Jobs-to-be-Done Interviews: the job, not the customer

Demographic and psychographic segments correlate with behaviour without explaining it. Knowing that a buyer is 34, urban and design-conscious does not tell you why she bought this week rather than last year, or what she would have bought instead. The job does: she needed to arrive at a client meeting looking prepared after an overnight flight, and the alternatives she weighed were a hotel iron and buying a shirt at the airport.

Three consequences follow. Competition is redefined — the competitive set is whatever else could do the job, which frequently includes a manual workaround, a spreadsheet, or doing nothing. Circumstance beats attributes: the same person hires different things for the same nominal product category depending on the situation. And the non-consumption case becomes visible, since the most common alternative to any product is that the job goes undone.

The framework distinguishes functional, emotional and social dimensions of a job. The functional dimension is usually easy to elicit and rarely explains the switch on its own; the emotional and social dimensions are where the actual decision often sits and where interviewing has to work harder.

JTBD in circulation is two related traditions with different methods, and practitioners frequently mix their vocabulary without noticing.

The Christensen tradition is narrative and causal: understand the story of a switch, in the circumstance that produced it. The popular statement is Competing Against Luck — reviewed in this Vision assessment — and the method is the switch interview.

The Ulwick / outcome-driven innovation tradition is quantitative: decompose a job into discrete outcome statements, then survey for importance and satisfaction to find under-served outcomes. See the assessment of What Customers Want: Using Outcome-Driven Innovation. Bettencourt and Ulwick set out the applied version in Service Innovation — A Jobs-to-Be-Done Guide.

They answer different questions. The narrative tradition explains why a switch happened; the outcome tradition tells you which parts of a job are under-served across a population. Using switch-interview data to make ODI-style prioritisation claims is a category error, and it is common. Note also that outcome-driven innovation is a trademarked methodology, which is worth knowing before describing your own work as ODI.

Eight interviews with owners of small firms who had changed accounting software in the previous two months.

The stated reasons at the start of each interview were uniform and useless: "the old one was clunky", "we needed better reporting". Backward timelining changed the picture.

In six of eight cases the first thought traced to a single event, and in five of those the event was the same kind: an accountant or bookkeeper had asked for something the incumbent could not produce, in front of someone whose opinion mattered — a bank, an investor, a co-founder. The push was not the software's deficiency in general. It was a specific moment of looking unprepared.

The anxiety force was the migration of historical data, and in every case the delay between first thought and purchase — a median of four months — was spent on it. Habit was maintained by the bookkeeper, who knew the old system and was not consulted until late.

The alternative set included two competitors, and also "ask the accountant to produce it manually each quarter", which three of the eight had done for months as a workaround.

The resulting job statement — "when someone whose judgement matters asks me for a figure I cannot produce, I want to be able to answer immediately, so I do not look like I am not in control of my own business" — is not about reporting features. The product implications were a migration service and a bookkeeper-facing onboarding path, neither of which appeared in the feature-comparison research the team had been running.

• Interviewing about the product. The switch interview is about the person's situation. Once it turns into feature feedback, it is a different study.
• Starting at the purchase. Without the backward timeline you get the justification, not the cause.
• Skipping anxiety and habit. Two of the four forces, and usually the two that explain the timeline.
• Job statements written before the interviews. Then confirmed by them. This is validation theatre.
• Jobs pitched too abstractly. "I want to feel in control of my finances" is true of everyone and directs nothing.
• Mixing the two traditions. Narrative switch data cannot support ODI-style prioritisation claims.
• Non-consumption ignored. The most common competitor is nothing, and it is rarely in the alternative set as recorded.

The practical test is whether the write-up contains dated, situated specifics — the day, the trigger, the delay, the workaround — or whether it contains generalities that could have been written before the fieldwork.

Beyond that: were participants recent switchers, is the interval from first thought to purchase reconstructed, are all four forces present, does the alternative set include non-consumption, and does the job statement name a circumstance rather than a desire?

JTBD is a practitioner framework with a thin peer-reviewed evidence base relative to its popularity. Much of the literature is by consultancies with a commercial interest in it, and claims about its predictive power should be read accordingly.

It relies on retrospective accounts, which are reconstructions. The backward timeline and the push for concrete detail reduce this, and do not eliminate it — people are constructing a coherent story of their own decision, and coherence is not accuracy.

It also does not size anything. Switch interviews explain a pattern; they cannot tell you how many people share the job, which is exactly what a prioritisation decision requires and why the survey-based tradition exists.

And it fits some categories better than others. Considered switches with a discernible trigger suit it; habitual, low-involvement and impulse purchases largely do not.

Outcome-driven innovation is the quantitative branch, decomposing jobs into outcome statements and surveying for under-served ones.

Problem or opportunity interviews cover similar territory with less prescribed structure.

Kano model exploration classifies which attributes delight and which merely satisfy, which is the natural complement once the job is understood.

Contextual inquiry observes the job being done rather than reconstructing it afterwards, and is the stronger method where observation is feasible.

Eight to twenty switch interviews produce a set of timelines that have to be compared: where the first thought sat, what the trigger was, what the anxiety was, how long the gap ran. The finding is in the convergence across timelines, and reading for it by hand across twenty transcripts is where most JTBD research quietly stops at three vivid stories.

Coding by force and by timeline position, then retrieving across interviews, makes the pattern visible and testable. Evidano supports jobs-to-be-done interviews as a named methodology and analyses against the force and timeline structure rather than as generic feedback. Writing the job statement is a judgement about what the interviews add up to, and it is the piece worth arguing about as a team.