GA4 attribution models compared: which one matches how you actually run campaigns

GA4 supports five attribution models. Each one tells a different story about the same campaign, and the right model depends on what decision you are making with the data. Most teams pick one model, never revisit the choice, and end up with reports that match neither their intuition nor their stakeholders' expectations.

Here is the practical guide to which model fits which decision.

The five models GA4 supports

1. Last-click (cross-channel). Credit goes to the last channel the user clicked before converting. Simple, intuitive, the default in most analytics tools for the past two decades. Tends to over-credit bottom-of-funnel channels (branded paid search, direct, retargeting) and under-credit top-of-funnel (organic social, content marketing, brand-awareness display).

2. First-click (cross-channel). Credit goes to the first channel the user interacted with. The opposite of last-click. Tends to over-credit top-of-funnel channels and under-credit channels that close.

3. Linear. Credit is distributed equally across every touchpoint in the conversion path. A four-touch path gives 25 percent of conversion credit to each touch. Useful for diagnosing which channels appear in conversion paths at all, less useful for budget decisions.

4. Position-based (40/20/40). 40 percent credit to the first touch, 40 percent to the last, 20 percent distributed across the middle. The classic "open and close" attribution model that recognizes both top-of-funnel discovery and bottom-of-funnel conversion.

5. Data-driven. Google's machine learning assigns credit based on observed conversion patterns in your specific account. Touches that consistently appear in converting paths get more credit; touches that appear in non-converting paths get less. Requires a minimum conversion volume (300+ conversions per model in 28 days) to enable.

Which model GA4 uses by default

In 2026, GA4 defaults to data-driven attribution for accounts with sufficient conversion volume. For lower-volume accounts, it falls back to last-click.

The default applies to GA4's Conversions report, the Acquisition reports, and most dashboard widgets. You can override the default per report or use Explore to compare models side by side.

When to use each model

There is no "right" model for every decision. Pick by the question.

The honest take on data-driven attribution

GA4's data-driven model is a black box. You feed it conversion data; it returns attribution percentages. You cannot easily inspect why it gave Channel A 32 percent credit and Channel B 18 percent credit. The opacity is a feature in some sense (the model captures patterns humans miss) and a bug in another (you cannot defend the budget allocation to a stakeholder who asks "why?").

Three honest observations after using data-driven attribution at multiple companies:

1. It tends to redistribute credit toward early-funnel channels (organic, content, social) compared to last-click. Sometimes the redistribution is justified; sometimes it inflates the contribution of channels that just happen to appear in many sessions.
2. It needs volume. Below 300 monthly conversions per model, the data-driven model is statistically unstable and GA4 falls back to last-click anyway.
3. It does not explain itself. When a stakeholder asks why paid search dropped from 40 percent of credit to 25 percent of credit between Q1 and Q2, the answer is "the model recalculated based on new patterns" and that is unsatisfying.

For most teams, data-driven works for budget allocation decisions but stakeholder reports are easier with last-click. Use both.

Setting the default model

Admin → Property → Attribution Settings → Reporting Attribution Model. Pick from:

• Cross-channel data-driven attribution (default)
• Cross-channel last-click
• Cross-channel first-click
• Cross-channel linear
• Cross-channel position-based
• Cross-channel time decay

The choice affects the Conversions report, Default Channel Group reports, and most other standard surfaces.

Comparing models without switching the default

GA4's Explore reports let you compare attribution models side by side without changing the default.

1. 01

   Open Explore.

   GA4 → left nav → Explore. Create a new exploration.
2. 02

   Pick the Funnel exploration template or Free-form.

   Either works for attribution comparison.
3. 03

   Add a Source / Medium dimension.

   Drag from the left panel into Rows.
4. 04

   Add Conversions metric multiple times.

   Drag the conversions metric into Values, three or four times. Each instance can be configured with a different attribution model.
5. 05

   Configure each metric instance with a different model.

   Click each metric, set the attribution model. Now you have last-click conversions, data-driven conversions, and position-based conversions side by side per row.
6. 06

   Compare patterns.

   Channels that get higher credit under data-driven than last-click are top-of-funnel contributors that close-attribution under-counts. Channels that drop under data-driven are last-click hogs that inflate under simpler models.

This view tells you whether your current default is over-crediting or under-crediting specific channels. The signal is the gap between models. The absolute number under any single model matters less.

What the model does not fix

A common misconception: switching attribution models will make GA4's numbers match your ad platform's numbers. It will not. The structural gap between Meta-reported clicks and GA4-reported sessions (covered in Meta Ads attribution vs GA4) is on the data-collection side. Attribution models do nothing about it.

Attribution models slice the conversions GA4 already recorded. They do not recover conversions GA4 never saw because of cookie consent, ad blockers, or in-app browser drop-off.

Common attribution mistakes

• Switching models monthly looking for the "right" one. Pick a default for stakeholder reports and stick with it. Use Explore to compare for budget decisions.
• Assuming data-driven is always best. Data-driven needs volume. Below 300 conversions per month per model, it underperforms simpler models and the GA4 fallback to last-click is silent.
• Reporting the same campaign with different models in different docs. Stakeholders see two different numbers for the same campaign and lose trust in the reporting.
• Attributing to the channel without understanding the touchpoint count. A campaign with average 6 touchpoints has very different cross-channel dynamics than one with 1.5 touchpoints. Look at touchpoint count alongside the attribution percentages.

For the broader question of why platforms disagree with GA4 in the first place, Meta Ads attribution vs GA4 covers the structural gap. For the GA4-side setup that minimizes attribution noise, UTM tracking in GA4 is the foundation.