Last Click Is Dead. Your Default Report Still Runs on It.

Every Monday the same report lands: last-click revenue by channel, and the same argument follows. Paid search claims the sale. Email claims the sale. Organic claims the sale. The report is not wrong in the way people think. It is running a model that GA4 retired in 2023, when data-driven attribution became the default and last-click was removed as an option for most properties. The store never rebuilt its default views, so the team is still making budget decisions on a model that no longer exists.

What changed in 2023

In 2023 GA4 made data-driven attribution the default model and removed last-click, first-click, linear, and the other rule-based models from standard reporting. Data-driven attribution assigns credit across the touchpoints in a conversion path based on modeled contribution, not on a fixed rule. A shopper who saw a Meta ad, clicked a Google ad, and came back through email gets the sale split across the three, in proportions the model estimates from the pattern of similar paths. The change was not cosmetic. It changed what every channel's number means, and a team that still reads the numbers as last-click is reading a different model than the one that produced them.

The trap is that the old views survived. Saved reports, Looker Studio dashboards, and Monday-morning spreadsheets built on the old model kept running, because nobody rebuilt them. The numbers look familiar, the channel names are the same, and the totals still add up. The only thing that changed is what the numbers mean, which is the one thing nobody checked.

What the default view still hides

A last-click view hides the assist. Email looks like it does nothing, because it rarely gets the last click, and then the team cuts email, and revenue drops, and nobody connects the two. Paid search looks like it does everything, because it catches the shopper at the bottom of the funnel, and then the team pours money into the bottom and starves the top. The view also hides the traffic mix problem: blended conversion rate is a mix of new and returning, mobile and desktop, paid and organic, and a change in the mix looks like a change in performance. A store that reads blended CVR as a site problem is fixing the wrong thing.

The fix is not a new tool. It is a rebuild of the default view: data-driven attribution as the model, conversion paths as the report, and new versus returning split out before any channel comparison. The question is not which channel got the last click. The question is which channels were in the path, and what happens to revenue when one of them is removed.

The rebuild, in order

First, confirm the model: GA4 Admin, Attribution Settings, and check that data-driven is the model in use. Second, rebuild the Monday report on conversion paths, not on last-click revenue, and split new versus returning before comparing channels. Third, run a holdout before cutting anything the new view says is weak: pause the channel for two weeks in one region and watch total revenue, not the channel's own number. A channel that looks weak in last-click and strong in the holdout is a channel the old model was lying about. For the platform side of the same mismatch, see why Meta and GA4 never match, and for the test that settles it, incrementality tests without a data team.

Frequently asked questions

When did GA4 drop last-click attribution?

In 2023, when data-driven attribution became the default model and the rule-based models, including last-click, were removed from standard reporting. Saved reports and dashboards built on the old model kept running, which is why teams still see last-click numbers.

What is data-driven attribution, in one sentence?

It assigns credit across the touchpoints in a conversion path based on modeled contribution, instead of giving all the credit to one touchpoint by a fixed rule.

Why does email look useless in a last-click view?

Because email rarely gets the last click. It assists earlier in the path, and a last-click view gives the assist zero credit. Cut email on that evidence and revenue drops, because the assist was doing work the report could not see.

Should we compare channels on blended conversion rate?

No. Blended CVR mixes new and returning, mobile and desktop, paid and organic. Split new versus returning before comparing channels, or a change in the traffic mix will look like a change in performance.