Attribution doesn’t fail in isolation.
It reflects how your analytics system is structured—how data is collected, processed, and interpreted over time.
For a broader view of how attribution is actually determined:
The problem
Your attribution doesn’t make sense.
- Different platforms report different results.
- GA4 tells a different story than Google Ads or Meta.
- The same conversion appears to be credited to multiple channels—or none at all.
This is a common outcome in modern analytics.
What’s actually happening
Attribution is not a single source of truth—it is an interpretation.
That interpretation is shaped long before reporting—by how events are defined, how users are identified, and how logic is applied across systems.
It is an interpretation of events based on:
- how users are tracked
- how sessions are defined
- how credit is assigned
Each platform applies its own rules and assumptions.
As a result, the same user journey is interpreted differently across systems.
Why the numbers don’t agree
Attribution differences come from structural differences between systems:
- tracking gaps — users are not consistently identifiable across sessions or devices
- platform-specific logic — each system defines sessions and conversions differently (for example, Google Ads conversion tracking may count conversions based on its own attribution window and interaction rules, which differ from GA4)
- attribution models — last click, data-driven, and platform models assign credit differently
- data loss and distortion — missing signals change how journeys are reconstructed
Each system is internally consistent.
They are not consistent with each other.
This is why comparisons across platforms rarely align.
These differences are not random.
They are the result of how each system is designed and maintained.
This is not an error
It’s easy to assume attribution is broken.
In reality, attribution is working as designed—within each system.
The issue is that no single system sees the full picture.
Each platform reconstructs the user journey from partial data.
Each is directionally useful—but inherently incomplete.
Why it doesn’t fix itself
Attribution depends on the same system that produces your tracking data.
As that system changes:
- tracking becomes less complete
- user journeys become harder to reconstruct
- discrepancies between platforms increase
Without active management:
- attribution becomes less reliable
- comparisons across platforms become less meaningful
- decision-making becomes less certain
Left unmanaged, attribution drift increases over time.
What this means
If attribution doesn’t align, the issue is not the model.
It’s the data behind it.
Changing attribution models or settings does not resolve this.
It only changes how incomplete data is interpreted.
Reliable attribution depends on a system that produces a consistent signal across platforms.
The next step
Before comparing platforms or adjusting models, you need to understand how your measurement system is actually behaving.
An Evaluate engagement identifies:
- where tracking gaps affect attribution
- how different systems interpret the same events
- what is required to improve consistency
From there, you can move toward attribution that is directionally consistent and usable for decisions.
Start with Evaluate
Doug McCaffrey
Designs and maintains analytics systems that remain reliable over time.
Explore how this connects across your data estate: