Materials Generation Subskill

Purpose

Generate ready-to-use sales materials from strategic inputs with automatic impact scoring for autonomous deployment vs human review.

Operates on: ONE product per invocation Enforces: Problem-first ordering, feature-to-benefit translation, stage-appropriate messaging Autonomous deployment: Impact score < 0.8

Input Parameters

Required:

• product: Product name (glamyouup, detekta)
• segment: Segment name (dtc-fashion-smb, marketplace-reseller, department-store-early-adopter)
• canvas_path: strategy/canvas/
• icp_path: research/customer/icp/{segment}-icp.md

Optional:

• narrative_path: threads/sales/narrative/{segment}-*.md (recommended)
• stage: "mvp" | "growth" | "enterprise" (defaults to Canvas stage if present)
• force_regenerate: boolean (skip change detection, always regenerate)

Workflow

1. Load Context

Read Canvas files from strategy/canvas/:

• customer-segments.md - Who you're selling to
• problem.md - Pain points to emphasize
• solution.md - What you're selling
• unique-value-proposition.md - Why you're different
• key-metrics.md - Quantified outcomes

Read ICP from research/customer/icp/{segment}-icp.md:

• Observable firmographics/technographics
• Buyer personas (economic, technical, end-user)
• Qualification questions
• Problem signals

Read Sales Narrative from artifacts/sales/{segment}/narratives/ (optional):

• {segment}-economic-buyer.md - Economic buyer messaging
• {segment}-technical-buyer.md - Technical buyer messaging
• {segment}-objection-lib.md - Objection handling library

Read Optional Data:

• research/customer/icp/prospect-list.csv - Real prospect examples
• research/customers/proof-points.md - Customer success stories
• research/market/industry-stats.md - Industry data

Handle missing files:

• Canvas missing: ERROR (required)
• ICP missing: Use generic B2B SaaS patterns + flag for customization
• Narrative missing: Extract from Canvas + flag "TODO: Refine with sales-narrative skill"
• Optional data missing: Generate with placeholders + TODO markers

---

2. Detect Changes & Score Impact

Compare to previous generation (if exists):

Check artifacts/sales/{segment}/ for existing files:

• If files exist: Compare Canvas/ICP/Narrative last-modified timestamps
• If no files exist: First-time generation (impact score = 0.0, auto-deploy)

Calculate impact score (0.0 - 1.0):

High Impact (>= 0.8) - Requires human review:

• ICP target segment change (different customer profile)
• Pricing structure change (>20% difference in tiers/pricing)
• Value proposition reframing (quantitative to qualitative shift)
• Problem angle shift (focus on different pain point)
• Stage change (MVP → Growth or Growth → Enterprise)

Medium Impact (0.5 - 0.79):

• Added significant proof points (new case studies, major metrics)
• Buyer persona refinement (new decision-makers)
• Messaging tone shift (casual to formal or vice versa)

Low Impact (< 0.5) - Auto-deploy safe:

• Statistics updated (industry data refreshed)
• Wording improvements (grammar, clarity)
• Email template variants added
• Call script timing adjustments
• Formatting changes

If force_regenerate=true: Skip change detection, set impact=0.0 (always auto-deploy)

Example scoring:

Changes:
- ICP customer_size changed from "50-200" to "200-500" → +0.4 (major)
- Added industry stat to problem narrative → +0.1 (minor)
- Fixed typo in one-pager → +0.0 (trivial)
Total: 0.5 (Medium impact - requires review)

---

3. Generate Artifacts

Create four sales materials following frameworks from references:

3.1 Pitch Deck (pitch-deck.md)

Structure (10-15 slides, one idea per slide):

