoke test worth the ad spend? Sometimes. Interviews and surveys capture what people say. A smoke test captures what they do. You build a one-page site describing the product as if it existed, point $150 to $300 of targeted traffic at it, and measure how many visitors leave an email or click "start trial". It earns its place when you have a clear hypothesis and a single offer to test, for example "mid-market RevOps leads will pay $99 a month for automated pipeline reports". It wastes money when you are still exploring, because you cannot test five positions with $200. Set the pass mark before you launch. A 5% email capture from qualified traffic is encouraging, 1% is a no. Write the number down first, otherwise you will rationalize whatever you get. ## Where does community mining fit? Reddit threads, Slack groups, G2 reviews and Hacker News comments are free, public and unprompted. Nobody is being polite to a researcher. That makes them the best source for the exact words your buyers use to describe the problem. Spend three days on it. Copy 50 complaints into a spreadsheet, tag each with the job to be done, and count the tags. The top 3 tags are your messaging hypotheses, and the vocabulary goes straight into your outreach emails. The limit: communities over-represent the vocal 5%. Use mining to generate hypotheses and interviews to test them. ## How do you turn raw notes into an ICP? Collecting data is half the job. Most founders stall at synthesis because they have 15 transcripts and no method. Use affinity mapping: one insight per note, cluster by similarity, name each cluster in the customer's own words. ![Hands clustering sticky notes into themes during affinity mapping of interview insights](https://fdzlnqpwsaniezitwiuw.supabase.co/storage/v1/object/public/cms-media/gotomarket-ai/2026-11/8ab033-i1.webp) Then score each cluster on two axes. Frequency: how many of the 15 people raised it unprompted. Intensity: did they describe a workaround, a budget line, or a past purchase attempt. A problem that is frequent but has no workaround is a complaint. A problem with a workaround and a budget is a market. A note-taking workspace with a database view is enough for this. You do not need a research repository tool before 50 interviews. ## What should you skip? - **Focus groups.** A room of 8 strangers produces groupthink and costs $4,000 or more per session. Interviews give you cleaner signal for a fraction of the price. - **Paid analyst reports before the deck.** They size the market. They do not tell you why anyone would buy from you. - **Surveys before interviews.** You will measure the wrong thing with great precision. - **Anything that takes longer than 3 weeks.** Research that outlasts the decision it informs is a hobby. ## What would we run in a two-week sprint? Here is the framework. Adjust it to your context. - **1-2**: Method: Competitor desk scan; Output: 6-8 row comparison table - **3-5**: Method: Community mining; Output: 50 tagged complaints, top 3 themes - **3-10**: Method: 15 customer interviews; Output: Transcripts, 1 insight per note - **11-12**: Method: Affinity mapping and scoring; Output: Ranked list of problems - **13-14**: Method: One 8-question survey; Output: 100+ responses on the top 3 themes At the end you have a ranked problem list, the customers' own vocabulary, and a count for each theme. That is enough to write an ICP, a positioning draft and a first outreach sequence. If the numbers do not support your original idea, you spent $300 to find out instead of $300,000. Ready to turn that research into a sharp ICP? Run your interview themes through the ICP Generator at gotomarket-ai.com and get a first positioning canvas the same afternoon. ## FAQ ### How many market research methods does a startup need? Three are enough at pre-seed: customer interviews, a competitor desk scan and one small survey. Add community mining and win/loss calls if you have the time. Using more than five at once usually means you are collecting data instead of deciding. ### What is the cheapest market research method for a startup? Community mining and competitor desk scans cost nothing but time. Customer interviews are nearly free too, since you only pay for an optional $20 to $25 gift card per participant. A full two-week sprint can run under $300. ### How many customer interviews are enough? For a narrow segment, 12 to 20 interviews usually reach saturation, when new calls stop revealing new themes. If call 15 still sounds nothing like call 5, your segment is too broad and should be split. ### What sample size do I need for a startup survey? For a narrow B2B segment, 100 to 150 qualified responses are enough to rank problems. Quality matters more than volume: 150 respondents who own the budget beat 1,000 random clicks. ### Should I run a survey or interviews first? Interviews first. A survey can only count the options you already wrote down, while interviews reveal the options you missed. Use the survey afterwards to measure how common the interview themes are. ### When should a startup use paid research tools? When you have more than 10 serious competitors, more than 50 recorded sales calls a month, or an investor asking for a market size. Before that, a spreadsheet and manual scanning give the same answer.