Advertisement
Business

B2B Customer Discovery Interviews: Five Questions That Reveal Whether the Problem Is Real

By Rachel Thorne, Small Business Advisor • Nov 17, 2026 • 5 min read
B2B Customer Discovery Interviews: Five Questions That Reveal Whether the Problem Is Real

Early numbers flatter almost every business-to-business startup. Signups climb, a weekly active chart slopes upward, and it feels like the product is finding its market. What those charts never show is the story behind them: why one user went looking for help, what made the old way of working intolerable, and why another signed up, opened one screen, and never came back. Your first users can still answer those questions while the decisions are fresh. The window is short. Once the company grows, founders lose the ability to sit down with users one by one, and the chance to learn this cheaply disappears with it.

The mistake most founders make in these conversations is asking for opinions about a product that barely exists yet. Opinions are cheap and usually kind. What survives scrutiny is what a person actually did. The working rule behind every good discovery question is simple: ask about what actually happened, not what could happen in theory. A research question such as why trials stall during setup is a fine thing to wonder about, but asking it directly invites the person across the table to produce a tidy explanation. Asking them to describe the last time setup stalled gets you a story instead — the tools they had open, where they got stuck, who they asked for help. People routinely describe an idealised version of how they work when asked in general terms; specific incidents recover the shortcuts, workarounds, and failures that summaries leave out.

The five-question script

The sequence below comes from a Y Combinator lecture on talking to users, and it is designed to fit on a card beside the keyboard during a call. It is a route through a story, not a questionnaire to complete. A good answer to the second question can easily take most of the conversation, and that is fine.

First: what is the hardest part about doing the job your product helps with? Name the job, not the product. Suppose you built an uptime monitor: do not ask what is hardest about your alerting. Ask what is hardest about responding when a client-facing service goes down. The first version traps the person inside your interface. The second leaves room for them to say the hardest part is having to explain the incident to a non-technical client — which points at a completely different roadmap than better alert routing. Follow up with which part takes the most time, which part creates the most risk, who else gets involved, and how often it happens. Treat the first answer as a starting point, not a conclusion: a complaint like "reporting is frustrating" is still a summary, and your next job is to pin down a real instance of it.

Second: ask about the most recent time it happened. That single question is what turns a chat into evidence. Take the most recent instance and reconstruct it in order — what started it, what they did first, which tools they opened, where they paused or asked for help, and what happened next. Stay chronological. When the person jumps to a conclusion, gently pull them back to the sequence: what exactly did they click, who did they message, how long did it take? The details of one real episode are worth more than ten general statements about how things usually go.

Third: why was that hard? The point is the consequence underneath the surface problem. A slow approval step is annoying; a slow approval step that delays a client deliverable and costs a renewal is a problem worth paying to fix. Ask what the difficulty stopped them from doing. If the answer is "not much," you may be looking at a minor irritation rather than a buying problem, and that is useful to learn now rather than after building for it.

Fourth: what have you already tried — if anything? This question tests urgency and reveals the competition. If the person built a spreadsheet workaround, hired someone part-time, or pays for a tool that almost works, the pain is real enough to have produced spending. If they have done nothing at all, the problem may not be painful enough to support a product — or the existing alternatives may be so bad that doing nothing is rational. Ask what they tried first, and what happened when that attempt failed. The gaps in current solutions are where a new product earns its place.

Fifth: what do you not love about the solutions you have tried? This surfaces the trade-offs people already accept, which is the most honest form of product requirements. A tool that is powerful but slow, cheap but unreliable, or flexible but impossible to set up tells you exactly which compromise your product should attack first — and which ones your users will forgive.

Advertisement

What to do with twenty conversations

One interview proves nothing. The pattern across fifteen or twenty is the finding. After each call, write down the specific stories, not your impressions: the job, the incident, the consequence, the workaround, the complaint. When the same incident shows up in different companies with different words, you have a problem worth building for. When every story is different, you are still learning where demand is strongest, and the right move is more conversations, not more features.

Two practical rules protect the whole exercise. Say nothing about your product until the problem part of the conversation is finished; the moment you describe what you are building, every answer after it is contaminated by politeness. And listen more than you explain. The founder who talks through the interview learns what they already believed. The founder who asks about the past, follows up on specifics, and writes down what actually happened learns what the market will pay for.