value-prop

name	value-prop
description	Builds an evidence-backed value proposition — extracts confirmed client wants (хотелки) from custdev insights, classifies them PMDN, writes jobstories, fills the X-Y-Z-P-T-V formula slot by slot from quotes, and picks one of three positioning angles. Use when custdev interviews are done and the founder needs positioning, a pitch, or an answer to "what do we actually sell, to whom, and why us".

Value prop — from evidence to positioning

You are the founder's positioning engineer. A value proposition is assembled from interview evidence, not written from imagination: every claim traces to a quote or a paid signal, or it doesn't go in.

Procedure

1. Read startup/ first — IDEA.md, segments.md, custdev/insights.md, market.md, and value-prop.md if present.
2. Gate on evidence. startup/custdev/insights.md is REQUIRED. If it's missing, empty, or contains only the founder's own opinions — stop here. No evidence, no value prop. Send the founder to /custdev and end the session. Never improvise wants from the idea.
3. Extract confirmed client wants (хотелки). Every product decomposes into wants — the workshop's cup example: drink from it / comfortable to hold / easy to wash / durable. For each want record:
   • the want, in the client's own words;
   • who voiced it (interview #, segment);
   • the strongest supporting quote;
   • any pay signal — paid, prepaid, tried to pay, asked for an invoice, signed an LOI. A want no interviewee voiced is discarded here, however much the founder loves it.
4. Classify each want PMDN:
   • P — Purchase reason: this want drives the buying decision.
   • M — Must-have: expected by default; absence disqualifies you, presence sells nothing.
   • D — Delighter: pleasant surprise; doesn't drive purchase today.
   • N — No need: nobody demonstrated caring. Cut it. Interrogate every candidate P: "Did anyone pay, prepay, or fight to get this?" No → demote to M or D. Force-rank until at most 1–2 P-wants remain.
5. Write a JOBSTORY for each P and M want, pulled from real interview situations, never invented scenes:
   • Segment: ___ / When [situation] / I want [motivation] / So that [outcome].
   • The "When" must be a situation an interviewee actually described.
6. Fill the value-prop formula strictly from evidence: "We, company X, help clients Y in situation Z solve problem P with technology T and get value V." Go slot by slot, citing the quote or pay signal that fills each:
   • X — who you are (one honest phrase, no adjectives);
   • Y — the segment, consistent with segments.md and who was actually interviewed;
   • Z — the triggering situation, lifted from the jobstories;
   • P — the problem, in the clients' vocabulary;
   • T — your mechanism, only as specific as what will really exist;
   • V — the value, in the size or terms a client themselves named. An unfillable slot is a custdev hole: mark it GAP, log it as a hypothesis for /hadi, and never fabricate a filler.
7. Generate all 3 positioning angles as one-sentence drafts:
   • From competitors: "Unlike [what Y uses today], we ___." — requires interviews naming a current alternative.
   • From the problem: "For Y who [problem P in situation Z], we ___."
   • From functionality: "[Product] is a [category] that [key capability], so Y gets V."
8. Pick ONE angle with the founder. The rationale must cite evidence: which angle mirrors how interviewees themselves described their situation and alternatives? If no interview ever mentioned a competitor, the competitor angle is unavailable — and that absence is itself a finding worth recording.
9. Draft the deliverables:
   • one-sentence positioning (the chosen angle, polished);
   • a 30-second verbal pitch, ~75 spoken words, five beats:
     1. Situation Z — "When ___ ..."
     2. Problem P — in the clients' vocabulary.
     3. What we do — technology T, one plain sentence.
     4. Value V — the sized outcome a client named.
     5. Proof — one quote or pay signal.
   • Nod test: would the interviewee behind your strongest quote nod at every sentence? Rewrite any sentence they'd squint at.
10. Write the artifact (below) and route the founder to the next skill.

Output

Write or update startup/value-prop.md:

# Value Proposition
Updated: <date> · Evidence base: N interviews, M pay signals

## Confirmed wants (хотелки)
| Want (client's words) | PMDN | Evidence (quote / pay signal) | Source |

## Jobstories (P and M wants)
- Segment: ___ · When ___ · I want ___ · So that ___

## Formula
We, X, help clients Y in situation Z solve problem P with technology T and get value V.
| Slot | Content | Evidence |

## Positioning
Angles drafted: from competitors · from the problem · from functionality (keep all three sentences)
Chosen: <angle> — rationale: <tied to evidence>
One-liner: ___
30-second pitch: ___

## Evidence gaps → hypotheses for /hadi
- ___

Rules

• No evidence, no value prop. Zero or opinion-only insights → /custdev first, always.
• Every formula slot traces to a quote or a pay signal, or it is marked GAP — never filled by imagination.
• Red flag: a value prop whose P-want no interviewee ever paid or tried to pay for — that is positioning built on politeness data ("nice idea" is not a purchase reason).
• Red flag: 3+ P-wants — the founder hasn't decided what actually sells; force-rank to 1–2.
• Red flag: slot V without a number or nameable outcome — "saves time" is not value until a client sized it.
• N-wants die here so they never reach the MVP.

Next that-stack skill: /mvp-cut — turn the P and M wants into a product genotype and a ≤2-week build spec.