How attribution works

Attribution is the process of assigning credit to the marketing touchpoints that led to a conversion.

The attribution problem

A typical customer journey involves multiple touchpoints:

Day 1: Sees Meta ad (impression)
Day 3: Clicks Google ad → visits site
Day 5: Clicks Meta retargeting ad → visits site
Day 7: Direct visit → purchases

Question: Which channel gets credit for the sale?

Each model gives a different answer. None of them are correct.

TrustData's approach: data-driven attribution

TrustData uses Data-Driven attribution, which:

1. Considers all possible combinations of touchpoints
2. Calculates the marginal contribution of each touchpoint
3. Distributes credit fairly based on actual impact

Why data-driven?

Data-Driven attribution uses your actual conversion data to answer: "What is each channel's fair share of the credit?"

For marketing, this translates to: "What is each channel's fair contribution to the conversion?"

Total conversion value: $100

Without any marketing: $0 expected value
With Google only: $40 expected value
With Meta only: $30 expected value
With both: $100 (actual conversion)

Data-Driven calculation:
- Google's marginal contribution: ($40 - $0) + ($100 - $30) / 2 = $55
- Meta's marginal contribution: ($30 - $0) + ($100 - $40) / 2 = $45

Result: Google gets 55%, Meta gets 45%

This is mathematically provable to be the only "fair" way to distribute credit.

The attribution pipeline

TrustData runs attribution daily:

Events → Sessionization → Path Construction → Data-Driven Calculation → Attribution Table

1. Sessionization

Group events into sessions:

• 30-minute inactivity timeout
• New session on UTM change
• New session on referrer change

2. Path Construction

Build the customer journey:

user_123:
  session_1: google/cpc → pageview → product_view
  session_2: meta/paid_social → pageview → add_to_cart
  session_3: direct → pageview → purchase

3. Data-Driven Calculation

For each conversion, calculate channel contributions using the dbt Fractribution model.

4. Attribution Table

Store results for querying:

Attribution dimensions

Credit is attributed at multiple levels:

Comparing to platform attribution

Why TrustData numbers differ from ad platform dashboards:

TrustData shows the unified truth across all channels, while ad platforms show their own optimistic view.