--- title: "Customer research methods: 11 options and when to use each" description: "Eleven customer research methods sorted by the moment in the customer relationship, from support tickets to win-loss and churn interviews, and how to pick one." url: "https://raconte.ai/en/blog/customer-research-methods" --- # Eleven customer research methods, from first contact to cancellation Eleven customer research methods sorted by the moment in the customer relationship, from support tickets to win-loss and churn interviews, and how to pick one. September 25, 2026 Godefroy Key takeaways * Customer research is anything that teaches you what your customers need, why they buy, and why they leave. You run part of it yourself. The rest is already in your support tickets, sales calls, reviews and product analytics. * Pick the method by the moment in the customer relationship: before the purchase, during the decision, during use, or at cancellation. The moment narrows the choice to two or three methods. * When a salesperson and a lost buyer both explain a lost deal, their reasons often differ. Ask the buyer. * Twenty to thirty interviews bring out nine needs out of ten for one homogeneous customer segment. * Two one-hour interviews find about as many customer needs as one two-hour focus group. People in a group do not spark extra ideas off each other. Customer research methods are the ways a company learns what its customers need, why they buy and why they leave. You can choose from eleven. Four of them use evidence you already have: support tickets, sales calls, public reviews and product analytics. The other seven are research you run: discovery interviews, switch interviews, win-loss interviews, surveys, usability tests, feedback interviews and churn interviews. You can add focus groups when you want to hear people argue. The method depends less on your team than on the moment you are asking about: before the purchase, during the decision, while the customer uses the product, or when they cancel. This page sorts the eleven by that moment, with the question each one answers and what it misses. ## What customer research is, and how it differs from market and user research Customer research studies the people who buy from you, or could: their situation, their problem, how they choose, how they use what they bought and why they stop. Its output is a set of reasons you can act on. It follows your customers and prospects across the whole relationship, from the first contact to the cancellation, which is what separates it from market research and user research. * Market research sizes a market and maps its competitors. It answers “how big” and “who else”, often from secondary data you did not collect. * User research, or UX research, studies how people use a product. It overlaps with customer research during use, and leaves out the purchase and the cancellation. In practice, customer research belongs to several teams at once. Product runs discovery interviews, customer success tracks churn, sales owns the lost deals and marketing wants the words customers use. That spread is the main reason it goes wrong: each team reads only its own source, so nobody puts what the four teams learn side by side. If your question is about research technique in general, whatever the population, start from how [qualitative research methods](/en/blog/qualitative-research-methods) compare: interviews, usability tests, diary studies, surveys, focus groups and analytics. This page stays with customers, from the first contact to the cancellation. ## The eleven methods at a glance Method Moment What it answers What it misses Usually run by Support tickets and chats During use Where customers get stuck, in their own words Everyone who gave up without writing Customer support Sales calls and CRM notes During the decision The objections raised before the purchase The real reason, when the buyer keeps it to themselves Sales Reviews and public comments During use, after the purchase What customers tell other buyers Customers who are neither delighted nor angry, who rarely write one Marketing Product analytics During use What everyone did in the product Why they did it Product, data Discovery interviews Before the purchase Whether the problem exists and what it costs today Whether people will pay for your solution Product Switch interviews (jobs to be done) Right after the purchase What pushed someone to change, and what almost stopped them Customers who never considered switching Product, marketing Win-loss interviews Right after the decision Why a buyer chose you or a competitor Prospects who never reached the shortlist Sales, product marketing Surveys (CSAT, NPS) During use How many, how often, which option Why the score moved Customer success Usability tests During use Where people fail at a task Whether they wanted that task at all Design Feedback interviews During use Why a customer gave the score they gave, and what would change it How common that reason is across the base Product, customer success Churn interviews At cancellation What made the customer leave, and when it started Customers who are unhappy and still paying Customer success The table leaves out focus groups on purpose: they find no more customer needs per hour than interviews. ![Eleven customer research methods placed on the customer journey, from before the purchase to cancellation](/.netlify/images?url=_astro%2Fcustomer-research-journey-en.D0Ps0WLg.jpg&w=1376&h=768&dpl=6ad1cc428d394a5c89e0155b) ## Start with the evidence you already have Every company collects customer research without calling it that. Support tickets, sales call recordings, CRM notes, app store reviews and usage data pile up every day. They cost only the time to read them and give you something no research session can reproduce: what customers did, wrote or said with no researcher asking them anything. Support tickets are the richest of these sources. A customer who writes in describes a