by Braden Kelley and Art Inteligencia
What Jobs-to-Be-Done Prompts Should Product Teams Run? (Short Answer)
Nine jobs-to-be-done prompts every product team should run: (1) the struggling moment, (2) the progress definition, (3) the trigger, (4) the competing workaround, (5) the hiring and firing criteria, (6) the anxiety and trust barrier, (7) the dignity and emotional cost, (8) the constraints that govern, and (9) the behavior you will falsify. Run them before roadmaps freeze — in customer contact, synthesis, and review — or you will optimize features for a job nobody is actually trying to get done.
A persona describes who marketing hopes exists. A job prompt reveals who showed up trying to make progress — and what they hired instead of you.
Why Isn’t the Backlog the Job?
I have watched product reviews that felt like feature auctions. Requests on the left. Capacity on the right. Someone asked what progress a human was trying to make. The room reached for a persona slide the way a drowning person reaches for a brochure.
Jobs-to-be-done is not a workshop poster. It is a set of prompts you run — in interviews, synthesis, roadmap reviews, and falsification — before the backlog freezes. AI can summarize after contact. It cannot replace the prompts that force contact with reality. Specs and roadmaps can follow. They should not lead.
| Prompt | Surfaces | Costume version |
|---|---|---|
| 1. Struggling moment | Situation, stakes, real language | Generic pain points; scraped reviews as “research” |
| 2. Progress definition | The job in their words | “Users want faster reporting” |
| 3. Trigger | Why act now; switching energy | Evergreen needs with no trigger |
| 4. Competing workaround | Shadow tools, heroics, effort | Competitor cards without workaround archaeology |
| 5. Hiring and firing criteria | Trust, good enough, non-negotiables | Feature parity with no fire criteria |
| 6. Anxiety and trust | Fear, undo, recoverability | “Frictionless UX” that ignores stakes |
| 7. Dignity and emotional cost | Shame, trapped, alone | Sentiment tags without a story |
| 8. Constraints that govern | Policy, politics, system of record | Discovery that ends at the happy path |
| 9. Behavior to falsify | Observable success hypothesis | Roadmap items with no testable behavior |
1. What Is the Struggling Moment Prompt?
The prompt: “Walk me through the last time this was hard.” Ask for a specific recent episode — not opinions about the category.
Surfaces: Situation, stakes, context, and language that could not have been invented in the building.
Costume: Generic pain points; scraped reviews mistaken for contact; synthetic personas with fluent prose and no episode.
Run it: One interview rule — no hypotheticals until the story is concrete. If you cannot point to a moment, you do not have a job yet. You have a market segment.
2. How Do You Define Progress in the Customer’s Words?
The prompt: “What were you trying to get done?” Separate the job from the feature request; ask in their words, not your taxonomy.
Surfaces: Functional and emotional progress — not “use our dashboard.”
Costume: Requirements that name features while the decision or relief the human sought stays invisible.
Run it: Write the job as a verb phrase a human would say out loud. If it sounds like a roadmap item, rewrite it. For sharper framing before specs freeze, see 10 Design Questions That Beat a 40-Page Requirements Document.
3. What Is the Trigger Prompt in JTBD?
The prompt: “What changed that made you act now?” Why this moment, not last quarter.
Surfaces: Urgency, switching energy, events that open a hiring window.
Costume: Evergreen “needs” with no trigger — roadmap fiction that could ship any quarter.
Run it: Map triggers to release timing and onboarding moments. Timing is part of the job.
4. How Do You Surface Competing Workarounds?
The prompt: “What did you use instead — including the ugly fix?” Ask for spreadsheets, email chains, heroics, shadow tools — not only named competitors.
Surfaces: Real competition (often internal); effort, failure demand, and dignity cost of the workaround.
Costume: Competitor battle cards without workaround archaeology.
Run it: List every workaround; rank by frequency and dignity cost. For friction customers feel before any map names it, read 12 Friction Points Customers Feel Before Your Journey Map Does.
5. What Are Hiring and Firing Criteria in Jobs-to-Be-Done?
The prompt: “What would make you choose this — and what would make you leave?” Force tradeoffs; ask what “good enough” means and what breaks trust.
Surfaces: Minimum viable trust; switching costs; non-negotiables.
Costume: Feature parity matrices with no firing criteria.
Run it: Pair hire/fire lists before prioritization. If you cannot name fire criteria, you do not know the job — you only know your backlog.
