How to Interview Customers About the Job They Hire a Product For 2026

If you have ever sat through a customer interview where every answer was a feature request, you already know why the jobs-to-be-done method exists. Learning how to interview customers about the job they hire a product for means reconstructing one real purchase from the inside out, in their words, until the progress they were trying to make is clearer than the product they ended up buying. Not a job interview for a job opening, not a product demo. This is research about a decision someone already made.

The mechanics are unglamorous. Recruit people who bought or switched recently, aim for conversations of 30 to 45 minutes, plan on 7 to 10 of them before the same patterns start repeating, and work inside a roughly 60-day memory window because that is about how far back most people can recall a purchase with any accuracy. Below is the method end to end, including the questions that open each stage up and the rewrites that keep you from contaminating the answers.

What You Need

Six things, and only one of them is the recorder. Everything else is preparation that determines whether the conversation produces evidence or just opinions.

A defined question, not a defined idea

Write down the situation you are investigating, not the feature you hope to validate. “Why do teams churn from our reporting product in month three” is investigable. “Do customers want saved views” is a question that smuggles your answer into the room.

A screener that recruits on behavior

You need three or four qualifying questions up front: roughly when did you buy or switch, what did you use before, and who else was involved. Recruit on behavior and recency, not on enthusiasm. Someone who loves your product and volunteers to chat will give you a flattering story with no purchase tension in it at all.

A recruiting plan, because this is the hard part

Forum discussions about this method keep circling the same frustration: teams lack a pool of recent purchasers and their CRM data is thin or unsegmented. Have a backup before you start. Recent support tickets, community members who posted about a switch, trial users who upgraded, and sales calls where the buyer mentioned a competitor all work as sources when the CRM does not.

A recorder and an explicit consent line

Record everything. Detailed note-taking pulls your attention off the participant and onto your keyboard, and one-on-one interviews plus a dedicated note-taker are exhausting to sustain. Say it plainly at the start: I am going to record this so I do not have to write while we talk, and nothing gets shared outside the team. Offer a copy of the transcript to anyone who wants one.

A neutral guide, not a script

Semi-structured is the working model. You need an ordered spine so the conversation has a shape, plus permission to abandon the spine the moment someone hands you a better story. Treating the guide as a rigid questionnaire is the single most common way these sessions go wrong.

A place to put the evidence

Either a shared document where quotes land within a few hours of the call, or a transcript folder per participant. Whatever you choose, decide before the first interview, because the pile of unanalyzed recordings is the most reliable way to waste a research budget.

Step-by-Step: How to Interview Customers About the Job They Hire a Product For

Choose a Specific Situation to Discuss

Anchor the whole conversation in one recent, concrete episode involving your category. Ask the participant to describe what happened, what they were trying to get done, and what happened next. Do not ask them to evaluate your product yet, and do not start with a question about your product at all.

What you are really doing is reconstructing a timeline of progress, which is the backbone of the method. The table below is the version I keep printed during the call, with the stage on the left, what you are trying to recover in the middle, and the question that opens it up on the right.

Stage	What you are trying to recover	Question that opens it up
First thought	The first moment the category entered their awareness	When did you first start thinking about buying something like this? Was it a weekday or weekend, morning or evening?
Passive looking	Undirected browsing and category awareness, often over months	What were you doing in the months before that? What did you notice, read, or search about?
Trigger	The specific event that turned passive curiosity into a problem	What happened that made it a problem rather than an idea?
Active looking	The evaluation: who they considered, what they compared, who they asked	Tell me about the first three things you looked at.
Decision	The final push, including what almost stopped them	What nearly stopped you from going through with it?
After	Whether the progress happened, and what they would do differently	Has it done the job you needed it to do?

You can tell this step worked when the participant’s account has dates, times, and names in it rather than generalities. If everything they say could apply to any purchase they have ever made, the situation is still too abstract. Ask for the specific evening, the specific conversation, the specific person.

Separate the Goal From the Product

This is the move most worth practicing. The underlying progress a customer is seeking is not the same thing as the product they currently use, and the gap between the two is where your research value lives. When someone tells you they need a project management tool, that is a solution, and it is the least interesting sentence in the interview.

