Creative supply planner

Creative Supply Planner

Design the creative supply chain needed to execute campaigns without bottlenecks, asset gaps, QA failures, or performance fatigue. Treat creative assets like construction materials: the right assets must be specified, produced, versioned, delivered, checked, and replaced on time.

How This Skill Orchestrates

1. Script execution (scripts/asset_matrix_builder.py): Combinatorial calculator — channels × placements × audiences × messages × geos × languages = total asset matrix. Estimates production hours, identifies GenAI-eligible work, detects bottlenecks
2. Script execution (scripts/fatigue_refresh_calculator.py): Estimates when assets need rotation per channel based on empirical fatigue windows. Outputs a refresh schedule with variant gap analysis
3. Reference lookup (references/channel_format_requirements.md): Platform-specific specs (Meta, TikTok, YouTube, Google Display, LinkedIn, email, OOH, CTV) with sizes, file types, character limits, restrictions
4. Reference lookup (references/genai_production_guidelines.md): GenAI tool selection (Firefly, GenStudio, Claude, Runway, Midjourney), brand safety rules, when NOT to use GenAI, cost comparison vs traditional, production workflow
5. Upstream context (Claude): Pull channel allocation from media-routing-planner, journey channels from crm-journey-architect, audience segments from audience-segmentation-brief
6. Visual output (Visualizer): Asset matrix dashboard + production load chart + fatigue schedule + GenAI opportunities

Model decision: Combinatorial math + rule-based scheduling (Capacity layer). Not ML — we're counting permutations (channels × audiences × formats) and applying empirical fatigue windows. The GenAI eligibility scoring is rule-based (asset type → AI %, not a trained classifier).

70/30 on creative production: ~70% AI-operated (resizing, adaptation, copy variants, template fills, GenAI generation with human review, naming/tagging, spec QA). ~30% human (concept ideation, art direction, brand judgment, compliance review, final approval, complex video editing).

Quick Reference

Resource	Purpose	Usage
scripts/asset_matrix_builder.py	Combinatorial asset matrix (12 channel/placement types, GenAI eligibility per asset type, production hours estimation, bottleneck detection, 70/30 split calculation, QA checklist)	python asset_matrix_builder.py --input config.json --output matrix.json
scripts/fatigue_refresh_calculator.py	Fatigue window scheduling per channel/asset type, variant gap analysis, refresh date calculation, risk scoring	python fatigue_refresh_calculator.py --input assets.json --output schedule.json
references/channel_format_requirements.md	Platform-specific specs: Meta (feed/stories/reels), TikTok, YouTube (pre-roll/bumper/shorts), Google Display (standard + HTML5), LinkedIn, email, OOH/DOOH, CTV	Read for format specs before building the matrix
references/genai_production_guidelines.md	GenAI tool selection, brand safety rules (8 rules), when NOT to use GenAI (7 scenarios), 10-step AI production workflow, cost comparison table (traditional vs GenAI)	Read when estimating GenAI savings and setting production workflow

How to Use the Scripts

# Step 1: Build the asset matrix
from scripts.asset_matrix_builder import AssetMatrixBuilder

config = {
    "client_name": "AcmeAuto MX",
    "campaign_name": "Q3 2026 Summer Campaign",
    "channels": ["meta_feed", "meta_stories", "tiktok_infeed", "youtube_preroll", "google_display", "email"],
    "audiences": ["New prospects", "Existing owners"],
    "messages": ["Summer offer", "Trade-in benefit"],
    "geos": ["Mexico City", "Monterrey"],
    "languages": ["es"],
    "production_budget_hours": 120,
    "timeline_days": 21,
}
matrix = AssetMatrixBuilder(config).build()

# Step 2: Check fatigue refresh needs
from scripts.fatigue_refresh_calculator import FatigueRefreshCalculator

asset_groups = [
    {"channel": "Meta", "asset_type": "static", "current_variants": 3, "launch_date": "2026-07-01", "campaign_duration_days": 90},
    {"channel": "TikTok", "asset_type": "video", "current_variants": 2, "launch_date": "2026-07-01", "campaign_duration_days": 90},
]
schedule = FatigueRefreshCalculator(asset_groups).calculate()

Process

Step 1 — Define campaign objective and funnel role

What is the campaign trying to achieve? Brand awareness needs high-impact hero assets. Conversion needs variant-rich, personalized formats.

Step 2 — Map channels and placements

Pull from media-routing-planner output. List every channel and placement that needs creative assets.

Step 3 — Identify required formats and specs

For each channel/placement, lookup specs in references/channel_format_requirements.md. Note file types, sizes, character limits, safe zones.

