How to Run Jobs to Be Done Interviews at Scale in 2026

Last updated: 23 September 2026

Quick Answer: To run jobs to be done interviews at scale, recruit recent switchers, walk each through one fixed timeline, and code the four forces consistently. Field within 90 days of the switch, while the story is still recallable. At hundreds of interviews, forces stop being anecdotes and become counts you can compare across segments.

Focus groups told a Detroit-area condo developer to add bay windows. Sales stayed flat. When Bob Moesta talked to actual buyers, the dining room table kept coming up: they could not move until they had decided what to do with it. The developer added moving and storage help, raised prices $3,500 and, per the Christensen Institute's account, grew its business 25 percent by 2007 as industry sales fell 49 percent.

The answer sat in what buyers did, not what prospects wanted, and that is why the method is hard to scale: only recent switchers can tell the story, and their memory of it fades fast.

What Does a Jobs to Be Done Interview Uncover That a Survey Cannot?

A jobs to be done interview uncovers the circumstances that made someone change, told as a sequence of events: what pushed them off the old way, what pulled them to the new one, what made them hesitate and what habit held them back. A survey records the tidier reasons people give afterward.

Those are the four forces of progress, as Moesta and Chris Spiek teach them:

1. Push of the situation: the event that made the current way intolerable.
2. Pull of the new solution: what the person pictured after switching.
3. Anxiety of the new: doubts that the new option would work.
4. Habit of the present: the comfort and sunk effort that kept them put.

Forces come out of the story. Nobody answers "what was your anxiety?" usefully, but almost everyone can say what they did the night before they bought.

Depth does not require a tiny sample. In one direct test of interviews at scale against surveys, Felix Chopra and Ingar Haaland ran 395 AI-assisted interviews and found they yielded richer and more robust conclusions than a traditional-size qualitative sample or a single open-ended survey question.

How Does the Switch Timeline Structure the Interview?

The switch timeline walks the respondent forward through dated events, from the first thought about changing to the first weeks of use. Each question asks what happened next, where and with whom, so forces surface from events, not opinions assembled on the spot.

Moesta and Spiek's timeline, condensed into five stages for a fixed guide:

1. First thought: when changing first crossed their mind.
2. Passive looking: noticing options without acting.
3. Active looking: comparing, asking around, trying.
4. Deciding: the event that ended the search.
5. Buying and first use: what the first days confirmed.

Moesta and Spiek budget 45 to 90 minutes. At volume, a fixed guide matters more than length, because identical stages let timelines compare line by line.

The Copyable Timeline Interview Guide

Timings total 40 minutes.

1. Frame (2 minutes): "I'm reconstructing what happened when you switched. Small details matter."
2. Anchor (3 minutes): "When did you buy it? Where were you? Who was involved?"
3. First thought (6 minutes): "When did you first think you needed something different? What was going on?"
4. Passive looking (5 minutes): "Before you searched, what did you notice?"
5. Active looking (8 minutes): "What did you compare? What almost made you stop?"
6. Deciding (6 minutes): "What happened right before you chose? What would have made you wait?"
7. First use (6 minutes): "What did the first week confirm or contradict?"
8. Close (4 minutes): "If you had not switched, what would you be doing now?"

Probe every stage with "What happened next?" and "You said [their phrase]; what did that look like?"

Why Recruit Recent Switchers Instead of Loyal Fans?

Recruit people who switched within the last 90 days, because only switchers can narrate a decision; loyal customers describe a relationship. Switchers are also scarce. Even when invited to compare every year, most people stay put: KFF found that among 11.3 million Medicare Advantage enrollees in 2019, 10 percent voluntarily switched plans for 2020. So find switchers by a dated event, not by asking a panel who changed brands lately.

Memory is the other constraint. Comparing self-reports with insurance claims for more than 6,000 employees, Short and colleagues found perfect agreement of 75 percent for monthly doctor visits but 30 percent for annual ones. Routine events blur within months, and passive looking is routine. The same perishability drives the critical incident technique, and it makes Moesta and Spiek's 30 to 90 day window a validity rule.

Three pools feed a switch program:

• Switched in: new customers, invited within weeks.
• Switched away: cancellations and downgrades, invited within days.
• Almost switched: trials and abandoned carts where anxiety and habit won.

Channel decides who answers inside the window. Alchemic interviews inside WhatsApp with no link and no app, by outbound AI phone call or by web link, publishes 57+ languages including Hindi, Tamil and Telugu, and runs managed fieldwork or bring your own sample across 14 markets including the USA and the UK.

Can an AI Moderator Run a Switch Interview at Volume?

Yes, if the guide is built around events and the system can follow the timeline rather than a fixed question order. A switch interview lives on "what happened next?" and on noticing when a date does not add up. How far a system will leave the guide to chase a date that does not fit varies by tool, so pilot a handful of timelines before fielding hundreds.

Three things change at volume:

1. Timelines run in parallel, so the last respondent is still inside the recall window.
2. The guide is fixed, which makes forces comparable across segments.
3. Interviewing in pairs becomes reviewing in pairs: the second listener Moesta and Spiek recommend moves to transcript review.

The risks are specific: a tidy reason instead of an event, a skipped stage, a date that contradicts an earlier one. Once it has the client's brief, Alchemic designs and tailors the discussion guide to handle these risks before fielding, rather than leaving that work to the buyer. See our note on the discussion guide for an AI moderator.

How Do the Four Forces Turn Into a Product Decision?

The forces become a decision once they are coded per interview and counted per segment. All four appear everywhere, so ask which is strongest among the switchers you want more of, and which stopped the people who almost switched.

Sorted alphabetically by force:

At 12 interviews these are impressions; at 300 across segments they are counts. If setup anxiety shows up in most almost-switchers and few switchers, onboarding is the product decision. Alchemic codes themes while fieldwork runs and ties each coded force to the verbatim or voice clip behind it, and its knowledge base carries into the next wave. Jobs also make sturdier segments than demographics, the step persona discovery and our guide to customer personas from real interviews cover.

Where Do Jobs to Be Done Interviews Fail?

Switch interviews fail where there is no switch to reconstruct, and scale does not repair that. Four cases where another approach is the better choice:

1. New-to-world categories. With no switchers yet, observing the current workaround tells you more.
2. Habitual purchases. An impulse buy has no first thought to recall; shopper observation works better.
3. Ranking needs across a market. Interviews find the forces; an outcome survey of the kind Tony Ulwick's method uses ranks and sizes them.
4. A few high-stakes buying committees. For eight enterprise deals, volume adds nothing; a skilled human pair reading internal politics may fit better.