Learn · methods
How to interview customers at scale
Customer interviews produce the most actionable insight available to any business. A single well-conducted conversation with a customer who recently churned, upgraded, or abandoned a purchase tells you things that months of analytics cannot. The problem is that running interviews at scale has always required more moderator time than most teams have, which means most businesses run far fewer customer conversations than they should.
That constraint is no longer structural. This guide covers how to design customer interviews that produce reliable findings, how to run them at a volume that reflects what you actually need to know, and how to get from raw conversations to decisions that improve the business.
Why customer interviews produce better insight than surveys
Surveys measure what you already know to ask about. You define the questions, you define the response options, and customers pick from what you've given them. When the real reason a customer churned doesn't appear in your dropdown, they pick the closest option, and you spend three months solving the wrong problem.
Customer interviews surface what you didn't know to ask. A customer explaining why they nearly cancelled before deciding to stay will cover territory that no survey designer anticipated. The insight comes from following the thread, probing the unexpected answer, and letting the conversation go somewhere the interviewer didn't plan.
The tradeoff has always been scale. A survey reaches 500 customers in an afternoon. An interview reaches one customer in 45 minutes. That tradeoff made surveys the default for most businesses even when interviews would produce more useful information.
AI-conducted interviews change the economics. The same structured, probing conversation that a skilled researcher would conduct can now run with many customers simultaneously, producing consistent depth across every session without requiring moderator time per participant. Fieldwork runs these conversations with Sofi, with topic coverage tracked per session so analysis starts from structure rather than a cold transcript read.
What to interview customers about
Customer interviews are most valuable for questions where the answer depends on context, motivation, or experience that a survey cannot capture. The highest-value interview topics for most businesses:
Why customers chose you. Not the stated reason ("your pricing was competitive") but the actual sequence of events that led to the decision. What triggered the search? What alternatives did they consider? What was the moment they committed? This conversation produces the insight that sharpens positioning and shortens sales cycles.
Why customers stay. The feature customers say they value most is often not the feature that would cause them to leave if you removed it. Interviews that probe the "what would happen if" question surface the real retention drivers that CSAT scores and NPS miss.
Why customers leave. Covered in more depth in our churn guide, but the short version: exit surveys produce socially acceptable answers. Interviews produce honest ones.
What customers are trying to accomplish. The job your product does for customers is rarely the job you think you're doing. A customer using your project management tool to manage their anxiety about a chaotic workplace is not the same as a customer using it to coordinate a team. Both needs are real. Only one of them shows up in a feature request.
Where the experience breaks down. Not "did you find the onboarding helpful?" but "walk me through what happened the first week you had the product." The story customers tell about their early experience surfaces the specific moments where expectations diverged from reality.
How to design a customer interview
The design is what determines whether a customer interview produces insight or just a pleasant conversation that gets filed and forgotten.
Start with a specific question, not a topic. "Understand the customer experience" is not a research question. "Understand what stops new customers from reaching their first successful outcome within the first two weeks" is. The more specific the question, the more the interview can be structured to answer it.
Define what you need to learn at each topic. For each area the interview covers, write down what evidence would constitute a complete answer. "The customer has described a specific moment where they felt uncertain, named what caused that uncertainty, and indicated what would have resolved it" is a completeness criterion. "The customer has talked about onboarding" is not.
Open every topic with a question that invites a story. "Walk me through what happened when you first tried to set up the integration" produces a story. "How was the integration setup experience?" produces a summary. Stories contain specific, analysable detail. Summaries contain impressions that are difficult to act on.
Plan for the unexpected answer. The most valuable customer interview finding is usually something you didn't anticipate. Leave room in the structure for a thread to be followed when the customer says something that wasn't in the guide but clearly matters.
Running interviews at scale
The traditional constraint on customer interview volume is moderator time. A researcher who can conduct four interviews in a day, process the recordings, and write up findings is operating at a pace that produces maybe 20 interviews per month. For a business that needs to understand hundreds of churned customers, or wants ongoing insight from thousands of active ones, that pace is a structural limitation.
Running customer interviews at scale requires either more moderators or a different execution model.
Always-on interviews. Rather than running a study at a point in time, deploy a customer interview that runs continuously. Customers who complete onboarding, reach a key milestone, or show a churn risk signal are automatically invited to a short interview. Sessions accumulate over weeks and months, building a continuously updated picture of what customers are experiencing.
Post-event interviews. After any significant customer action: a purchase, a cancellation, a renewal, a support ticket: trigger a short interview that captures what happened and why. The timing matters: a customer interviewed within 48 hours of an event gives richer, more accurate answers than one interviewed two weeks later.