Neutral prompts that pull the goal out include: what were you trying to change, avoid, accomplish, or feel at that point? what did you want to be true when this was all over? what were you doing instead before? what was the situation that made this worth solving? Ask about the outcome, not the object. A participant who says they want better reporting may turn out to need to avoid a specific recurring embarrassment in a meeting, and those two statements produce completely different products and messages.

It helps to write the three layers down separately as you listen, because they are easy to blend. The functional job is the practical progress: stop re-entering the same numbers twice a week. The social job is how the change looks to other people: be the person in the room who has the answer. The emotional job is what it feels like inside: stop dreading the Monday review.

How to Interview Customers About the Job Without Leading Them

Leading questions are not dishonest, they are just useless as evidence. They hand the participant your theory and invite them to agree or politely disagree, and both answers are worthless. Two habits do most of the damage: asking why, and reacting visibly to an answer.

The problem with why is that it pushes people out of the register where memory lives. While describing what happened, someone is partly feeling their way through a reconstruction. Ask why and the conversation switches to reasoning, and the reasoning is reconstructed on the spot, tidy and plausible and largely invented. Practitioners in customer research communities describe this switch from emotional to cognitive register constantly. Ask what instead, and the memory keeps its detail.

Instead of	Ask
Why did you buy that?	What was going on at work that made you start looking?
Why did you choose them over us?	What was the deciding factor between the last two you looked at?
Why did you leave your old tool?	What was the last straw, the thing that finally made you switch?
Would you pay for a feature that does X?	How do you handle X right now, step by step?
How important is this feature to you?	Tell me about the last time that came up.
Do you like the current dashboard?	Walk me through the last time you used it.
What would make you buy from us?	What would have to be true for you to start looking again?
Who else is involved in the decision?	Who had a say in this, and who had the loudest voice?

Three techniques do more work than any question wording. Context reinstatement means reconstructing the moment physically: put them back in the room, back at the desk, at the time of day. Many researchers report that a silence of 10 to 20 seconds is often what surfaces a detail nobody had mentioned. And the third is managing your own face: no raised eyebrows, no sighing at a competitor name, no laughing at an answer you expected. Warm and interested, but factually neutral, and if you catch yourself about to pitch, say thank you and write the objection down instead.

You can tell the step worked when the participant volunteers a detail you did not ask for. That kind of surplus material is the signal you stopped leading somewhere around minute twenty, which is also why short interviews miss the good stuff.

Map the Existing Alternatives and Forces

Now find out what they were doing before, including the option of doing nothing. People hire products to beat four forces: the status quo, the itch of a new solution, and the pull of the new solution versus the anxiety of giving something up. Ask what they tried before, what made the alternative attractive, what made it hard to live with, and what would make the switch worth the hassle.

Be specific about the alternatives. The real competitor list is usually longer than the research team’s, and it includes a spreadsheet, an intern, a contractor, a feature toggle in another tool, and doing the task badly but quickly. Teams that only map head-to-head competitors miss the fact that the status quo wins on switching costs and familiarity more often than it loses on capability.

Each interview should produce a picture of one situation, with the alternatives and what is holding them in place. When several interviews produce a map with the same shape, you have something usable. When the maps look nothing alike, either you recruited for different situations or your question was too broad to begin with.

Probe for Stakes, Emotions, and Trade-Offs

Now find out what is at stake if the job is done badly. Ask what happens if nothing changes, who else notices, what the cost of that is, and what the participant has already given up to solve it. Then ask what they would trade to keep the progress: money, time, learning something new, asking their boss for help, accepting a worse result in some other way.

Trade-off questions are unusually productive because they expose what the participant values, and values rank priorities without you having to ask. Nobody will say a feature is important. Almost everybody will describe how many hours they have already burned on the workaround, which tells you the same thing without the flattery.

Watch for the moment the story gets specific about money, deadlines, or a person. That is where the emotion of the situation sits, and it is usually described matter-of-factly, as a minor irritation, which is why it needs the follow-up rather than the reaction.

