Jobs to be done explained by Bob Moesta, co-creator of the framework

Executive overview

Most product teams study what customers say they want — but people can't accurately report why they buy. Jobs to be done (JTBD) is a demand-side framework built on understanding the struggling moments that cause people to change behaviour, not the features that attract them.

The core shift: stop asking what people want and start understanding the context and outcome driving them to act. Products aren't bought — they're hired to make progress. The competitive set only becomes visible when you look from the demand side, not the supply side.

Competing against the wrong things is the default; the job reveals the real competition.

What jobs to be done actually means

• People don't buy products, they hire them to make progress in a specific context toward a specific outcome
• Snickers competes with protein shakes and Red Bull (missed-meal replacement); Milky Way competes with a glass of wine or a run (post-emotional comfort) — not with each other
• A struggling moment causes demand; the product doesn't create it
• Supply and demand are less connected than assumed — struggling moments can exist for years before anyone builds a solution
• Southern New Hampshire University found 50–60 anomalous online-only students in 2010; studying them revealed a job that scaled to 200,000+ students

The four forces of progress

Behaviour change only happens when the first two forces exceed the second two:

• F1 — Push: the context making the current situation intolerable
• F2 — Pull: the appeal of a new outcome or solution
• F3 — Anxiety: fear of the new, triggered by anything unfamiliar
• F4 — Habit of the present: the inertia of the existing product or behaviour

Adding features increases F2 but also F3 — anxiety rises with complexity. Reducing friction (F4) often does more than any new feature. Bob raised condo prices to include moving and two years of storage, removing a key friction point; sales increased 30%.

Context and outcome, not pain and gain

• The standard "pain and gain" model is incomplete — context is what makes value legible
• A vector of progress = the context someone is in + the outcome they're trying to reach
• Value is relative to starting point: moving from far below the goal creates more perceived value than starting close to it
• QuickBooks has half the features and double the price of alternatives — meeting customers where they are matters more than feature superiority
• "Bitching ain't switching" — complaints don't predict action; only people who already tried to change reveal real demand signals

The six phases of buying

Most sales processes map to how the seller wants to sell, not how the buyer actually moves:

1. First thought — problem becomes conscious
2. Passive looking — problem aware, solution unaware; gathering background
3. Active looking — problem and solution aware; framing alternatives
4. Deciding — making trade-offs
5. First use — initial experience with the new product
6. Ongoing use — building the new habit

Auto Books restructured demos around where each prospect sat in this sequence. Three distinct demo types replaced one generic pitch. Sales cycle halved; conversion quadrupled.

How to interview customers

Who to interview: only people who have already tried to make the progress — not people who say they want to. For zero-to-one products, interview people who made a similar switch with a competing product.

How many: 10–12 interviews per round. Patterns repeat around 7–8. Two rounds of 12 beats one round of 24.

Three tips:

• Read Never Split the Difference (Chris Voss) for mirroring and questioning techniques
• Don't use a discussion guide — follow the most meaningful threads, not a fixed question list
• Use the pushes/pulls/anxieties/habits framework as a mental scaffold, not a script

Questioning approach:

• Extract the full story — not just the moment of purchase but the 18 months leading up to it
• Ask "tell me more about that" and "give me an example" rather than "why"
• Play back information incorrectly — people will correct you and elaborate
• Bracket when they run out of language: "was it more like X or more like Y?" forces further articulation
• Reach the edge of language — the point where they have no more words for it

Three layers of customer language:

1. Pablum layer — surface pleasantries ("it was great")
2. Fantasy/nightmare layer — exaggerated positive or negative ("it was the best/worst")
3. What actually happened — the investigative layer you need

A $137 coat rack took 18 months to decide on. The buyer reported buying it "in a week." The real story is always longer than they remember.

Clustering, not segmenting

• Don't look for themes across interviews — find causal pathways
• Pushes travel with specific pulls; sets of reasons work together
• Most products are hired to do 3–5 different jobs — often in conflict
• Intercom found four distinct jobs; instead of four products, they turned off irrelevant features per pathway
• Each pathway competed with a different product: one vs. HubSpot, another vs. Zendesk — so pricing was restructured to match each competitive frame

When JTBD doesn't work

• Where there's no real choice (e.g., employer-assigned health insurance) — people don't go through a deliberate decision process
• Deep habitual behaviour (e.g., chewing gum) — people can't recall purchase moments, only use moments
• When a company wants JTBD to validate a predetermined conclusion — the method reveals demand-side reality; it can't be reverse-engineered to justify a supply-side decision
• For habitual or no-choice contexts: ethnography and friction-mapping are more useful

Different flavours of the framework

• Bob Moesta's method: qualitative, organic, tactical — focused on how to get jobs data and what to do with it; suited to startups and zero-to-one work
• Clay Christensen's approach: theoretical and strategic — a thinking framework and philosophy; derived from the same underlying data set
• Tony Ulwick's ODI (Outcome-Driven Innovation): functional and supply-side — asks what jobs a product can do, not what jobs people have; more prescriptive, suited to regulated or complex systems

Sitting in a conference room and hypothesising jobs is "100% wrong" — the irrational details only emerge in conversation.

Zero to one applications

• Study what people will fire when they hire your new product — interview users of whatever they'd stop using
• Facebook Marketplace (unconfirmed) emerged from studying eBay and Craigslist sellers and buyers
• There are no new jobs — only better delivery on long-standing struggling moments
• Jobs existed 10 years ago and will exist 20 years from now; technology changes how they're fulfilled, not what they are

The dining room table insight

Bob built condos for downsizers who said they didn't want a dining room table. He built a space for one anyway. Sales rose 22%.

The table was the emotional bank account of their family life — if there was no place for it, they wouldn't move. What people say they don't want and what's blocking them from acting are different things.

Three things to take away

• Study struggling moments — they are everywhere; innovation happens when people change direction, not when they maintain momentum
• Understand the customer's standard, not your own — their context, their outcome
• Choose what to suck at — make deliberate trade-offs that match the trade-offs your customer is already making; most product failures come from misaligned trade-offs