6. How Do You Design for Anxiety and Trust?
The prompt: “What were you afraid would go wrong?” Ask what almost stopped them — embarrassment, rework, blame, compliance fear.
Surfaces: Anxiety as design input; recoverability, undo, and escalation at the moment of truth.
Costume: “Frictionless UX” that ignores stakes; demos that assume infinite confidence.
Run it: Design recovery paths from named anxieties, not generic “error states.” Trust is a job outcome.
7. Why Ask About Dignity and Emotional Cost?
The prompt: “Where did this make you feel stupid, trapped, or alone?” Ask where the experience taxed dignity — repetition, visibility, powerlessness.
Surfaces: Emotional job; shame and confidence; EX/CX overlap when employees deliver the experience.
Costume: Sentiment tags without a human story; NPS without a “why.”
Run it: One dignity cost per journey; owner and measure if you claim to be human-centered. Dignity is not soft. It is design surface.
8. What Constraints Should Govern the Job Map?
The prompt: “What rules, tools, or politics could you not ignore?” Ask what policy, staffing, time, or system-of-record reality shaped the attempt.
Surfaces: The real design surface — not infinite “yes” in the prototype.
Costume: Discovery that ends at the happy path; surprise policy after “done.”
Run it: Constraint map beside job map before solutioning. For what AI can assist versus what humans must own in the craft, see 5 Elements of Human-Centered Design That AI Cannot Own.
9. What Behavior Will You Falsify?
The prompt: “What will people do differently if this job gets easier?” Name one observable behavior — complete in one try, abandon workaround, time-to-confidence.
Surfaces: Testable hypothesis; adoption signal; anti-demo theater.
Costume: Roadmap items with no falsifiable behavior; applause demos that teach nothing.
Run it: One behavior per bet; measure what people do, not what they clap for. For method costume that skips falsification, see 7 Ways Design Thinking Gets Misused.
How Do Product Teams Check JTBD Before the Roadmap Commits?
Before the next roadmap commit, run five go/no-go questions. If you cannot answer them with contact-backed evidence, you are prioritizing features, not jobs:
- Did we run contact-backed prompts — not only desk research and generated personas?
- Can we state the job in their words — as a verb phrase, not a feature cluster?
- What workarounds are we competing with — including the ugly internal ones?
- What anxieties and dignity costs are in scope — with recovery designed in?
- What behavior falsifies success — not what demo are we showing?
Optional 90-minute JTBD sprint: One journey, one room, nine prompts as a wall — answers as evidence, not slogans. Then a thin backlog that traces back to the prompts. Requirements should be the receipt of understanding — not the substitute for it.
Run the prompts before the backlog owns you.
Frequently Asked Questions
What are jobs-to-be-done prompts?
Jobs-to-be-done prompts are structured questions product teams run in discovery — about struggling moments, progress sought, triggers, workarounds, hire/fire criteria, anxiety, dignity costs, constraints, and falsifiable behaviors — to understand what job a human is trying to get done before freezing features or roadmaps.
How do product teams use JTBD?
Product teams use JTBD in customer interviews, synthesis, roadmap reviews, and testing — running prompts that surface real episodes, language, workarounds, and trust barriers, then tying backlog items to observable behavior change rather than feature requests alone.
What is the difference between JTBD and personas?
Personas often describe demographic or attitudinal segments. JTBD focuses on the progress a person is trying to make in a specific situation — including competing workarounds, triggers, anxieties, and firing criteria. Personas can inform marketing; job prompts should inform product bets.
How do you run JTBD interviews?
Start with a concrete struggling moment — no hypotheticals. Walk through progress sought, what triggered action, what they used instead, hire/fire criteria, fears, dignity costs, and governing constraints. Close by naming one behavior that would change if the job got easier, then test for it.
Can AI replace jobs-to-be-done research?
AI can assist after contact — summarizing interviews, clustering themes, drafting job maps — but it cannot replace lived episodes, workaround archaeology, or judgment about trust and dignity. Synthetic insights without contact produce fluent decoration, not product discovery.
Image credits: Google Gemini
Content Authenticity Statement: The topic area, key elements to focus on, etc. were decisions made by Braden Kelley, with a little help from Google Gemini and Cursor to clean up the article, add images and create infographics.
Sign up here to get Human-Centered Change & Innovation Weekly delivered to your inbox every week.