Test the Story With a Contradictory or Edge Case

Every account is a hypothesis until something breaks it. Ask when it went wrong, when the person considered not buying at all, when they nearly bought something else, and whether there is a day when they would not need a solution like this at all. If the need genuinely disappears in some circumstances, that is a finding about the job’s boundaries rather than a failure of the interview.

Exceptions are how you define the situation properly. A participant who once bought quickly and now stalls tells you what changed in their world, and that change is usually the trigger. A participant who went around the official buying process entirely tells you your funnel has a hole. One contradiction per interview is enough to sharpen the definition; do not turn the last ten minutes into an interrogation.

If a participant tells you they just needed a specific tool with no fuss, treat it as useful rather than disappointing. Ask what happened the last time before they decided to buy it, because the trigger still exists, and check whether the thing they bought replaced something they had already tolerated for years.

Synthesize the Job Into an Insight Statement

Compare the accounts side by side and mark what repeats: the circumstances, the motivations, the barriers, the outcome they wanted. Then write a provisional job statement in the standard form, and keep a second sentence for the situation and the emotional payoff.

The template is: when [situation], I want to [motivation], so I can [expected outcome]. Add a second sentence for the social and emotional layer: and so I can feel [feeling] about [identity or group].

A filled example: when my team grew past twelve people, I want to see where projects are slipping without asking three people for updates, so I can spend my week on the work instead of the status chasing. And so I can look prepared in front of leadership. Functional, social, emotional, all in one artifact.

Support the statement with evidence rather than adjectives. Quote the participant who described the Monday review dread, count how many of ten said the same thing about the trigger, and note the ones who disagreed. Disagreement is not noise to be smoothed away; it usually marks a segment boundary, and preserving it in the write-up is what stops the statement from becoming too vague to test. Hold the statement as provisional until the next round of interviews either confirms it or breaks it.

Common Mistakes

These are the errors that quietly ruin a study, paired with the correction that gets you unstuck.

Asking what customers want

You will get a feature request, and you will get it with confidence. The correction is to ask what they did about it last time instead, and to keep coming back to the outcome rather than the object. Requests are data about wording and priorities, not about what to build.

Treating praise as evidence

Compliments are noise. They cost the participant nothing and they are what people say to be polite. Ask for the time something went wrong, the workaround they built, the thing they gave up. Evidence has a cost attached to it, and that is how you tell it apart from a compliment.

Prompting with solution language

Saying a brand name out loud changes the rest of the answer, because the participant starts defending or attacking rather than recalling. Save brand questions for the alternatives-mapping step, and phrase everything else around situations, times, places, and people.

Pitching in the room

If you find yourself explaining how your product works, the session has become a demo and the data is contaminated. The move is to acknowledge the point, write it down verbatim, and thank them for being straight with you. If they ask what you want them to say, say that you would like the unvarnished version, including the parts that would be hard to hear.

Overgeneralizing from one participant

One vivid story is a hypothesis. It becomes a pattern when the same circumstance, motivation, or barrier shows up across several interviews in the same situation. Count before you conclude, and keep a list of what you are still unsure about rather than smoothing it into a neat statement.

Writing a job statement that describes a product

If your statement could be rewritten as a feature, it is not a job. When I want a dashboard that supports saved filters is a product. When my team grew past twelve people and I want to see where projects are slipping without asking three people is a job. The test is whether the statement survives a change of solution, which a feature statement does not.

Stopping when you have enough notes

The habit that pays off most: put quotes into the shared document within a few hours, while you still remember the tone. The second habit is counting codes. When new interviews stop producing a new code and participants start all sounding the same, you have hit saturation and can stop. Most teams reach that point somewhere between 7 and 10 interviews within the same situation, though new segments reset the count.

Frequently Asked Questions

Conclusion

Start smaller than feels rigorous. Recruit three people who bought or switched in the last two months, ask each one to walk you through a single recent episode from the first thought to the decision, and get to the progress before you mention your product once. Three conversations will not produce a validated job statement, but they will show you immediately whether your screener is finding real buyers or just polite fans. Then write the statement in the when, I want to, so I can form and take it back out to the next three people you interview.