Attribution model selection is a reporting governance decision, not simply a configuration preference. In an enterprise environment, the selected model influences how marketing channels are valued, how budgets are assessed, and how Google Analytics 4 (GA4) figures are interpreted alongside advertising, finance and customer data.

GA4 currently provides three attribution models within its attribution reports:

• Data-driven attribution
• Paid and organic channels last click
• Google paid channels last click

The first-click, linear, time-decay and position-based models are no longer available in GA4 attribution reporting as of November 2023, according to Google’s official documentation.

The appropriate model depends on the reporting objective, conversion volume, customer journey, data quality and required alignment with other platforms.

1. Attribution models in GA4

An attribution model determines how credit for a key event is assigned to the touchpoints in a user’s conversion path. A key event may represent a purchase, lead submission, application start or another action that has been defined as commercially significant.

GA4’s three available models have different purposes.

Data-driven attribution

Data-driven attribution distributes credit across touchpoints based on observed conversion and non-conversion paths. The model uses machine learning to assess the contribution of interactions such as advertisements, clicks, device types, timing and touchpoint order.

The model is specific to the relevant property and key event. It does not apply a fixed rule such as assigning all credit to the first or last interaction. Instead, it estimates how the presence of a touchpoint changes the likelihood of a key event.

Data-driven attribution is generally suitable for organisations with:

• Complex, multi-touch customer journeys
• Sufficient conversion volume
• Multiple paid, owned and organic channels
• A requirement to understand upper-funnel contribution
• A need to evaluate channels beyond the final interaction

Google Analytics uses data-driven attribution by default for many key event reports. The model can produce fractional credit, meaning that a single key event may be distributed across several channels.

Paid and organic channels last click

Paid and organic channels last click assigns 100% of the credit to the last non-direct channel before the key event. Direct traffic is excluded unless the complete path consists of direct visits.

For example, a path consisting of display advertising, organic search and email would assign the full value to email. A path consisting of paid search, direct and conversion would assign the value to paid search, because direct traffic is ignored in a non-direct model.

This model is straightforward to understand and does not require the same volume of data as data-driven attribution. It can therefore be useful where:

• Conversion volumes are limited
• Stakeholders require a simple rules-based approach
• Reporting must be easily reconciled with a last-click framework
• The primary question concerns the final identifiable marketing interaction

Google paid channels last click

Google paid channels last click assigns credit to the last Google Ads channel in the conversion path. If no Google Ads interaction is present, the model falls back to paid and organic channels last click.

This model is most relevant when you assess Google Ads performance in isolation or when reporting must reflect the final Google Ads interaction rather than the final interaction across all paid and organic channels.

2. The reporting scope must be understood

A common source of confusion is the assumption that one attribution setting governs every dimension and report in GA4. It does not.

The selected reporting attribution model affects reports and explorations that use event-scoped traffic dimensions, such as:

• Source
• Medium
• Campaign
• Default channel group

User-scoped and session-scoped dimensions, such as first user source or session source, are not changed by the property’s reporting attribution model. These dimensions continue to use their own attribution logic.

This distinction is important when comparing dashboards. A report using event-scoped campaign data may show fractional credit under data-driven attribution, while a report using session-scoped source data may continue to show a last-click value.

Google’s documentation on attribution settings also states that changing the reporting attribution model applies to historical and future data. This means that historical comparisons can change after a model switch. Governance documentation should therefore record the model in use during each reporting period.

Recommended action: Every enterprise reporting specification should identify the attribution model, dimension scope, key events, lookback window and reporting date basis. Without these definitions, apparently conflicting reports may be technically correct but operationally difficult to interpret.

3. Comparing the models for enterprise reporting

No model provides a complete description of customer behaviour. Each model answers a different question.

Data-driven attribution addresses the estimated contribution of touchpoints across a conversion path. Paid and organic last click identifies the final non-direct interaction. Google paid channels last click isolates the final Google Ads interaction.

These outputs should not be presented as competing versions of an objective fact. They are different analytical views of the same recorded journeys.

4. What attribution modelling does not solve

Attribution modelling cannot correct weaknesses in the underlying implementation.

It does not solve:

• Missing or duplicated key events
• Incorrect revenue or transaction values
• Broken cross-domain measurement
• Inconsistent campaign parameters
• Referral contamination
• Consent-related data loss
• Poorly defined conversion events
• Incorrect Google Ads or GA4 configuration
• Identity loss between websites, applications and account areas
• The absence of incrementality or controlled testing

A data-driven model applied to incomplete data remains dependent on incomplete data. Similarly, a last-click model may be easy to reconcile but still produce misleading conclusions if channel tagging is inconsistent or important interactions are not captured.

Attribution also does not establish causation. Credit allocation indicates how a model values recorded touchpoints; it does not prove that a channel independently caused a conversion. Incrementality testing, media mix modelling, holdout tests and broader commercial analysis may be required for investment decisions.

Recommended action: Attribution model selection should follow an implementation and data-quality review. The review should test event collection, consent behaviour, campaign governance, cross-domain journeys, revenue values and reconciliation with advertising platforms.

5. How to use GA4’s attribution reports

GA4 provides reports for comparing the impact of different attribution models on key events and revenue.

The current workflow is:

1. Open Advertising in GA4.
2. Navigate to Attribution.
3. Open the Attribution models report.
4. Select the relevant key events.
5. Select the reporting dimension, such as primary channel group, source, medium or campaign.
6. Compare the available attribution models.
7. Review changes in attributed key events and revenue.
8. Apply filters for regions, devices, campaigns or other relevant segments.

Google’s Key event attribution models report documentation explains that the report can show changes in key event and revenue values between models.

For an enterprise analysis, the comparison should not rely on a single aggregate total. It should assess:

• The channels gaining or losing credit
• The effect on high-value key events
• Differences by market and device
• Changes in branded and non-branded search
• The impact on upper- and lower-funnel campaigns
• Differences between new and returning users
• Variance between GA4 and Google Ads

The analysis should be retained as part of the reporting record rather than treated as a one-off investigation.

6. Google Ads reconciliation requires a separate view

Google’s official documentation states that Google Ads uses last-click attribution for conversions based on GA4 key events. Google Ads creates conversions only for key events where it’s the last non-direct click, even if a different model is selected elsewhere.

As a result, GA4 data-driven attribution and Google Ads conversion reporting should not be expected to match automatically.

For a like-for-like comparison with Google Ads, Google recommends selecting paid and organic channels last click in GA4. However, differences can still arise from:

• Different conversion definitions
• Different counting methods
• Different conversion windows
• Time-zone differences
• Consent and modelling behaviour
• Reporting by event time versus interaction time
• Data processing and reporting delays

A reconciliation layer should therefore document which figures are used for optimisation, executive reporting, financial analysis and platform-level campaign management.

Recommended action: Maintain separate labels for “GA4 modelled contribution”, “GA4 last-click comparison” and “Google Ads reported conversions”. Do not combine these measures under a single unqualified conversion total.

7. A practical enterprise approach

A controlled approach to attribution reporting can be structured as follows.

1. Define the reporting questions

Select the model based on the decision being supported.

For example:

• Budget allocation across the full marketing mix may require data-driven attribution.
• A finance reconciliation may require a clearly defined last-click view.
• Google Ads bid optimisation may require Google paid channels data.
• Executive reporting may require both a simple primary view and a documented supporting analysis.

2. Confirm implementation quality

Before comparing models, validate:

• Key event definitions
• Transaction and revenue values
• Campaign naming and UTM governance
• Cross-domain configuration
• Consent signals and consent mode behaviour
• Duplicate event prevention
• Google Ads linking and conversion imports

TagDataTrust’s analytics data quality service and Google Analytics consulting service address the implementation controls required for reliable reporting.

3. Select a primary model

Data-driven attribution is generally appropriate for a mature, high-volume property with complex journeys and a need for cross-channel analysis. Paid and organic last click may be more appropriate where volumes are limited or reconciliation and interpretability are the primary requirements.

The choice should be recorded as a governance decision rather than left as an undocumented default.

4. Retain a comparison view

Even where data-driven attribution is the primary model, last-click comparisons remain useful. They identify how much channel valuation depends on the selected methodology and provide a reference point for platform reconciliation.

5. Review the model periodically

The review should be triggered by:

• Significant changes in conversion volume
• New channels or campaign types
• Major consent changes
• Website or app restructuring
• New markets or domains
• Changes to key event definitions
• Material differences between GA4 and advertising platforms

Conclusion

Attribution model selection in GA4 should be treated as a reporting governance decision with implications for interpretation, reconciliation and budget evaluation. The appropriate model depends on the reporting question, the quality of the underlying implementation and the level of alignment required with advertising platforms and other business reporting.

For enterprise teams, the practical requirement is not to identify a universally correct model, but to define a primary view, document its scope and maintain comparison views where needed. That approach reduces ambiguity and supports more consistent decision-making.

If you’d like to hear more on this topic and set up a 1:1 call with us, please use the contact form. Contact a Consultant