real episode, with the date, the screen and the words they use for the problem. Tag a month of tickets by theme and you get a first map of where customers get stuck. The limit is who writes: people who got stuck and still cared enough to ask. The customer who gave up quietly never appears there. Sales calls and CRM notes hold the objections raised before the purchase: the price, the missing integration, the competitor the prospect compared you with. Recorded calls are more reliable than notes, since a note records what the salesperson understood. The “loss reason” field in the CRM is the least reliable source of all, because the reason a salesperson logs often differs from the one the buyer gives. Public reviews show which words customers use when they recommend you to someone else, which is what your marketing copy should borrow. They skew towards the delighted and the furious, so read them for vocabulary rather than proportions. Product analytics records everyone, all the time. It can show, for example, that most accounts never make a second export. It records no intent, so it only shows where to look. To learn why, you need a conversation. None of the four sources answers your own question. Customers write tickets and reviews about whatever they choose, salespeople ask the questions that move the deal forward, and analytics records no reasons at all. To get the answer, including from the customers who never wrote in, you have to ask. A [voice of the customer program](/en/blog/voice-of-the-customer) repeats both steps every month: you read these sources, then you interview customers about what you still cannot explain. ![Evidence a company already has compared with research it runs: who answers, what it costs, who it misses](/.netlify/images?url=_astro%2Fevidence-vs-research-en.Dn77Bsop.jpg&w=1376&h=768&dpl=6ad1cc428d394a5c89e0155b) ## Before the purchase: discovery interviews Discovery interviews check that a problem exists, and matters, before you build the solution. You ask people who have the problem to tell you about the last time it came up: what they did, how long it took, what it cost, what they already use to get around it. Keep to the past: when you ask about the future (“would you use a tool that…”), people answer with compliments rather than facts. Product teams that hold one interview a week decide on fresh evidence rather than on memory. Most teams give up after three weeks. A set method and fifteen ready questions help you [keep up weekly customer discovery interviews](/en/blog/customer-discovery-interviews). ## During the decision: switch interviews and win-loss interviews Two methods study the purchase itself, from two angles. A switch interview asks a new customer how they came to change. A win-loss interview asks a buyer who just decided, for you or against you, why. ### Switch interviews show what pushed someone to change Switch interviews come from the jobs-to-be-done approach, which Clayton Christensen, a professor at Harvard Business School, made famous with a fast-food chain’s milkshakes. The chain first asked customers to describe the ideal milkshake, thicker, fruitier or more chocolatey, then changed the recipe to match. Sales [did not improve](https://www.library.hbs.edu/working-knowledge/clay-christensens-milkshake-marketing). One of Christensen’s fellow researchers then spent a day in a restaurant, noting who bought milkshakes and when. Forty percent were bought first thing in the morning, by commuters who took them to go. The next morning he interviewed them as they left. They explained that the milkshake kept one hand busy and lasted the whole drive, which a bagel did not. Bob Moesta, co-founder of The Re-Wired Group and one of the originators of the approach, runs these interviews along the buyer’s timeline, from the first thought to the first use. He reads the decision as four forces: the push of the current situation, the pull of the new solution, the anxiety about switching and the habit of the old way. In the interview, you look for all four in one real story. Ask a new customer when they first thought about changing, what happened that day, what they looked at and what nearly made them stay put. Run them within a few weeks of the purchase, while the person still remembers the order of events. ### Win-loss interviews tell you why a deal was won or lost A win-loss interview asks a buyer, shortly after they decided, how they compared the options and what tipped the choice. Most companies rely on the loss reason the salesperson logged in the CRM, a field that is often wrong. Clozd, a company that runs win-loss programmes, reports that sellers were wrong about the real reason for the outcome [85% of the time](https://www.clozd.com/blog/win-loss-analysis-whats-the-best-source-of-win-loss-data) when it compared CRM entries with buyer interviews. Corporate Visions, a sales training firm, gives [50% to 70%](https://corporatevisions.com/blog/the-business-case-for-win-loss-analysis/) of lost deals where sellers and buyers cite different reasons. Both figures come from vendors who sell this service, so take the size of the gap with caution. Its direction is consistent, though: “price” is the reason a salesperson logs most easily, and a buyer often says something else once someone outside the deal asks. That is why the person asking matters. A buyer tells the salesperson who lost the deal a polite version. Someone from product, a third party, or an interview the buyer answers alone gets closer to the real reason. ## During use: surveys, usability tests and feedback interviews Once someone is a customer, the question becomes how well the product works for them and what would make it better. ### Surveys count what you already know how to ask A satisfaction survey, CSAT or NPS, tracks a trend across your whole base. It works when you already know the options and want to know how many people pick each one. Response rates limit it: across 4,332 in-product