Needs, Motivation and Jobs-to-be-Done Analysis

Why can't I just ask people what they need?

Because articulating a need requires a kind of introspective access nobody has on demand, so people answer with a solution instead, usually an incremental improvement on what they already use. The answer is genuinely informative about how the person describes their situation, and it is not a need. Elicit indirectly instead: look for workarounds they have built, points where they abandoned something, substitutes they use including non-category ones, consequences they paid, and compensating behaviours like over-checking or arriving early. A workaround is the strongest single indicator of an unmet need, because somebody spent real effort to compensate for something.

What is the difference between a want and a need?

Evidence of consequence. A want is something a person would take if offered. A need is something whose absence produces a consequence you can point at: a workaround, an abandonment, a complaint, a substitute, or a cost paid in money, time, effort or standing. A second test helps: would the person notice if it were removed? A preference disappears without a trace; a need leaves a hole. Candidates that fail both are preferences, and reporting them as such is honest and useful. Applying this test removes most of what usually appears on a needs list.

How do I write a job statement?

Describe the situation and the progress the person is trying to make, without naming a solution. "When I have just been told I need a procedure, help me work out what happens next, so I can tell my family something definite." If a solution appears in the statement, it is a feature request. Job statements are interpretations of evidence and should be labelled as such, with the evidence beneath them, never presented as something a participant said.

Is jobs-to-be-done a proven theory?

It is a framing, and an honest treatment says so. There is no agreed operational definition, no established measurement, and no body of evidence that people experience their behaviour this way. What it does buy is a real discipline: the job is stable while solutions change, so a framework built on jobs survives a technology shift that would invalidate one built on product attributes, and it forces you to notice non-category substitutes. It organises evidence well in some categories and strains badly in others, particularly habitual, hedonic and socially embedded consumption. Test it on two or three cases before committing, and say which it is doing.

What is laddering and how far should I go?

Laddering moves from what someone mentions (an attribute) to what it does for them (a consequence) to why that matters (a value). Ladder from a specific instance rather than a general opinion, or you get a general ladder that fits anyone. Stop at the last distinctive rung, usually the second or third step. Keep going and every ladder in every category terminates in security, freedom, belonging or self-esteem, which is true, universal and useless for a decision. Record which rungs came from the participant's own words and which are your inference, because laddering is where an analyst most easily supplies the rungs.

Why does the social dimension of a job matter so much?

Because it explains behaviour that looks irrational when only the functional task is considered. People will accept a worse functional outcome to avoid being seen as someone who does not understand, cannot cope, or needs help. A functional analysis will never find this, because the person will not volunteer it and the behaviour looks like a technology problem or a preference. Look for behaviour that is costly functionally and cheap socially, for concealment, and for what changes when other people are present.

How do I tell a stated barrier from a real one?

Four checks. Does behaviour change when the stated barrier is removed? If price falls and adoption does not move, price was not operative. Do people who face the barrier behave differently from people who do not? Does the barrier appear in the account of what actually happened, or only in answer to "why not"? And do non-users give the same barrier that users give retrospectively? Price is the most over-stated barrier in research because it is a socially safe answer that closes the conversation, while complexity, embarrassment and not knowing where to start are systematically under-stated.

Who should be in the sample for barrier research?

People who did not adopt, who left, or who chose something else. This sounds obvious and is routinely violated, because samples are drawn from customer lists. Barriers reported by users are the barriers users overcame, which is a different thing from the barriers that stopped everyone else. Where rejecters cannot be recruited, say plainly what the barrier findings cannot cover rather than presenting user recollection as though it explained non-adoption.

What are the four forces in a switching decision?

The push of the current situation, meaning what became intolerable; the pull of the new option; the habit holding the person in place, including inertia, sunk cost and arrangements already made; and the anxiety about the new option, meaning what might go wrong or has to be learned. The value of the framing is that the last two are usually stronger than the first two and are routinely absent from analysis, so interventions get designed to increase attraction when the operative constraint was fear of change. Two cautions: switching accounts are retrospective and partly rationalised, and the framing organises evidence rather than measuring it, so do not attach weights to forces nobody measured.

Should needs that conflict with each other be reported?

Yes, and they are usually the most valuable material in the study. Real people want a decision made for them and want to stay in control, want speed and want reassurance, want privacy and want help. A tidy needs list has had the conflicts removed, and the removal destroys the part that matters, because tensions are exactly where a proposition has to choose and where competitors are usually failing. Look at how participants resolved the conflict in practice; the resolution behaviour tells you which need dominates in that situation.

What should I do when the evidence for a need is thin?

Shorten the list. Do not raise the level of abstraction, which is the tempting move because an abstract need is unfalsifiable and therefore always supportable. A candidate with two mentions and no consequence evidence is a hypothesis with its validation named, not a finding. A short needs list with evidence and prevalence attached to every entry is something a product team can act on; a long abstract one is something they will politely ignore.