Customer Research and the Job to Be Done

The job, not the demographic

Clayton Christensen and colleagues argued in Harvard Business Review in 2016 that firms have never known more about their customers yet innovate hit-or-miss, because they organize their knowledge around customer attributes and product features rather than around what the customer is trying to get done. The Christensen Institute's statement of the theory describes a job as the progress a person or organization is trying to make in a particular circumstance, with functional, social and emotional dimensions — and contrasts it with conventional marketing's focus on demographics or product attributes.

The circumstance is the part that matters most for marketing. 'Finance directors at mid-sized firms' is a demographic. 'A finance director whose board has just asked for monthly cash forecasts and who has three weeks to produce the first one' is a job in a circumstance. The second tells you when the buyer enters the market, what they search for, what they compare you to (often a spreadsheet, a contractor or doing nothing), and what success feels like — which is the raw material for the buying situations described in Lesson 1.1.

Jobs also expose your real competition. The thing a customer fires when they hire you is frequently not a direct competitor at all. Knowing that is the difference between a message that says 'better than Competitor X' and one that says 'the first board pack in a week, without a new hire'.

Interviews that collect behaviour

Nielsen Norman Group's guide to user interviews describes the method plainly: a one-on-one conversation to learn who people are, what their experiences are like, and what they need and value. Its practical advice is the part most teams get wrong. Ask about specific events rather than general processes — 'tell me about the last time you had to produce that report' — because recalling a particular incident nudges memory and produces usable detail. Use open-ended questions. Avoid leading ones; the article warns that the quality of the data depends heavily on the interviewer, and an interviewer who asks many leading questions gets answers that reflect the question.

A serviceable script for a marketing interview has five parts. The trigger: what happened that made you start looking? The alternatives: what else did you consider, including doing nothing? The decision: who else was involved, and what nearly stopped you? The words: how would you describe what we do to a colleague? And the outcome: what is different now? The fourth question is the most valuable for marketing, because it hands you the customer's vocabulary — usually plainer and more specific than yours.

Talk to recent buyers (memory is fresher), to people who chose a competitor or nothing, and to churned customers. There is no magic sample size for this kind of discovery work. Our rule of thumb — a recommendation, not a research finding — is to start with five to eight conversations per segment and stop when the answers stop surprising you, not at a quota.

The limits of what people tell you

Interviews are an attitudinal method: they collect reported behaviour, not observed behaviour. Nielsen Norman Group lists the consequences — memory is flawed, so people do not recall events fully, especially distant ones; participants do not know which details matter; and people describe what they think they did, or should have done. Never ask a prospect to predict their own future behaviour ('would you pay for this?') and treat the answer as a forecast. Ask what they did last time they had the problem, and what it cost them.

Pair interviews with behavioural evidence wherever you can: which pages people actually read before booking a call, which reply in a cadence produced the meeting, what support tickets say in customers' own words. When an interview and the behavioural data disagree, believe the behaviour and go back to ask why.

Making research a team asset

Research that lives in one founder's notebook decays the moment that founder is busy. The fix is a lightweight record: for each interview, the date, the segment, the trigger, the alternatives considered, three verbatim quotes and the job statement you would write from it. After every five interviews, write a one-page synthesis: the jobs you saw most often, the words customers used, and what surprised you.

Two rules keep it honest. Quotes are verbatim, never paraphrased into your own language — the value is precisely that they are not your words. And the synthesis records disagreements and outliers rather than smoothing them away; the one interview that does not fit is often the next segment.

Research of the outside world complements interviews but does not replace them. Reading analyst reports, competitor pages and community threads tells you what is said in public; only conversation tells you what made someone act.