Multi-Touch Attribution
Understand which marketing channels actually drive conversions with sophisticated attribution modeling.
Facebook says it drove 100 conversions. Google says it drove 100 conversions. But you only had 120 total conversions.
Who is right? Both are - and neither are. That is why you need multi-touch attribution.
Why Attribution Matters
The Multi-Touch Reality
Most customers interact with multiple channels before converting:
| Touch | Channel | What Happened |
|---|---|---|
| 1 | Facebook Ad | First awareness |
| 2 | Google Search | Research |
| 3 | Retargeting | |
| 4 | Direct | Returned to buy |
Question: Which channel gets credit for the sale?
The Problem with Platform Reporting
| Platform | Claims | Reality |
|---|---|---|
| Meta | 100 conversions | Counts anyone who saw an ad |
| 85 conversions | Counts last-click from their ads | |
| 45 conversions | Counts click-through purchases | |
| Total Claims | 230 | You only had 120 conversions |
Platforms over-count because they each take full credit for shared conversions.
Attribution Models Explained
1. Last Click
| Model | Description |
|---|---|
| How it works | 100% credit to final touchpoint |
| Best for | Simple analysis, bottom-funnel focus |
| Limitation | Ignores awareness channels |
Example: Facebook → Google → Email → Direct (100%)
2. First Click
| Model | Description |
|---|---|
| How it works | 100% credit to first touchpoint |
| Best for | Awareness campaign evaluation |
| Limitation | Ignores conversion channels |
Example: Facebook (100%) → Google → Email → Direct
3. Linear
| Model | Description |
|---|---|
| How it works | Equal credit to all touchpoints |
| Best for | Fair, unbiased view |
| Limitation | Does not reflect reality |
Example: Facebook (25%) → Google (25%) → Email (25%) → Direct (25%)
4. Time Decay
| Model | Description |
|---|---|
| How it works | More credit to recent touchpoints |
| Best for | Short purchase cycles |
| Limitation | Undervalues awareness |
Example: Facebook (10%) → Google (20%) → Email (30%) → Direct (40%)
5. Position Based (U-Shaped)
| Model | Description |
|---|---|
| How it works | 40% first, 40% last, 20% middle |
| Best for | Balanced emphasis on intro and close |
| Limitation | Arbitrary weighting |
Example: Facebook (40%) → Google (10%) → Email (10%) → Direct (40%)
6. AI-Powered (Data-Driven)
| Model | Description |
|---|---|
| How it works | ML model based on YOUR conversion data |
| Best for | Most accurate for your business |
| Limitation | Needs 100+ conversions for accuracy |
Example: Facebook (35%) → Google (25%) → Email (28%) → Direct (12%)
Channel Performance Analysis
Attribution Report
See true channel value with your chosen model:
| Channel | Conversions | Revenue | ROAS | Spend | Profit |
|---|---|---|---|---|---|
| 45.2 | $12,430 | 3.1x | $4,000 | $8,430 | |
| 38.7 | $10,890 | 2.8x | $3,900 | $6,990 | |
| 28.5 | $8,240 | 12.4x | $665 | $7,575 | |
| Organic | 15.1 | $4,120 | - | $0 | $4,120 |
Channel Interaction Analysis
See how channels work together:
| Path Pattern | Conversions | Avg Value |
|---|---|---|
| Social → Search → Direct | 34 | $142 |
| Email Only | 28 | $89 |
| Search → Direct | 25 | $112 |
| Social Only | 18 | $76 |
Insight: Multi-channel paths have 60% higher order value.
Assisted Conversions
Channels that help but do not close:
| Channel | Last-Click Conv | Assisted Conv | Assist Ratio |
|---|---|---|---|
| 32 | 68 | 2.1 | |
| Display | 8 | 45 | 5.6 |
| Blog Content | 5 | 38 | 7.6 |
Insight: Display and content appear weak in last-click but drive significant assists.
Configuring Your Model
Choosing a Model
| Business Type | Recommended Model |
|---|---|
| E-commerce, short cycle | Time Decay or Last Click |
| E-commerce, considered purchase | Position Based or AI |
| B2B, long sales cycle | Linear or AI |
| Brand-focused | First Click or Position Based |
Setting Attribution Window
| Window | Use Case |
|---|---|
| 7 days | Impulse purchases |
| 14 days | Standard e-commerce |
| 30 days | Considered purchases |
| 60 days | High-value items |
| 90 days | B2B, complex sales |
Configuration API
****
bash PUT /api/v1/settings/attribution { "model": "ai_powered", "windowDays": 30, "crossDevice": true, "includeViewThrough": true, "viewThroughWindow": 1 } ****Recap
Here is what you learned:
- Platform reporting over-counts - Each claims full credit
- 6 attribution models - From simple to AI-powered
- Choose based on business - Purchase cycle and focus matter
- Analyze channel interaction - Channels work together
- Use data for decisions - Allocate budget based on true value
Attribution is not about finding the "right" model - it is about understanding how your channels work together to drive growth.
Next step: Set up your integrations to ensure complete attribution data.
Key Takeaways
- 1Understand the true value of each marketing channel
- 2Choose from 6 attribution models including AI-powered
- 3See how channels work together, not in isolation
- 4Make data-driven budget allocation decisions
Frequently Asked Questions
Which attribution model should I start with?
How does cross-device attribution work?
What about view-through conversions?
Was this article helpful?
Continue Reading
Understanding Customer Journeys
See the complete path from first click to purchase. Understand which touchpoints actually drive conversions.
trackingCustomer Intelligence
Understand your customers at the individual level. AI profiles, segments, and lifetime value predictions.
trackingAI-Powered Insights
Let AI analyze your data and surface actionable insights you might have missed.