surveys analysed by the survey vendor Survicate in 2025, the [median response rate was 9.98%](https://survicate.com/reports/survey-response-rate-benchmarks/). The nine customers out of ten who skip it are not a random sample. You can copy [a feedback form template for each moment of the relationship](/en/blog/customer-feedback-form): after a purchase, a support ticket, onboarding or a cancellation. The choice of score matters less than it seems. Timothy Keiningham, of the research firm Ipsos Loyalty, and three academic co-authors followed 21 companies and more than 15,500 survey interviews from the Norwegian Customer Satisfaction Barometer. NPS [predicted revenue growth no better](https://www.ipsos.com/en-us/paper-longitudinal-examination-net-promoter-and-firm-revenue-growth-awarded-2007-marketing-science) than the satisfaction measures it was meant to replace. A score out of ten has other flaws too, which matter when you [write a customer satisfaction survey](/en/blog/customer-satisfaction-survey). ### Usability tests show where people fail at a task A usability test asks a customer to complete a task while you watch: set up an integration, export a report. You see where they hesitate, which a survey or an interview never shows. Five people per round find most of the problems when you test again after each change. The test assumes the task is worth doing, so you learn nothing about whether the customer wanted it. ### Feedback interviews give the reason behind the score A feedback interview asks a current customer about a recent episode of use: what they did last week, what slowed them down, what they would change. Ask for the score, then spend the rest of the interview on what justifies it and what would move it up one notch. You learn why the number moved. ## At cancellation: churn interviews A churn interview asks a customer who just cancelled what made them leave, and when it started. The cancellation form only records a checkbox, “too expensive” or “missing feature”, ticked by someone already on their way out. Ask the same person to tell the story and you hear the sequence: the new manager who switched tools, the integration that broke in March, the three weeks nobody logged in. Two habits make churn interviews useful: * Run them within days of the cancellation. Two months later, the story has been simplified into a single reason. * Ask about the first moment of doubt, rather than about the day of the cancellation. People usually decide weeks before they click. One churn interview proves nothing. Ten, read side by side, show the patterns: a segment that leaves after onboarding, a competitor that keeps coming up, a price step that makes customers reconsider. ## Focus groups find no more customer needs per hour than interviews Focus groups appear in most lists of customer research methods. Abbie Griffin, then a marketing professor at the University of Chicago, and John Hauser, a marketing professor at MIT, compared them with one-to-one interviews in their 1993 study of the voice of the customer, counting how many customer needs each surfaced. Two one-hour interviews found about as many as one two-hour focus group (51% against 50% of the needs), and four interviews about as many as two groups (72% against 67%). They saw [no sign of group synergy](https://mitsloan.mit.edu/shared/ods/documents?PublicationDocumentID=5259), the idea that people spark ideas off each other. A focus group still earns its place when the disagreement is what you want to hear: two names for a product, two ways to package an offer, a room reacting to a price. For understanding one customer’s reasoning, interviews go further. ## How to choose a customer research method Write your question in one sentence, then find the moment it belongs to. The moment narrows the choice to two or three methods. Then decide whether you want to know what customers say or what they did. ```mermaid graph TD A["What do you want to know about your customers?"] --> B{"Which moment?"} B -->|"Before they buy"| C["Discovery interviews"] B -->|"While they decide"| D{"Did they just decide?"} D -->|"Yes, for us or against us"| E["Win-loss interviews"] D -->|"Yes, they switched to us"| F["Switch interviews"] D -->|"Not yet"| G["Sales calls and CRM notes"] B -->|"While they use the product"| H{"Do you know the options?"} H -->|"Yes, you need counts"| I["Survey"] H -->|"No, you need the reasons"| J["Feedback interviews and support tickets"] H -->|"You need to see them do a task"| K["Usability test"] B -->|"When they leave"| L["Churn interviews"] ``` Then pair the method with a source you already have. A churn interview explains what support tickets only hint at. A survey tells you how common a reason is once the interviews have named it. Most studies need two methods: one that shows what happened, and one that explains why. ## How many customers to talk to Griffin and Hauser, in the same study, found that interviews with twenty to thirty customers identify 90% to 95% of the needs of one homogeneous segment. Twenty interviews already surfaced more than 90% of what thirty did. When you look for themes rather than an exhaustive list of needs, around twelve interviews with one group of users are often enough. Two rules matter as much as the number of interviews: * Count each segment separately. Small businesses and enterprise buyers are two segments. * Ask everyone the same things, in roughly the same order. Twenty interviews that each ask something different leave you with twenty anecdotes. For win-loss and churn, there is no fixed count: interview every lost deal and every cancellation above a certain size, month after month, rather than running one study a year. ## Run customer research without a research team Most companies have no researcher, yet every interview method needs someone to recruit, schedule, run and write up each conversation. So teams fall back on surveys and CRM fields, where one more answer costs nothing, while every