Segment-specific studies. Rather than trying to understand all customers at once, run focused studies on specific segments: customers who upgraded in the last quarter, customers who have been active for more than 12 months, customers in a specific industry or company size. Each segment study produces findings that are specific enough to act on.
Getting from conversations to decisions
The goal of customer interviews is not transcripts. It is decisions that improve the business. The path from one to the other has three steps.
Pattern identification. Read across sessions looking for what appears consistently. Not "one customer mentioned X" but "seven of twelve customers described the same experience at the same point in the journey." Patterns that appear across multiple customers from multiple segments are findings. Single-customer observations are data points worth noting but not acting on alone.
Root cause, not symptom. "Customers are confused by the onboarding" is a symptom. "Customers don't understand that they need to connect their data source before the dashboard populates, and the interface gives no indication that this is the missing step" is a root cause. Decisions made on symptoms produce solutions that don't work. Decisions made on root causes produce solutions that do.
Specific next action. Every interview finding should map to a specific action someone owns. "Improve the onboarding" is not an action. "Add a data source connection prompt to the empty dashboard state before showing any other options" is an action. The specificity of the action is a test of whether the finding was specific enough.
What this looks like in practice
A direct-to-consumer subscription business is losing 8% of customers in the first 30 days. The team has tried multiple onboarding email sequences without meaningfully moving the number.
A researcher designs a 15-minute interview focused on a single question: what happens in the first week that determines whether a new customer stays or leaves? The interview runs with 20 new customers who signed up in the past 30 days, split equally between customers who are still active and customers who cancelled.
The finding is specific: customers who cancel in the first 30 days almost universally describe the same experience. They signed up expecting the product to work immediately, encountered a setup step they didn't anticipate, and never completed it. Customers who stayed described the same setup step but had a specific trigger that got them through it: typically a notification that something was waiting for them, or a moment where they happened to return to the product and saw the setup prompt.
The fix is not a new email sequence. It is a persistent in-product prompt that stays visible until the setup step is completed. That prompt goes live. First-30-day churn drops from 8% to 5.3% in the following quarter.
The email sequences ran for six months without moving the number. The interview study took two weeks and produced the specific finding that led to the change that worked.
Frequently asked questions
How many customer interviews do you need to find reliable patterns?
For a well-scoped question with a relatively homogeneous customer group, 8 to 15 interviews is typically enough to reach saturation: the point at which new interviews stop producing new patterns. For questions that span multiple customer segments, or for churn studies where you want to understand multiple failure modes, 20 to 30 sessions gives you more confidence that you've captured the full range of what's happening.
What is the best way to recruit customers for interviews?
A direct, personalised request from a real person converts better than a generic survey invitation. Email or message the customer specifically, explain what you're trying to understand and why their perspective matters, keep the ask small (15 to 20 minutes), and offer an appropriate incentive. Customers who have had a strong experience with your product are more likely to agree. Customers who have churned are often willing to explain why if the request feels genuine rather than defensive.
How do customer interviews differ from NPS or CSAT surveys?
NPS and CSAT measure a customer's current sentiment as a single number. They tell you that satisfaction has dropped but not why. Customer interviews surface the specific experiences, events, and unmet expectations behind the number. The two methods complement each other: NPS identifies that something has changed, interviews explain what changed and why.
Can small businesses run customer interviews without a dedicated researcher?
Yes. The founder or a product manager asking genuine questions and listening carefully can produce useful customer interview findings without formal research training. The risk is confirmation bias: people asking their own customers questions tend to hear what they expect to hear. Structured interview guides, recording conversations for review, and interviewing customers across a range of experiences (not just the enthusiastic ones) reduce that risk significantly.
How often should a business run customer interviews?
Continuously, at low volume, is better than periodically at high volume. An always-on interview that captures 5 to 10 customer conversations per week produces a richer and more current picture of what customers are experiencing than a quarterly study of 50 people. The cadence should be calibrated to how fast the product and customer base are changing: faster change requires more frequent listening.
What is the difference between customer interviews and user research?
User research typically focuses on how people interact with a specific product or interface, often in a controlled setting. Customer interviews are broader: they explore what customers are trying to accomplish, why they chose you, what their experience has been, and what would make them stay or leave. Both are valuable. Customer interviews inform strategy and positioning. User research informs product and interface decisions.
Related on Fieldwork
- Why are customers churning?
- Customer experience research
- Run customer interviews at scale with Fieldwork
Last updated: 2026-07-11