Attribution is the art of assigning credit to marketing touchpoints that contributed to a conversion. In GA4, you pick an attribution model that determines how that credit is distributed across channels. Most marketers either leave the default in place without thinking about it or switch models without understanding the implications. Both approaches produce misleading data.

The Models Available in GA4

GA4 currently offers four attribution models:

Data-driven attribution (DDA) uses machine learning to assign fractional credit to touchpoints based on their actual contribution to conversions, as measured across your account's data. This is the default for accounts with enough conversion data. It is the most accurate model if your data volume supports it.

Last click gives 100% of credit to the last touchpoint before conversion. This is what old Google Analytics defaulted to, and it systematically undervalues upper-funnel channels like display, social, and organic search discovery terms.

First click gives 100% of credit to the first touchpoint. Useful if you want to understand acquisition channel effectiveness in isolation. Rarely the right model for ongoing optimization.

Linear distributes credit equally across all touchpoints in the path. Better than last-click for multi-channel businesses, but the equal weighting assumption is usually wrong.

Time decay gives more credit to touchpoints closer to conversion. Reflects the reality that a retargeting ad the day before purchase contributed more than an awareness display ad three weeks prior. Reasonable for short sales cycles.

Position-based (U-shaped) gives 40% credit to first touch, 40% to last touch, and distributes the remaining 20% across middle touchpoints. A pragmatic choice that acknowledges both acquisition and closing channels.

Why Data-Driven Attribution Is the Right Default (With Caveats)

For accounts with sufficient conversion volume, DDA is the most accurate model available. It does not assume that any touchpoint position is inherently more valuable. It learns from your actual conversion data.

The caveat is data volume. GA4 requires a minimum of 400 conversions in a 30-day window for DDA to activate on a specific conversion event. For lower-volume businesses or infrequent conversion types (like phone calls or form submissions for high-ticket services), DDA may not have enough data to be meaningful. In those cases, position-based or time-decay models are more defensible choices.

The Reporting vs. Attribution Model Distinction

Here is where most GA4 users get confused: GA4 has two attribution settings that operate differently.

Reporting attribution model: Used for standard GA4 reports (Acquisition reports, etc.). You can change this in Admin > Attribution Settings. This does not change historical data -- it changes how future reports are calculated.

Ads-preferred last click: Applied to Google Ads reporting specifically. When your GA4 and Google Ads accounts are linked, conversion data exported to Google Ads uses last-click attribution by default unless you override this. This means your Google Ads campaigns may be getting more or less credit than your GA4 reports suggest.

Run a comparison: look at your Traffic Acquisition report with the default model, then switch to last click in the comparison view. If there is a significant difference, you have touchpoints that are being over or undervalued in your current default model.

Setting Up Attribution Correctly

1. Go to Admin > Property Settings > Attribution Settings
2. Set your reporting attribution model. For most businesses with multi-channel acquisition, position-based or data-driven (if you meet the threshold) is the right starting point
3. Set the lookback window. Default is 30 days for non-direct channels and 90 days for direct. For longer B2B sales cycles, extend to 60 or 90 days for non-direct
4. In Google Ads, confirm that imported conversions are using the attribution model you intend

The Multi-Channel Problem GA4 Still Has Not Solved

GA4 is session-scoped by default. If a user clicks an organic result on their phone and converts on desktop three days later, GA4 may not connect those sessions unless the user is logged in to a Google account (enabling User ID graph). This means your cross-device, cross-session attribution is incomplete regardless of which model you choose.

For businesses where this matters (most e-commerce), you need to implement user ID tracking. Push your own user ID into GA4 using the user_id parameter when users are logged in. This dramatically improves cross-device attribution accuracy.

What to Actually Report to Stakeholders

Attribution models are for internal optimization decisions, not for explaining your entire marketing ROI to a CFO. The right framing for leadership reporting: show total channel contribution across models (DDA), show cost per conversion by channel, and flag attribution uncertainty where it exists.

Avoid the trap of choosing an attribution model that makes your favorite channel look good. Last-click attribution systematically favors brand paid search and retargeting, which are often the cheapest channels to optimize to but the most dependent on upper-funnel work you are not crediting properly.