How GA4 Conversion Attribution Really Works

GA4 doesn't use a single attribution model. It uses different models depending on the report and, more precisely, on the scope of the dimension you're looking at. The same key event can be credited to one channel in one report and spread across several in another. Both can be correct.

Other platforms add their own logic on top. Google Ads, Meta and most CRMs apply their own models, lookback windows and counting rules, so numbers across platforms will almost never match.

What does Google Analytics 4 do?

GA4 attributes traffic at three scopes:

• User scope: the channel that first acquired the user.

• Session scope: the channel that started each session.

• Event scope: the channels that contributed to a key event, credited using your property's reporting attribution model.

That model is data-driven by default, and very few properties change it. But it only applies to key events viewed with event-scoped traffic dimensions. Most standard reports use session or user scope, so they never show data-driven attribution at all.

Which reports use which attribution model?

Rule of thumb: check the dimension name. A "Session" prefix means session scope, and a "First user" prefix means user scope. No prefix means event scope, which uses your property's reporting model.

This is why teams sometimes change the attribution model and see nothing move in their main acquisition reports: those reports were never using it.

Read Google's documentation on attribution in GA4. The best explanation we have found, with examples and screenshots, is on the MeasureSchool blog.

You can adjust GA4's attribution settings in Admin: under Data display, click Events, then Attribution settings. You need the Marketer role or above at property level to change them.

There are four settings:

• The reporting attribution model (data-driven by default)

• The channels eligible to receive credit for web conversions shared with Google Ads

• The lookback window for acquisition key events

• The lookback window for all other key events

Changing the reporting attribution model applies to historical as well as future data. Your past key event reports will be recalculated the moment you save, so tell anyone who reports from GA4 before you change it.

Learn more.

We (highly) recommend keeping the data-driven model. The results are harder to explain, but they give a more accurate reflection of how your marketing drives key events.

For example, if a user clicks a paid search ad and then returns the next day through organic search, the paid search activity probably drove that organic visit.

The data-driven model gives the paid ad a share of the credit. That gives you a truer view of how your channels work together, and better data to optimise your marketing with.

Understanding data-driven attribution - Google Support.

GA4 now offers three reporting attribution models:

• Data-driven: uses your property's historical data to work out how much credit each touchpoint deserves. The credit shifts depending on each path to conversion. This is the default.

• Paid and organic last click: gives all the credit to the last channel the user clicked through from before the key event, ignoring direct traffic.

• Google paid channels last click: gives all the credit to the last Google Ads click. If there was none, it falls back to paid and organic last click.

Google has removed the first click, linear, time decay and position-based models, so you may see them in older guides but not in GA4.

Tip: if you want a simple last-click view, switch the reporting model to paid and organic last click. Remember that the change applies to your historical data too.

Learn more.

The data-driven model affects all key event and revenue data reported with event-scoped dimensions.

There are two main parts to the data-driven methodology:

• Analysing the available path data to build conversion rate models for each of your key events

• Using those conversion rate predictions as input to an algorithm that attributes credit to each touchpoint

Together, these assign fractional credit to your marketing touchpoints. In plain terms, the model looks at every touchpoint to show how each channel affects the likelihood of converting. That is why you may see more credit for paid search than a last-click view would give it. For example, if a user arrived via paid search yesterday and via social today, a last-click model credits social. The data-driven model may credit paid search, or split the credit across both.

You can choose which channels can receive credit for conversions shared with Google Ads.

It's always worth peeking at the handy infographics in the settings to see how the model will work.

This setting affects all conversions shared with Google Ads.

What this means

• Google paid channels: only Google Ads paid channels can receive conversion credit.

• Paid and organic channels: paid and organic channels are both eligible for credit, but only the credit assigned to Google Ads channels appears in your Google Ads account.

In plain terms, if you select paid and organic channels, GA4 shows all channels, but Google Ads only shows its own share of the credit.

The lookback window sets how far back GA4 looks for touchpoints to credit. The default is 90 days for most key events and 30 days for acquisition key events, such as first_open and first_visit.

We strongly suggest keeping the defaults. If you do change them, note that the change is not applied retrospectively.

Consent choices now have a big effect on attribution. When a user declines analytics or advertising cookies, GA4 can't link their visits into a single journey, so fewer paths are observed in full.

With Consent Mode set up, Google tags still send cookieless pings for users who decline. GA4 and Google Ads use these to model the behaviour and conversions they can't observe directly.

That fills some of the gaps, but it means part of your attributed data is an estimate. If your consent rates change, your attribution can shift even when your marketing hasn't.

Attribution tells you which touchpoints appeared on the path to a key event. It doesn't tell you whether the key event would have happened anyway. Some things it can never see: TV, podcasts, word of mouth and offline activity that produce no click.

For big budget decisions, pair attribution with methods that measure incremental impact:

• Incrementality testing: geo or holdout tests that switch activity off in one group to measure the difference it makes.

• Marketing mix modelling (MMM): statistical models that estimate each channel's contribution from spend and outcomes over time. Google's open-source MMM, Meridian, has made this far more accessible.

Use attribution for day-to-day optimisation, and incrementality and MMM for deciding where the budget goes.

The GA4 interface only gives you its built-in models. With the GA4 export to BigQuery, you have the raw event data and can build attribution with your own rules. For example, you could adjust channel groupings, fix known tracking gaps or join in CRM outcomes.

That work also pays off when you use AI to analyse your data. Our analytics agent, Inspectre, answers questions in plain English from a cleaned, structured reporting table in BigQuery. It doesn't work from raw GA4 data, so your channel groupings and attribution fixes are applied consistently to every answer. [LINK: Inspectre page]

Good attribution starts with good data collection, so it's worth revisiting your measurement plan too.

If GA4 still doesn't match your other platforms, the usual causes are:

• Attribution models: each platform credits touchpoints its own way.

• Lookback windows: GA4, Google Ads and Meta use different defaults.

• Counting: GA4 can count a key event every time it happens or once per session, while ad platforms may count per click or per conversion action.

• View-through conversions: ad platforms may credit ad views without a click; GA4 does not.

• Conversion date: Google Ads reports a conversion against the date of the ad click, while GA4 reports it on the date it happened.

• Consent and modelling: each platform models unobserved conversions differently.

• Time zones and currency: property and account settings that don't match.

Learn more about why your GA4 numbers differ from Google Ads.

GA4 attribution isn't one number. It changes with the report, the dimension scope and the model in your settings. Most standard reports use last-click or first-touch logic, while data-driven attribution only appears for key events viewed with event-scoped dimensions. Once you know which view you're looking at, most of the apparent discrepancies explain themselves.

No attribution model captures everything, though. Consent, cross-device journeys and offline activity all leave gaps. The businesses that get the most from their data use attribution for optimisation, test for incrementality, and build on well-planned, well-structured data.

We can help you get attribution you can trust.

Training: choose from our Essentials and Advanced GA4 training courses to build your team's confidence with GA4 reporting.

GA4 health check: find out whether your GA4 setup is collecting the right data, and what to fix first. [LINK: GA4 health check]

Talk to us: if your numbers still don't add up, talk to us about how we can help.