extra interview needs a person on the call. We built [Raconte](/en) so that nobody on your team has to be on the call. You write what you want to ask once, as a brief, then send a link by email or share it publicly. Each customer opens it in their browser when it suits them and talks for a few minutes. The AI asks the questions, follows up when an answer stays vague, and brings people back to a real episode when they drift into generalities. The ready-made templates include a discovery interview and two feedback formats, one after a demo and one qualitative NPS. A churn or win-loss brief takes a few lines to write. Every interview comes back with the transcript, the audio, a summary and per-message sentiment, plus reports that aggregate every answer to the same brief. You can pull them into your own tools through the [API and the MCP server](/en/docs/api). The first 60 interview minutes each month are free, with no credit card. Beyond that, you pay [by the interview minute](/en/pricing). Raconte covers the five interview methods: discovery, switch, win-loss, feedback and churn. An AI interview is voice only, so a usability test still needs someone watching the screen. A strategic account about to leave deserves a call from a person who can negotiate. And support tickets, sales calls, reviews and analytics are yours to read, whatever tool runs the interviews. ## Where to start Pick the moment where your team knows least. For most B2B teams it is the decision: the CRM shows “price”, but nobody has asked a lost buyer to confirm it in months. For a product that loses customers after onboarding, it is the cancellation. Then take the evidence you already have for that moment, a month of tickets or the last twenty lost deals, and read it before writing a single question, so you know what to ask. Interview ten customers of one segment with the same guide. Ten is enough to see the main reasons come back. Put the results next to the source you started from. To list every need of that segment, go on to twenty or thirty. ## Frequently asked questions ### What are the methods of customer research? The eleven that cover most needs are support ticket analysis, sales call and CRM analysis, review mining, product analytics, discovery interviews, switch interviews, win-loss interviews, surveys, usability tests, feedback interviews and churn interviews. Focus groups are often added to the list, although interviews surface as many customer needs per hour as a focus group. The first four use evidence you already have. The other seven are research you run. ### What are the 4 types of customer research? The usual split is primary research (data you collect yourself) against secondary research (reports and data someone else collected), crossed with qualitative research (reasons, in the customer’s words) against quantitative research (counts and scores). A more useful split for choosing a method is by moment: before the purchase, during the decision, during use and at cancellation. ### What is the difference between customer research and market research? Market research sizes a market and maps its competitors, often from secondary data. Customer research studies the people who buy from you or could: their problem, how they choose, how they use the product and why they leave. ### How do you conduct customer research? Write your question in one sentence, find the moment of the customer relationship it belongs to, and read the evidence you already have for that moment. Then pick the method that answers what the evidence leaves open, write one guide, and interview ten to thirty customers per segment, depending on whether you want the main reasons or every need. Put the findings next to the source you started from. ### How many customer interviews are enough? Griffin and Hauser found that twenty to thirty interviews identify 90% to 95% of the needs of one homogeneous customer segment, and twenty already surface more than 90% of what thirty do. Count separately for each segment. For win-loss and churn interviews, interview continuously, every lost deal and every cancellation above a size you choose. ### Who should run customer research? The team that will act on the answer should run it: product for discovery and feedback, customer success for churn, product marketing for win-loss. Win-loss and churn interviews work better when the person asking had no part in the deal or the account, since a buyer gives the salesperson who lost the deal a polite version. Ready to launch your first interview? Describe your interview, share the link: an AI agent runs it by voice with each participant. You get back the transcript and analysis. [Try for free](/login) ## Continue reading [![Customer discovery interviews: method and 15 questions](/.netlify/images?url=_astro%2Fthumbnail.w6TuGzWB.jpg&w=1376&h=768&dpl=6ad1cc428d394a5c89e0155b) September 24, 2026 ## Customer discovery interviews: method and 15 questions What customer discovery is, its four phases, 15 questions that keep people on facts, and how to hold a weekly interview cadence without drowning in logistics. ](/en/blog/customer-discovery-interviews)[![Qualitative research methods: which one for which question](/.netlify/images?url=_astro%2Fthumbnail.-XcEzSVW.jpg&w=1376&h=768&dpl=6ad1cc428d394a5c89e0155b) August 28, 2026 ## Qualitative research methods: which one for which question Four qualitative research methods and two quantitative ones side by side: the question each one answers, the one it cannot, and how to choose between them. ](/en/blog/qualitative-research-methods)[![Customer satisfaction survey: an example and its limits](/.netlify/images?url=_astro%2Fthumbnail.BVaPkpjh.jpg&w=1376&h=768&dpl=6ad1cc428d394a5c89e0155b) August 19, 2026 ## Customer satisfaction survey: an example and its limits A customer satisfaction survey template you can copy, what every question tells you, and the four blind spots of a score out of 10. ](/en/blog/customer-satisfaction-survey)