Customer Discovery Questions That Reveal Real Buying Intent
Nobody on a discovery call says "I would never pay for this." They say it looks great, they'd definitely use it, send over a deck. Then they stop answering email. The questions that reveal real buying intent are about the past and the present, not the future: when the problem last happened, what it cost, what they already tried, and what they pay today. Future-tense questions collect compliments, and compliments don't close.
This article covers the questions. The full process, from planning the calls to acting on them, lives on our customer discovery page.
Why "would you use this?" fails
A hypothetical yes costs the other person nothing. Saying yes is polite, it ends the topic, and it gets them off the call on time. Rob Fitzpatrick's The Mom Test is built on this point: ask about their life, not your idea, because people will lie to you to be nice. The fix is to replace future questions with past ones. The past already happened, so the answers have details you can check: dates, names, numbers, consequences.
Start from the decision, not a question list
Before you write a single question, name the decision the calls are supposed to settle. Is this segment worth the next quarter? Is the problem painful enough that someone will pay to fix it? Then write questions whose answers would actually change that decision. A question whose answer changes nothing is chitchat with an agenda.
The questions, grouped by what they prove
Did the problem actually happen?
- "When did this last come up? Walk me through it."
- "What did you do about it?"
- "What did it cost you?" Hours, money, a missed deadline, someone's weekend.
If they can't point to a specific time it happened, you're discussing a theory, and theories don't have budgets.
What have they already spent?
- "What are you using for this today?"
- "What does that cost you, in money or in someone's time?"
- "Who maintains that spreadsheet, and how long does it take them?"
A workaround is intent you can see. People don't build Tuesday-night spreadsheets for problems they don't care about. Current spending, in dollars or in hours, is the strongest intent evidence a discovery call can produce, because it already happened.
Where does it rank?
- "What happens if you do nothing about this for six months?"
- "What else is competing for that budget this quarter?"
If honestly nothing happens in six months, you've found a real problem that nobody will pay to fix this year. Better to learn that in week two than after you build.
How do they buy?
- "Who else would be involved in a decision like this?"
- "How did the last tool your team bought get approved?"
- "What would kill this internally?"
Someone with real intent either knows these answers or offers to find out. Someone being polite changes the subject.
End with an ask that costs them something
Intent shows up when a person gives something up: time, reputation, data, money. End every discovery call with a concrete next step that's slightly uncomfortable to grant: thirty minutes with the person who owns the budget, an intro to the colleague who feels the pain daily, or real data to run through a pilot. A yes to any of those means something, because it spends their credibility, not just their politeness. "Send me a deck" is a no with good manners.
Grade the answers like evidence
Not every answer on the same call is worth the same.
- Strong: specific past behavior with consequences attached. The kind of answer that names the quarter it happened and the person who cleaned it up.
- Medium: a fair inference. They described the problem secondhand, or the cost is real but vague.
- Weak: compliments, hypotheticals, anything in the future tense.
Count the strong answers, and count distinct people behind each pattern. One enthusiastic person who repeats a point five times is one data point. How many interviews you need before the pattern is trustworthy has the stopping rule.
Don't leak the answer you want
"Wouldn't it be faster if this were automated?" Everyone says yes to faster. Leading questions manufacture intent, and manufactured intent is worse than no data, because you'll build on it. Ask "walk me through the last report you built" and you get the truth either way: either the pain shows up in the details, or it doesn't and you just saved yourself a quarter.
Write down the exact words
Memory inflates. By the time you debrief your cofounder, "maybe next quarter" has become "they're in." Keep the verbatim answer, who said it, and the timestamp, so your team can check the quote instead of trusting your recap. If the calls are recorded, analyzing the transcripts with AI covers how to do this without re-listening to everything.
Where UpSight fits
UpSight keeps discovery answers as evidence. Record calls with the desktop recorder, run built-in surveys when you can't get someone on a call, import audio, video and text files (mp3, mp4, vtt, srt, txt and more) from any tool, or have Otter send transcripts over automatically. Every answer points back to the quotes behind it, so you can check it: the exact words, the speaker, the timestamp, and a confidence label on each piece of evidence. When three calls mention the same workaround, the theme links to the three people who said it, not to a summary of them.
Pricing is published on the pricing page: Free at $0, Pro at $29 per month, Team at $39 per user per month. The Free plan covers your next round of calls.
FAQ
What are the best customer discovery questions for buying intent?
Ask when the problem last happened and what it cost, what they use and pay today, what happens if they do nothing for six months, and who else would be in the decision. Then end with a concrete ask. Past behavior and current spending are evidence; future-tense answers are not.
How do I know if a prospect actually intends to buy?
Two signs together: they've already spent something on the problem (money, hours, a workaround someone maintains), and they say yes to a next step that costs them something, like an intro to the budget owner. Either one alone is interest. Both together is intent.
Should I ask "would you pay for this?"
It's the weak form of a better question. What someone already pays is a fact; what they say they'd pay is a guess made to be agreeable. Ask what the problem costs them today. When you're ready to test price for real, quote one and ask for a commitment, which is validation rather than discovery. Customer discovery vs customer validation covers when to switch.
How many discovery calls before the answers mean anything?
Plan for 10 to 15 per segment, and stop when two or three calls in a row add nothing new. Fewer real conversations with open questions beat thirty calls full of leading ones. The full numbers and the stopping rule are in How Many Customer Interviews Is Enough?
Next step
Book five calls with one segment. Ask about the past, grade the answers, end each call with an ask that costs them something. If you want every answer kept with the quote behind it, run your next five discovery calls through UpSight, or see how it analyzes calls you've already recorded.
Try this on three of your own calls.
Free plan, no card. Every answer shows the quote behind it.
Start Free TrialFounders and product leads: see the customer discovery platform. Consultants: see customer discovery for consultants. Already have recordings? See customer call analysis.