| name | attribution-model |
| description | Designs marketing attribution models with channel mapping, weighting logic, reporting recommendations, and implementation steps. Use when understanding which channels drive conversions. |
| allowed-tools | Read Write Glob |
| metadata | {"author":"matthewhitcham","version":"1.0"} |
Attribution Model
When to Use This Skill
Use this skill when you need to:
- Understand which marketing channels drive the most conversions
- Choose an attribution model for budget allocation decisions
- Design a multi-touch attribution framework
- Report on marketing ROI by channel
DO NOT use this skill for setting up tracking pixels, configuring ad platforms, or building attribution software. This is for designing the model and reporting framework.
Core Principle
PERFECT ATTRIBUTION IS IMPOSSIBLE — PICK A MODEL THAT IS GOOD ENOUGH TO MAKE BETTER BUDGET DECISIONS THAN GUESSING, AND BE HONEST ABOUT ITS LIMITATIONS.
Phase 1: Brief
Required Inputs
| Input | What to Ask | Default |
|---|
| Channels | "What marketing channels are you using? (paid ads, SEO, email, social, referral, direct)" | Must be provided |
| Conversion type | "What is a conversion? (purchase, signup, demo booking, lead form)" | Lead form submission |
| Sales cycle | "How long from first touch to conversion? (same day, days, weeks, months)" | 1-2 weeks |
| Budget | "Monthly marketing spend across all channels?" | $1,000-$5,000 |
| Tracking setup | "What tracking do you have? (UTMs, GA4, CRM, pixel tracking)" | GA4 + UTMs |
| Decision need | "What decision will this help you make? (where to spend more, what to cut)" | Budget allocation |
GATE: Confirm brief before proceeding.
Phase 2: Design
Model Selection
Present the options with pros and cons for the user's context:
| Model | How It Works | Best For |
|---|
| Last touch | 100% credit to final channel | Short sales cycles, simple setups |
| First touch | 100% credit to discovery channel | Understanding awareness drivers |
| Linear | Equal credit across all touchpoints | Fair baseline, multi-channel |
| Time decay | More credit to recent touches | Longer sales cycles |
| Position-based | 40% first, 40% last, 20% middle | Balanced awareness + conversion credit |
| Data-driven | Algorithmic based on actual paths | High volume, sophisticated setups |
Recommended Model Logic
For most solopreneurs and small businesses: start with position-based or last-touch with a first-touch overlay report. Graduate to data-driven when you have 500+ conversions per month.
GATE: Present model recommendation and wait for approval.
Phase 3: Build
Deliverables
1. Attribution Model Document
- Selected model with rationale
- Channel map with all touchpoints and how they are tracked
- Weighting logic explained with examples
- Known blind spots and limitations
2. Channel-Conversion Path Map
- Typical customer journeys by channel combination
- Example paths: Social ad → Blog post → Email → Purchase
- Touchpoint definitions: what counts as a "touch" per channel
3. Reporting Template
- Monthly attribution report structure
- Metrics per channel: conversions, assisted conversions, cost per acquisition, ROAS
- Comparison: attributed conversions vs. platform-reported conversions (they will differ)
4. Implementation Checklist
Phase 4: Polish
Model Validation
After 30 days, check:
- Do the attribution numbers make intuitive sense given your spend and effort?
- Are any channels getting zero credit despite known activity?
- Does the model change your budget allocation decisions?
Quarterly Review
Reassess the model fit every quarter as channels, spend, and volume change.
Example 1: Solo Service Business (3 channels, short sales cycle)
Channels: Google Ads, Instagram organic, referral
Model: Last-touch (simple, decisive, sufficient for 3 channels)
Report: Monthly cost per lead by channel, compare to close rate
Example 2: E-commerce Brand (6 channels, multi-touch)
Channels: Meta ads, Google ads, email, SEO, influencer, direct
Model: Position-based (40/20/40) with 30-day attribution window
Report: Weekly ROAS by channel, assisted conversion count, path analysis
Anti-Patterns
- Platform-reported only — every ad platform takes full credit. Facebook and Google will both claim the same conversion. Use independent tracking.
- Ignoring assisted conversions — a channel with zero last-touch conversions may be generating all your first touches. Check assisted conversions before cutting.
- Over-engineering early — data-driven attribution with 50 monthly conversions is noise, not signal. Match model complexity to data volume.
- Set and forget — attribution models need recalibration as your channel mix changes.
- Treating the model as truth — all models are wrong. Use them to make better decisions, not perfect ones.
Recovery
- No UTM tracking in place: Start tagging all links today. You cannot attribute what you do not track. Provide a UTM setup guide.
- Too few conversions for multi-touch: Use last-touch attribution and supplement with qualitative data ("How did you hear about us?" survey).
- Platform numbers do not match GA4: This is normal. Document the discrepancy and use one source of truth for decisions.
- User wants to attribute everything perfectly: Set expectations that 70-80% accuracy is excellent for attribution. Perfect is not achievable.