Step 4 — Map creative variants

Define the versioning dimensions: audiences × messages × geos × languages. The asset matrix builder calculates the combinatorial total.

Step 5 — Identify production source per asset

Classify each asset: net new production, adaptation from existing, localization, resize, GenAI generation, or template-based. This determines hours per asset.

Step 6 — Identify GenAI opportunities

Review references/genai_production_guidelines.md. Score each asset type for GenAI eligibility. Flag assets where GenAI should NOT be used (regulated content, hero imagery, celebrity likeness).

Step 7 — Define QA requirements

Apply the 13-item QA checklist from the asset matrix builder: brand, legal, copy, specs, UTMs, accessibility, naming, localization, mobile, personalization, client approval.

Step 8 — Define approval workflow

Who approves what? Client approval for concepts, legal for claims/disclaimers, brand for visual identity, platform for ad policy compliance.

Step 9 — Calculate fatigue and refresh schedule

Run the fatigue calculator to determine when each channel's creative needs rotation. Identify variant gaps (current variants vs. needed for campaign duration).

Step 10 — Produce the creative supply plan

Generate the full matrix, production load, bottlenecks, GenAI savings, refresh schedule, and recommendations.

Output Format

Produce in TWO forms: inline visual dashboard (Visualizer) and structured markdown plan.

Visual Dashboard (Primary)

Render as inline HTML widget:

• Header with campaign name, total assets, total production hours, GenAI savings %
• Asset count cards — one card per channel showing asset count, production hours, GenAI-eligible count
• Versioning formula display — "2 audiences × 2 messages × 2 geos × 1 language = 8× multiplier"
• Production load bar — stacked by production source (new / adaptation / resize / GenAI)
• Fatigue schedule timeline — per channel, showing refresh dates and variant status (🟢 sufficient / 🔴 gap)
• Bottleneck alerts — capacity, video-heavy, version overload flags
• sendPrompt() buttons: "Run campaign QA for these assets" → campaign-launch-qa, "Check team capacity for this production load" → fte-capacity-sizing

Markdown Plan (Secondary)

## 🎨 CREATIVE SUPPLY PLAN — [Campaign] — [Client]

### Creative supply verdict
[Sufficient / At risk / Insufficient — with rationale]

### Asset requirement matrix
| Channel | Placement | Format | Audience | Message | Geo | Source | Hours | GenAI | Status |
[Full matrix — sorted by channel then placement]

### Versioning summary
| Dimension | Variants | Rationale |
[Audiences, messages, geos, languages, total multiplier]

### Production load
| Source type | Asset count | Hours | % of total |
[New, adaptation, resize, localization, GenAI, template]
Total: [X] assets | [Y] hours → [Z] hours with GenAI ([W]% savings)

### Creative fatigue schedule
| Channel | Type | Current variants | Needed | Refresh window | Next rotation | Risk |
[Per channel/type — sorted by risk]

### GenAI opportunities
| Asset type | Count | Hours saveable | Tools | Risk | Human review |
[Per asset type — sorted by hours saveable]

### Bottlenecks
[Each bottleneck: type, description, severity, fix]

### QA checklist
[13-item checklist]

### Recommendations
[What to produce first, what to cut, what to GenAI, what to delay]

Examples

Example 1 — Simple social campaign: User: "Need assets for a Meta + TikTok campaign targeting 2 audiences with 1 message. 3 weeks to produce." → Matrix: 2 channels × ~3 formats × 2 audiences = ~12 assets. Production: 48h (38h with GenAI). Fatigue: 3 Meta variants needed for 4-week campaign, 5 TikTok variants. Verdict: Tight but feasible.

Example 2 — Multi-market full-funnel: User: "Campaign across Meta, TikTok, YouTube, Google Display, email. 3 audiences, 2 messages, 3 geos, 2 languages." → Matrix: 5 channels × ~12 formats × 3 × 2 × 3 × 2 = 216+ assets. Bottleneck: version overload + video production heavy. Recommendation: reduce to 2 geos, consolidate audiences, maximize GenAI on display/email.

Skill Chaining

Direction	Skill	Connection
Upstream	media-routing-planner	Channel allocation → defines which channels need creative
Upstream	audience-segmentation-brief	Segments → defines audience versioning dimension
Upstream	crm-journey-architect	Journey channels → defines CRM content needs
Downstream	campaign-launch-qa	Asset list → QA readiness check per asset
Downstream	fte-capacity-sizing	Production hours → can team absorb this workload?
Downstream	weekly-control-tower	Creative fatigue → monitoring trigger