Why last click attribution is still wrong - and what to do about it
Last click attribution persists in paid search reporting partly because it is simple and partly because switching to something better is harder than it sounds. But it causes real strategic mistakes.
Here is a situation I see regularly. A Display campaign appears to have terrible CPA on last-click attribution. An agency recommends pausing it. The client agrees. Branded search volume drops a month later. Direct traffic reduces. Overall conversion volume declines. The Display campaign, which was doing top-of-funnel awareness work and influencing later conversions that got attributed to brand search, was more valuable than last-click data suggested.
The specific distortion in paid search
Branded search almost always looks excellent on last-click because it captures conversion credit from customers who were influenced by other channels earlier in the journey. Non-brand search looks good because it captures mid-funnel searchers who were already aware. Upper funnel activity - Display, YouTube, