Attribution Modelling · How GA4 Credits Conversions to Channels

Attribution directly determines how channel performance is reported — and therefore where marketing budgets are allocated. Under last-click attribution, all conversion credit goes to the final channel before the conversion — typically direct or branded search. Under first-click attribution, all credit goes to the channel that first acquired the user — typically organic search or paid social. Under data-driven attribution, credit is distributed based on each channel's actual contribution to conversions. The channel that looks best depends entirely on which model is applied.

This matters because marketing teams typically allocate budget based on reported performance. A model that under-credits awareness channels (like paid social or display) relative to their actual contribution will result in under-investment in those channels; a model that over-credits last-touch channels will result in over-investment in channels that capture conversions already in progress but do not generate them. Choosing the right attribution model is not a technicality — it affects real budget decisions.

A touchpoint in GA4's attribution system is any session that occurred within the attribution lookback window before a conversion. If a user visited the site from organic search on Monday, from a Google Ad on Wednesday, and then directly on Friday before converting — those are three touchpoints in the conversion journey.

GA4 attributes conversions to channels rather than specific campaigns or landing pages at the top-level attribution reports. The conversion path — the sequence of channels — is visible in the Attribution section of GA4 (Advertising → Attribution → Conversion Paths). This report shows the most common multi-touch paths to conversion, revealing which combinations of channels tend to produce the most conversions.

GA4's attribution uses sessions as the touchpoint unit, not individual interactions within sessions. A session from a specific channel (paid search, organic social, email) is one touchpoint — regardless of how many pages were viewed in that session.

Data-driven attribution (DDA) is a machine learning-based model that analyses the actual conversion paths in an account and determines the incremental contribution of each touchpoint — the probability that removing that touchpoint from the path would have prevented the conversion. Credit is distributed proportionally based on this incremental contribution analysis.

Google has documented that data-driven attribution outperforms rule-based models in driving more efficient advertising outcomes because it more accurately reflects which channels are actually driving incremental conversions rather than simply appearing in journeys that would have converted anyway. A channel that appears in many conversion paths but consistently on paths that would have converted without it receives less credit under DDA than under linear attribution.

Data-driven attribution requires a minimum volume of conversion data to operate — Google's documentation indicates that a conversion action needs at least 300 conversions in 30 days before DDA is available for that conversion. Below this threshold, GA4 falls back to last-click attribution for that conversion action.

GA4 uses data-driven attribution as the default model for reporting on conversions in most reports. The attribution model and lookback window are configured at the property level: Admin → Data Display → Attribution Settings.

Lookback windows in GA4

The lookback window defines how far back in time GA4 looks for touchpoints to attribute to a conversion. GA4 provides separate lookback window settings for:

• Acquisition conversion events: the default is 30 days
• All other conversion events: the default is 30 days for click-based interactions; 1 day for view-through interactions (when a user saw but did not click an ad)

Longer lookback windows (90 days is available) capture more touchpoints in longer consideration cycles — appropriate for B2B or high-value purchases where the research process spans weeks or months. Shorter windows are appropriate for impulse purchases or services with short consideration cycles.

GA4 offers two variants of most attribution models: "ads-preferred" and "cross-channel" (also referred to as "Google Paid Channels Last Click" and "Cross-channel" respectively in some GA4 documentation).

Ads-preferred models: when a paid Google click and an organic interaction occur in the same session (or close in time), the paid click receives preferential credit. This benefits Google Ads' reported conversion numbers — which is why it is important to understand whether you are using ads-preferred or cross-channel attribution when comparing paid and organic performance.

Cross-channel models: credit is distributed without preferential treatment of paid clicks. These give a more balanced view of the contribution of paid vs organic channels. Google's recommendation is to use cross-channel data-driven attribution for the most accurate picture of channel contribution.

GA4's attribution analysis is in the Advertising section (left navigation → Advertising). Key reports:

• Attribution → Model Comparison. Compares conversion counts under different attribution models side by side — showing how the reported credit for each channel changes when switching from last-click to data-driven to first-click. Essential for understanding how sensitive your channel performance conclusions are to the attribution model choice.
• Attribution → Conversion Paths. Shows the most common sequences of channels leading to conversions — the actual multi-touch paths. Reveals which channel combinations most frequently appear in converting journeys and which channels most often initiate or close journeys.
• Advertising Snapshot. A high-level summary of advertising performance connecting Google Ads spend to GA4 conversion data.

Change the attribution model: Admin → Data Display → Attribution Settings → Reporting attribution model. Changing the attribution model affects how historical data is reported in GA4's conversion reports — it is not retroactively fixed; GA4 recalculates historical reports using the new model. This means a model change can create an apparent discontinuity in conversion reporting history.

For this reason, changing attribution models is significant and should not be done casually. If you are evaluating a model change, use the Model Comparison report to understand what the current data looks like under the proposed model before switching — so you can set appropriate expectations for how reports will change.

All attribution models, including data-driven attribution, have inherent limitations that are important to understand when making budget decisions based on attribution data:

• Unmeasured touchpoints. Attribution only credits touchpoints GA4 can track — sessions that fire the GA4 tag with a source/medium. Offline touchpoints (sales calls, physical ads, word-of-mouth) are not captured. Attribution data is only as complete as the tracking implementation.
• Cross-device gaps. Users who research on mobile and convert on desktop appear as two separate users in GA4 unless User ID or Google Signals (both requiring user identification and consent) are implemented. Cross-device journeys are systematically under-measured in standard GA4 attribution.
• Consent-gap modelling. Non-consenting users are modelled, not directly measured. The modelled conversions in GA4 (visible when Consent Mode is implemented) are estimates, not observations. Attribution accuracy is reduced proportionally to the non-consent rate.
• Correlation vs causation. Attribution credits channels that appear in conversion paths — it cannot directly measure whether those channels caused the conversion. A channel that consistently appears in paths but whose removal would not affect conversion volume (because users would have converted through another channel) receives credit it does not deserve. True incrementality requires controlled testing, not attribution modelling alone.

For most marketing teams, a practical attribution strategy combines several approaches rather than relying on any single model:

• Use data-driven attribution in GA4 as the primary performance model — it is the most sophisticated available within GA4 for accounts with sufficient conversion volume.
• Use the Model Comparison report quarterly to validate that data-driven attribution is not materially distorting channel performance vs last-click — specifically to confirm you are not over-crediting or under-crediting any single channel.
• Use the Conversion Paths report to understand the most common multi-touch journeys — this qualitative view of the funnel complements the quantitative credit assignment of attribution models.
• Supplement attribution data with incrementality testing (holdout experiments) for major channel budget decisions — specifically for decisions about whether to increase or decrease investment in a channel significantly. Attribution tells you what correlates with conversion; incrementality testing tells you what causes conversion.