Article in English.
GA4 attribution assigns credit for key events to eligible touchpoints. Understand the available models and dimension scopes before comparing channel performance.
Record the reporting assumptions
Keep the selected key event, date range, attribution settings and report dimension beside an exported chart. Compare changes in credit separately from changes in actual orders or leads. The traffic-source guide explains why acquisition reports answer a different question.
Which attribution models are available?
Google documents three models: data-driven, paid and organic last click, and Google paid channels last click. First-click, linear, time-decay and position-based models were retired in 2023. Do not plan a current GA4 implementation around those retired options.
See Google's attribution overview and attribution models report for the current definitions.
Attribution and acquisition answer different questions
An acquisition report may describe the source associated with a user or session. An attribution report assigns credit to eligible interactions for selected key events. Check the dimension scope before comparing charts.
For example, asking how a user was first acquired is different from asking how a recorded enquiry's credit is distributed. These can both be useful without producing the same channel totals.
How to review a model comparison
- Confirm the event represents a useful outcome and is being recorded correctly.
- Open the Advertising attribution reports and choose the relevant key event.
- Record the available date range, model and reporting settings.
- Review where channel credit changes and which paths are represented.
- Compare the finding with customer and campaign evidence before changing spend.
An increase in a channel's attributed credit does not establish that the channel generated additional orders. Keep changes in recorded outcomes separate from changes in credit assignment.
Common interpretation mistakes
- Comparing differently scoped reports as if they describe the same thing.
- Treating an analytics model as a controlled experiment.
- Ignoring incomplete measurement, consent choices or unobserved interactions.
- Interpreting a model switch as a change in customer demand.
- Cutting an early-stage channel solely because it receives little last-click credit.
A reviewable reporting note
Alongside the chart, write the question, selected event, report scope and relevant configuration changes. Include any known tracking gaps. This helps another person reproduce the view and judge its practical limits.
For a low-volume service business, also review actual enquiries and follow-up outcomes. A small number of observed paths may not justify a confident allocation of marketing budget.
Continue the measurement review
Use the website traffic-source guide to understand acquisition dimensions and the contact-form tracking guide to validate the underlying outcome.
If you need help implementing the measurement plan, review our SEO and analytics-related services.