# Slide 1: Title + Problem Hook
[Product Name]
[One-sentence problem hook from Canvas]

# Slide 2: Problem Statement
[3-5 quantified pain points from Canvas problem.md]

# Slide 3: Current Alternatives
[How prospects solve today, why they fail - from Canvas]

# Slide 4: Your Solution
[High-level approach from Canvas solution.md]

# Slide 5: How It Works
[3-step process with feature → benefit mapping]

# Slide 6-7: Demo/Screenshots
[Visual proof - TODO if not available]

# Slide 8: Value Proposition
[Quantified benefits from Canvas key-metrics.md]

# Slide 9: Why You/Team
[Credibility signals - from Canvas or stage-appropriate]

# Slide 10: Traction
[Stage-appropriate metrics: MVP=pilots, Growth=customers, Enterprise=scale]

# Slide 11: Pricing
[From Canvas or ICP-based pricing guidance]

# Slide 12: Next Steps
[Clear CTA: trial, pilot, demo]

Quality checks:

• Problem (slides 2-3) comes BEFORE solution (slides 4-5)
• Each feature has corresponding benefit
• No vague claims ("best-in-class" without proof)
• Stage-appropriate language (MVP="pilot results", Enterprise="200+ customers")
• One idea per slide (not walls of text)

See references/pitch-deck-structure.md for detailed guidance.

---

3.2 One-Pager (one-pager.md)

Format constraints:

• Single page when printed (8.5" x 11")
• Readable in B&W (no color dependency)
• 12pt+ font throughout
• Skimmable in 30 seconds

Structure:

# [HEADLINE: Problem in 10 words or less]

## The Problem
- [Pain point 1 from Canvas]
- [Pain point 2 from Canvas]
- [Pain point 3 from Canvas]

## The Solution
- [How you solve it - approach not features]
- [Key differentiator from Canvas UVP]
- [For: ICP summary]

## Key Benefits
- [Quantified benefit 1 from Canvas metrics]
- [Quantified benefit 2]
- [Quantified benefit 3]

## How It Works
1. [Step 1 - customer perspective]
2. [Step 2]
3. [Step 3]

## Pricing
[Starting point or "Contact us"]

## Contact
[Company] | [Website] | [Email] | [Phone]
[CTA: Book a 15-minute demo]

Quality checks:

• Fits on one page
• Readable in B&W (no colored text)
• Problem before solution
• Benefits quantified (not vague)
• No jargon requiring explanation

See references/one-pager-rules.md for layout examples.

---

3.3 Call Scripts (call-scripts.md)

Three scripts with strict timing:

Discovery Call (30 minutes):

# Discovery Call Script (30 minutes)

## Opening (1 minute)
[Permission-based opening]

## Qualification (5 minutes)
[ICP qualification questions - yes/no or numeric]
[If disqualified → polite exit]

## Problem Exploration (15 minutes) ← MOST IMPORTANT
[Open-ended discovery questions]
[Listen for pain, quantify impact]

## Light Pitch (5 minutes)
[Map your solution to THEIR problems mentioned]

## Next Steps (4 minutes)
[Schedule demo with specific times]

Demo Call (30 minutes):

# Demo Call Script (30 minutes)

## Recap (2 minutes)
[Confirm problems from discovery]

## Customized Demo (15 minutes) ← MOST IMPORTANT
[Problem 1 → Feature → Benefit → Their impact (5 min)]
[Problem 2 → Feature → Benefit → Their impact (5 min)]
[Problem 3 → Feature → Benefit → Their impact (5 min)]

## Objection Handling (5 minutes)
[Common objections + responses from references]

## Pricing (3 minutes)
[Their tier + ROI calculation]

## Close (5 minutes)
[Direct ask: trial, pilot, or contract]

Follow-up Call (15 minutes):

# Follow-up Call Script (15 minutes)

## Opening (1 minute)
[Reference previous conversation specifics]

## Address Questions (8 minutes)
[Answer open questions with proof]

## Provide Proof (2 minutes)
[Case study, data, ROI calculation]

## Clear Ask (4 minutes)
[Specific proposal: trial, pilot, contract]
[Timeline and next steps]

Quality checks:

• Time limits specified per segment
• Questions map to ICP qualification criteria
• Problem explored before solution pitched
• Objection responses prepared
• Clear CTA in each script

See references/call-script-timing.md for detailed timing management.

---

3.4 Email Templates (email-templates.md)

Max 5 sentences per email (hard constraint):

Cold Outreach:

Subject: [Problem-focused, not company name]

Line 1: [Problem hook with quantification]
Line 2: [Brief solution + credibility signal]
Line 3: [Specific ask for 15-min call]
Line 4: [Easy out]
Signature

Follow-up Sequence:

# Follow-up #1 (3 days later)
Subject: Re: [Original subject]
[Value-add: stat, insight, article]
[How it relates to them]
[Soft ask]

# Follow-up #2 (1 week later)
Subject: [New angle on problem]
[Different problem perspective]
[How you solve it]
[Direct ask + calendar link]

# Follow-up #3 (2 weeks later - breakup)
Subject: Should I close your file?
[Acknowledge no response]
[Easy out or timeline question]
[Leave door open]

Meeting Request (after initial contact):

Subject: 15 min to discuss [specific problem]?

[Thank them + reference their response]
[What you'll cover]
[Specific time options or calendar link]

Post-Demo Follow-up:

Subject: Great chatting - [Their Company] + [Your Company]

[Thank + specific reference from demo]
[Attach resources promised]
[Quantified next step proposal]
[Make it easy]

Quality checks:

• Max 5 sentences (count them)
• Subject line is problem-focused
• Specific not generic (uses ICP characteristics)
• One CTA per email
• No vague claims without quantification

See references/email-formulas.md for detailed formulas and A/B testing.

---

4. Quality Validation

Run all quality checks before finalizing:

Problem-First Ordering:

• Pitch deck: Problem (slides 2-3) before solution (slides 4-5)
• One-pager: Problem section before solution section
• Call scripts: Problem exploration before pitch
• Emails: Problem hook in line 1

Feature-to-Benefit Translation:

• Every feature has corresponding benefit
• Benefits explain "why it matters"
• Impact quantified where possible
• No standalone feature lists

Vague Claim Detection:

• No "best", "leading", "innovative" without proof
• All claims either quantified or flagged as TODO
• Comparisons are specific
• Credibility signals are concrete

Format Compliance:

• Pitch deck: One idea per slide
• One-pager: Fits one page, readable B&W
• Call scripts: Time limits realistic
• Emails: Max 5 sentences each

Stage-Appropriate Messaging:

• MVP: "5 pilot customers" not "thousands of users"
• Growth: "35 customers" not "scrappy startup"
• Enterprise: "200+ customers" not "founder-led"

See references for detailed quality criteria:

• references/problem-first-ordering.md
• references/feature-to-benefit-translation.md
• references/stage-appropriate-messaging.md

Flag quality issues as TODOs in output:

TODO: Quantify this claim - "significant improvement" → "35% reduction"
TODO: Translate feature to benefit - "AI-powered" → "Reduces returns by 35%"
TODO: Add proof point - "industry-leading" needs supporting data

---

5. Version & Deploy

Based on impact score, deploy or flag for review:

Impact Score < 0.8 (Auto-Deploy)

1. Archive old materials (if they exist):

   • Move existing files from artifacts/sales/{segment}/materials/ to artifacts/sales/{segment}/materials/archive/{date}/
   • Only if files already exist (first-time generation skips this)

2. Write files to artifacts/sales/{segment}/materials/:

   • pitch-deck.md
   • one-pager.md
   • call-scripts.md
   • email-templates.md

3. Return success:

   Status: Deployed
   Location: artifacts/sales/{segment}/materials/
   Segment: {segment}
   Impact: {score} (Low - auto-deployed)
   Files: [pitch-deck.md, one-pager.md, call-scripts.md, email-templates.md]
   

---

Impact Score >= 0.8 (Requires Review)

1. Write files to artifacts/sales/{segment}/drafts/{date}/:

   • pitch-deck.md
   • one-pager.md
   • call-scripts.md
   • email-templates.md
   • REVIEW-NEEDED.md explaining high-impact changes

2. Flag in ops dashboard:

   # ops/today.md
   
   ## 🚨 Review Required: Sales Materials
   
   **Segment**: {segment}
   **Impact Score**: {score} (High)
   **Reason**: {changes that triggered high score}
   **Location**: artifacts/sales/{segment}/drafts/{date}/
   **Action**: Review materials, approve deployment, or request changes
   

3. Return pending status:

   Status: Pending Review
   Location: artifacts/sales/{segment}/drafts/{date}/
   Segment: {segment}
   Impact: {score} (High - requires review)
   Reason: {specific high-impact changes}
   Action: Review at artifacts/sales/{segment}/drafts/{date}/REVIEW-NEEDED.md
   

---

Example Impact Scoring

Example 1: Low Impact (0.2)

Changes:
- Updated industry statistic in problem slide (2023 data → 2024 data)
- Fixed typo in one-pager ("recieve" → "receive")
- Adjusted call script demo timing (16 min → 15 min)

Impact Calculation:
- Stat update: +0.1
- Typo fix: +0.0
- Timing tweak: +0.1
Total: 0.2

Action: Auto-deploy

Example 2: Medium Impact (0.6)

Changes:
- Added major case study (ChicThreads 35% return reduction)
- Buyer persona refinement (added "E-commerce Manager" to ICP)
- New email template variant for enterprise prospects

Impact Calculation:
- Major proof point: +0.3
- Persona refinement: +0.2
- Template variant: +0.1
Total: 0.6

Action: Auto-deploy (under 0.8 threshold)

Example 3: High Impact (0.9)

Changes:
- ICP target changed from "50-200 employees" to "200-500 employees"
- Pricing increased from $400/month to $800/month
- Value prop shifted from "reduce costs" to "increase revenue"

Impact Calculation:
- ICP segment change: +0.4
- Pricing change (100%): +0.3
- Value prop reframe: +0.2
Total: 0.9

Action: Flag for human review (exceeds 0.8 threshold)

---

Output Structure

artifacts/sales/
├── dtc-fashion-smb/
│   ├── materials/
│   │   ├── pitch-deck.md
│   │   ├── one-pager.md
│   │   ├── call-scripts.md
│   │   ├── email-templates.md
│   │   └── archive/
│   ├── narratives/
│   │   ├── dtc-economic-buyer.md
│   │   ├── dtc-technical-buyer.md
│   │   └── dtc-objection-lib.md
│   ├── sequences/
│   │   └── archive/
│   └── drafts/
│       └── {date}/ (if high impact)
│           ├── pitch-deck.md
│           ├── one-pager.md
│           ├── call-scripts.md
│           ├── email-templates.md
│           └── REVIEW-NEEDED.md
├── marketplace-reseller/
└── department-store-early-adopter/

---

Error Handling

Canvas missing: STOP - Canvas is required input

ERROR: Canvas not found at strategy/canvas/
Required files: problem.md, solution.md, unique-value-proposition.md
Action: Create Canvas first using canvas-design skill

ICP missing: Use generic patterns + flag

WARNING: ICP not found at research/customer/icp/{segment}-icp.yaml
Using generic B2B SaaS ICP patterns
Action: Generate ICP using icp-generator skill for better targeting

Narrative missing: Extract from Canvas + flag

WARNING: Sales narrative not found at threads/sales/narrative/
Extracting messaging from Canvas directly
Action: Refine messaging using sales-narrative skill for better storytelling

Invalid segment: ERROR

ERROR: Segment '{segment}' not found
Available segments: [list from existing Canvas directories]

---

References

This subskill relies on detailed frameworks in references/:

• pitch-deck-structure.md - 10-15 slide structure, examples
• one-pager-rules.md - Format constraints, layouts
• call-script-timing.md - Timing management per call type
• email-formulas.md - Cold outreach, follow-up patterns
• problem-first-ordering.md - Correct vs wrong ordering
• feature-to-benefit-translation.md - Translation patterns
• stage-appropriate-messaging.md - MVP vs Growth vs Enterprise
• objection-handling-patterns.md - Common objections + responses

Note: These references are in the parent sales-materials skill directory and are accessible via relative path.

---

Success Criteria

Well-generated materials enable:

1. Pitch deck presentable in 15 minutes without edits
2. One-pager understandable in 30-second skim
3. Call scripts keep calls on time and on track
4. Email templates get >20% response rate (industry benchmark)
5. All materials pass quality checks (problem-first, quantified, specific)