An attribution model answers: which marketing touchpoint gets credit for this conversion?
Most customers don't convert on their first interaction. A typical path might include an organic blog post, a retargeting ad, a branded search, and a direct visit before the final purchase. Attribution models determine how much credit each touchpoint receives — and by extension, which channels appear to drive the most value in reporting.
Common Attribution Models
First-click attribution All credit goes to the first touchpoint. Best for understanding which channels create initial awareness and start purchase journeys.
Last-click attribution All credit goes to the final interaction before conversion. The default in most analytics platforms — simple but systematically over-credits bottom-funnel channels like branded search and direct traffic.
Linear attribution Credit is split equally across all touchpoints. More complete than single-touch models, though it treats every interaction as equally important regardless of actual influence.
Time-decay attribution More credit is given to touchpoints closer in time to the conversion. Works reasonably well for short sales cycles where recency correlates with influence.
Position-based (U-shaped) attribution 40% to first touch, 40% to last touch, 20% distributed across middle touchpoints. Reflects the common view that discovery and close are the most meaningful moments.
Data-driven attribution Uses machine learning to assign credit based on observed conversion patterns across channels. Requires sufficient volume (~1,000+ conversions per month) to be statistically reliable. Google Analytics 4 defaults to this model when data is sufficient.
Attribution in CRO
Attribution models directly affect how A/B test results are interpreted. If a landing page experiment improves paid search conversions but users open a follow-up email before completing the conversion, last-click attribution will credit email — making the test look less impactful than it is.
When running experiments, clarify:
- Which conversion event is being measured
- What attribution window applies (7-day, 30-day, etc.)
- Whether the testing platform and analytics platform use the same model
Inconsistent attribution logic is a common source of conflicting results between experimentation platforms and business intelligence tools.
Practical Starting Points
- Start with last-click because it's available everywhere and easy to explain
- Add first-click as a comparison to identify channels doing upstream awareness work
- Graduate to data-driven when conversion volume supports it
- Use UTM parameters consistently — without them, cross-channel attribution is built on incomplete data