End-to-End Example: From Product Input to Decision Report
End-to-End Example: From Product Input to Decision Report
Illustrative example. “NotesAI” is a fictional product invented to explain the UseCaseify workflow. Every quote, number, and response below is fictional. This is not a customer case study, and the fictional prospect answers are not real market data.
1. Product input
The team behind NotesAI — a fictional AI meeting-minutes tool for Japanese businesses (transcription, structured minutes, action items; around ¥1,500/user/month) — enters the product URL and a short description. They have a working product and zero customer case studies.
2. Product profile
UseCaseify drafts a versioned product profile: category, core problem, capabilities, target hints, and open questions. The team corrects one field — the draft assumed on-premise deployment was available; it is not — and confirms the profile. Nothing downstream runs against an unconfirmed profile.
3. Research questions
The research plan asks, among others: Who complains publicly about writing minutes? Do teams already pay for alternatives? Where does resistance to recording meetings appear?
4. Evidence collected
Supporting evidence
- Posts by sales managers about spending evenings writing call summaries and follow-ups (repeated signal across several forums).
- Job postings assigning minutes duty to junior staff — evidence that companies already pay for this work with salaried time.
- Comparison articles about paid transcription tools, showing an existing paid category.
Contradicting evidence
- Multiple threads arguing that bundled AI features in existing office suites are “good enough” for internal meetings.
- IT administrators expressing concern about recording customer-facing calls at all.
Unknowns
- Willingness to pay among licensed professionals (tax accountants, labor consultants) — no usable public signal found.
5. Opportunity candidates
The system generates ten candidates, merges duplicates, and red-team review removes weak ones. Four opportunity cards reach human review:
- A. Sales-call minutes with same-day follow-up for B2B sales teams
- B. Client-meeting records for licensed professionals
- C. Council and committee minutes for local government
- D. Internal recurring-meeting minutes rolled out by IT
6. Opportunity scoring
Each card is scored on ten dimensions and summed deterministically (fictional values):
| Card | Total | Confidence | Evidence level |
|---|---|---|---|
| A. Sales follow-up | 72 | medium | repeated signal |
| B. Professionals | 68 | low | single signal |
| C. Government | 55 | low | insufficient |
| D. Internal meetings | 61 | medium | repeated signal |
The team rejects C outright (no channel into government sales). B scores well but its confidence is low — the “unknown willingness to pay” from step 4 is visible instead of hidden inside the score.
7. Selected opportunity
The system names A the primary recommendation (highest score at medium confidence, no unresolved contradiction) and states the reasons and the main counter-argument: recording resistance on customer calls.
8. GTM test assets
For A, nine asset types are generated with unsupported-claim checks. Three examples (fictional):
- Value proposition: “Send the follow-up while the deal is still warm — minutes and action items ready before you’re back at your desk.”
- Headline: 「商談が終わった瞬間、議事録とフォローメールの下書きまで。」
- Cold outreach opening: a two-sentence email referencing the evening hours sales managers spend on call summaries.
The claim check flags one draft sentence — “trusted by sales teams across Japan” — as unsupported (no customers yet) and it is removed.
9. Validation questions
The team publishes a five-question validation page (password-protected) and shares it with sales managers in their network: relevance of the problem, current handling, most valuable outcome, biggest concern, willingness to see a demo.
10. Prospect feedback
Eleven fictional responses come back. The pain is confirmed, but the resonant outcome is not “minutes written faster” — it is “the follow-up email goes out the same day.” Two respondents name recording resistance as a blocker; several report the current alternative is “a junior colleague does it.”
11. Learning update
The analysis clusters the responses and proposes score changes: urgency up, evidence quality up (evidence level rises to prospect feedback), and the recording concern is filed as contradicting evidence. The suggestions apply only when the team accepts them — they accept all three.
12. Recommendation change
The primary opportunity remains A, but the recommended value proposition pivots from “write minutes faster” to “same-day follow-up,” and the next test is cold outreach using the new hook. B stays queued for a later cycle. This is the loop working as intended: real feedback changed the message before any ad budget was spent.
13. Decision report
The team locks the decision into an immutable report: the chosen opportunity, the pivoted value proposition, the evidence for and against, the accepted score changes with provenance, and the next tests — shareable with advisors through a revocable link.
14. What remains uncertain
Recording resistance in regulated industries, actual willingness to pay (feedback is interest, not purchase), and channel economics. The report says so explicitly. A decision made on eleven responses is directional — the point is that it is traceable, and the next test is already defined.
Try UseCaseify at usecaseify.com — 2 credits on sign-up, no credit card required.