Proposal

Proposal

> Part of the discovery-phase skill pack · deliverables group · reads discovery-context.md (run profile-builder first if missing).

Agency-to-client proposal document. The output the client reads, signs (or doesn't), and budgets against.

Step 1 — Read context + prerequisites

Read discovery-context.md (sections 3. Engagement → Mode — if mode is staff aug or internal, suggest discovery-handoff instead, but allow override; 5. Expected deliverable).

Prerequisite artifacts — the proposal's substance:

• scope-doc.md — scope language
• estimation.md — effort + assumption list
• risk-assumption-map.md — risk register
• Recommended: problem-canvas.md, opportunity-tree.md

If discovery-context.md is missing, ask the BA inline: "(a) engagement mode (pre-sale / paid discovery / other); (b) client name + sector?" — tag [ASSUMED]. If 1–2 prerequisite artifacts are missing, ask for 3–5 stand-in bullets per artifact. If 3+ are missing, strongly recommend running the prerequisite skills first; otherwise proceed with [ASSUMED ] placeholders and a non-removable DRAFT-ONLY banner at the top of proposal.md (NOT for client send). Never block.

Step 2 — Pick proposal style

| Style | When | Length | |---|---|---| | Executive brief | Senior client decision-maker, pre-sale, time-pressured | 2-3 pages | | Detailed proposal | Procurement / RFP response, larger engagement | 8-12 pages | | Pitch deck | Initial conversation; supplements verbal pitch | Doc + companion deck |

Default to executive brief unless context section 3 (Mode) implies RFP.

Step 3 — Required sections (in this order)

1. Cover — engagement title, client name, agency name, date, version, primary contact
2. TL;DR — 3-4 bullets max. The client should be able to forward just this.
3. Problem we'd solve — one paragraph from problem-canvas.md, in client-language. NOT in agency-language.
4. What we learned (if discovery happened already) — 3-5 bullets from insight-matrix.md top-3 with one quote each. Supports the recommendation.
5. Recommended approach — what we'd build, why this and not alternatives. Pull from opportunity-tree.md chosen-direction. If client pre-proposed a different solution and our recommendation differs, handle this section carefully — see step 4.
6. Scope — pull "in" column from scope-doc.md. Include the success metrics.
7. Out of scope — pull "out" column + "explicitly deferred". Crucial — prevents post-contract scope wars.
8. Approach / phases — high-level timeline aligned to discipline matrix from estimation.md. Show milestones, not all tasks.
9. Team — who from agency works on what. Pull discipline columns from estimation.
10. Timeline — calendar weeks, key milestones, decision points
11. Effort and pricing guidance — from estimation.md adjusted total. Range, not a number. Specify whether fixed-fee, T&M, or phased.
12. What we need from you — the client-side assumptions and dependencies (from scope-doc.md outsourcing extras + risk-assumption-map.md [client]/[shared] rows)
13. Risks and how we'll handle them — from risk-assumption-map.md register, prioritized
14. Assumptions — verbatim from estimation.md "Assumptions that invalidate the estimate"
15. Next steps — concrete, signable: "we'll send a SoW within 5 business days of your verbal yes" or similar

Step 4 — Handle "we recommend something different than client asked"

If opportunity-tree.md was Validate-mode and chose against client's pre-proposed solution:

• Section 5 leads with empathy: "You came to us with ``. Here's what we learned and where it took us."
• Section 5 shows the alternatives — including client's idea — with honest scoring.
• Don't pretend the client's idea was never on the table; that breaks trust.
• Pre-draft the BA-to-client conversation script as an appendix the BA can rehearse.

Step 5 — Voice and tone

• Client-language, not agency-language. "We'll do X" → "We'll help you do Y for users".
• No internal jargon ("MVP", "v1", "story points"). Translate to client terms ("first release", "subsequent release", "effort estimate").
• Active voice. Short sentences. The client is busy.
• No buzzwords ("synergy", "leverage", "best-in-class"). They lower trust.

Step 6 — Output

./discovery/proposal.md per ./template.md.

Append to _log.md: [proposal | YYYY-MM-DD] style: ; total: ; sent_to: .

Anti-patterns

• Buried scope. Scope (and out-of-scope) are the two most-read sections. Don't hide them in section 11.
• Single-number price. Always range. Single number = commitment.
• Hiding "what we need from you". This is where the engagement actually lives or dies. Be explicit.
• Including discovery's open questions as "we'll figure this out". Either resolve them before proposal or surface them as risks.
• Agency-internal language. "Sprint 1, 2, 3" → "First 3 weeks". "Deliverable D1" → describe it in plain terms.
• No alternatives in Validate-mode handling. If you recommend differently than client asked, show your work. Hiding it = lost trust.

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: stanislavnianko
• Source: stanislavnianko/product-discovery-claude-skills
• License: MIT

Install and usage instructions live in the source repository linked above.