Pilot2Service Customer Vision to JTBD
Purpose
Turn a free-form, even unstructured vision of a product and customer into structured customer understanding before writing the PRD: who the customer is (ICP), what they're really trying to accomplish (JTBD), what needs lie beneath that (Need Themes), and where AI brings the biggest competitive advantage. The principle is to start from the customer, not the technology — a spec never starts from a technical description.
Based on
- The owner's AI-native Business Design workshop (the owner's own workshop), run 1–2 June 2026, Day 1 — Session 2 "Planning in the AI Era", steps 1–3 (vision → design partner → ICP/JTBD/Need Themes/AI-advantage scoring).
- The methodology of an external "AI-first SaaS Product" workshop (applied
by the owner to one own case, see
../../references/ai-first-saas-workshop-source.mdand../../cases/ai-decision-coach-mvp-case.md) — deepens this skill's steps 5–7: verb-driven JTBD phrasing, a more detailed 5-criteria NMB+AI-advantage score (replacing the earlier single AI-advantage score), and an explicit AI wedge selection criteria set. Note: this deepened part has so far been applied only once (the owner's own case) — not as broadly validated as the skill's ICP/JTBD backbone, which is based on a multi-participant workshop. - Ideal Customer Profile (ICP) and Jobs-To-Be-Done (JTBD) product-strategy frameworks (generally known, not the owner's own — the workshop applies them within a prompt chain run with an AI thinking partner).
- See
../../references/workshop-source.mdand../../references/prompt-library.md(prompts 3–6).
Method
- Write a free-form vision. Open a blank document and write freely: what is the customer's need, how does the service work from the customer's perspective, how significant is the problem, how does the customer use the service, what are your own values/principles/ focus. It doesn't need to be logical or in order — disconnected thoughts and unfinished sentences are acceptable. The goal is to get your own thinking out into a form the AI can work with.
- Hand the vision to the AI thinking partner for sharpening
(
../../references/prompt-library.md, prompt 3): ask it to state what is still unclear and needs a decision, ask at most 7 clarifying questions in priority order, and surface assumptions worth checking. Don't ask for a plan yet — at this stage the goal is sharpening your thinking. - Ideal Customer Profile (ICP). Ask the AI to sketch out multiple customer profiles that have this problem most strongly, describe each, and prioritize who to serve first and why.
- Jobs To Be Done (JTBD) — verb-driven. Dig into the customer's real behavior and the situation in which the problem needs to be solved, and what they're really trying to achieve — look past the product to the customer's true goal. Phrase every JTBD starting with a verb, as a problem-/solution-independent progress ("figure out", "assess", "translate" — not "gets a dashboard" or "uses feature X"). Produce typically 5-8 JTBDs. Test: a JTBD should remain true even if the whole product changed into a completely different solution.
- Distill into 5+2 Need Themes. Convert the JTBD analysis into 5
functional and 2 psychological themes in table form. Each theme is a
one- or two-word noun phrase that captures the core of the need (e.g.
affordability, reliability, confidence). Use a four-column
table: Need Theme / Type (Functional: Understand, Diagnose/Evaluate,
Plan/Structure, Communicate, Decide/Act — or Psychological: Confidence,
Uncertainty Reduction) / Underlying "why" / Related JTBD(s). See the
worked example
../../cases/ai-decision-coach-mvp-case.md, section 2. - NMB + AI-advantage scoring — 5 criteria, not one. Score each need
theme on five criteria, each 1–5:
- Need Depth — how deep/acute the need is when it's triggered.
- Frequency — how often the customer encounters this need.
- Market Coverage — how broadly the need touches the target market.
- Business Strength — how strong YOUR OWN starting position is for serving this need (data, experience, existing tools).
- AI Advantage — how much competitive advantage AI brings to
serving this particular need (a genuine differentiator, not a
nice-to-have).
Calculate a total score (max 25) and classify each need theme as a
Differentiator (high total score, competitors don't cover it well)
vs. a Table Stake (necessary but generic competitors/other AI tools
are available). See the worked example
../../cases/ai-decision-coach-mvp-case.md, section 3.
- Select the AI differentiator need ("AI wedge"). The AI wedge is a
need theme that SIMULTANEOUSLY satisfies: high Need Depth, high
Frequency, weak Market Coverage among competitors, high own Business
Strength, AND high AI Advantage. This is the one (or two) need(s)
around which solution ideation (see
../ai-differentiator-solution-ideation/SKILL.md) is built — don't try to serve all Differentiator-level needs at once. - (Deepening, optional) Deep research. Use an AI tool's deep research mode to deepen your understanding of the customer/problem, competing solutions, and market/research data before solution ideation. The result is a research draft with reports and references — not a final truth.
- Carry the result (ICP + JTBD + Need Themes + NMB score + chosen AI
wedge) into
../ai-differentiator-solution-ideation/SKILL.mdas the basis for ideating solution directions — or directly into../ai-buildable-prd-writing/SKILL.mdif the solution direction is already clear and the ideation stage isn't needed.
Gotchas
- Phrasing a JTBD around the product instead of the customer's progress ("gets a dashboard", "uses feature X") fails the method's own test: a JTBD should stay true even if the whole product changed into a completely different solution — if it doesn't, it's not verb-driven and solution-independent yet.
- A need theme can score well overall and still not be the AI wedge — step 7 requires it to SIMULTANEOUSLY hit high Need Depth, high Frequency, high own Business Strength, high AI Advantage, AND weak Market Coverage among competitors. A need with four strong criteria but a market competitors already cover well is a Table Stake, not a wedge.
- Deep research output (step 8) is a research draft, not verified fact — carrying its claims straight into the PRD without checking primary sources contradicts the skill's own "does NOT do" caveat.
- The 5-criteria NMB+AI-advantage scoring and AI wedge selection logic (steps 5-7) come from an external workshop the owner has so far applied to only one case — treat that part as less battle-tested than the ICP/JTBD backbone, which rests on a multi-participant workshop.
- Skipping the free-form vision step (step 1) and starting straight from a structured ICP/JTBD template reintroduces a technology-first framing — the whole point of step 1 is to get unstructured customer thinking down before any structure is imposed.
What this skill does NOT do
- Does not replace real customer research or interviews — the AI structures your thinking and existing knowledge, it doesn't produce new empirical customer data.
- Does not make the final ICP or priority choice for you.
- Deep research results are a research draft — verify primary sources before carrying them into the PRD or decision-making.
Continue from here
- Preceding skill in this pack:
../ai-native-opportunity-scan/SKILL.md - Next skill in this pack:
../ai-differentiator-solution-ideation/SKILL.md— ideates 3 solution directions for the chosen AI wedge. (If the solution direction is already clear, you can go directly to../ai-buildable-prd-writing/SKILL.md.) - Related skill in another pack:
../../../../opportunity-recognition/skills/opportunity-value-assessment/SKILL.md - Worked example:
../../cases/ai-decision-coach-mvp-case.md— the full own case for steps 4–7. - The pack's shared guardrails:
../../CLAUDE.md
References
../../references/prompt-library.md— prompts 3–6../../references/workshop-source.md— source information (AI-training-P6)../../references/ai-first-saas-workshop-source.md— source information (external workshop, NMB scoring and AI wedge deepening)../../cases/ai-decision-coach-mvp-case.md— worked example../../CLAUDE.md— the pack's shared guardrails