Last-Click Attribution

Last-click attribution gives all credit for a conversion to the final ad click before it happened. If a customer sees your display ad, clicks a Facebook post, and then clicks a Google search ad before buying, the search ad gets the full sale. Google's help page describes the model as giving all credit to the last-clicked ad and corresponding keyword. Google Ads still offers it, alongside data-driven attribution. Its appeal is that anyone can recompute it by hand from a click log, with no modeling and no hidden assumptions. Google's help page notes that most advertisers are used to measuring on a last-click basis, and plenty of old benchmarks and agency reports were built on it.

The rule is mechanical: look back from the conversion, find the most recent click that qualifies, and credit it. Everything earlier receives nothing, no matter how many touches there were. Because the last click is often a branded search or a retargeting ad, last click tends to favor channels near the end of the funnel. It also feeds Smart Bidding the same way: if the attribution model is last click, bidding learns from final-click patterns only, and keywords that introduce customers rarely get rewarded in the bidding data. Last click is easy to audit. Anyone with the click log can reproduce the numbers by hand, which is useful when you need to settle a dispute over a report.

In Google Ads, the model is set per conversion action, and you can change it, with the new rule applying to conversions counted going forward, not to past ones. In GA4, the equivalent choices are paid and organic last click, which ignores direct visits, and Google paid channels last click, which credits only Google paid clicks. These are separate from Meta, which counts conversions by its own click and view windows. That means the same sale can be claimed by Google under last click and by Meta under a 7-day window, so platform totals add up to more than your actual orders. If you change a conversion action's model, past reports will not match new ones for the changed period, so write down the switch date next to any trend line.

The mistake is reading last click as cause. The final click is where the path ended, not necessarily what persuaded the buyer. A customer who searched your brand name after seeing three ads gives branded search all the credit, and cutting the upstream ads then dries up branded search a month later. The opposite mistake is dismissing last click as useless. For short purchase paths, such as an emergency locksmith call, the last click is often the only click, and the model is accurate enough. It also makes a poor judge of display, video, and social, which seldom get the last click even when they start the journey that ends in a search. If a stakeholder insists on last click, give them a one-line caveat on every report: this shows where journeys ended, not where they began. That sentence prevents a surprising number of bad budget decisions.

Use last click as a baseline. In Google Ads, open the attribution model comparison and view your campaigns under last click and data-driven side by side. Where numbers match, the path is short and the choice does not matter. Where they differ, the campaign plays an assist role. This week, sort your campaigns by the gap between the two models, and protect any campaign whose credit rises sharply under data-driven. If you keep last click for audits, say so in your reports so nobody compares it to a different model unknowingly. Use it as the audit view and a second model as the planning view. The two together tell you what finished the sale and what helped it along.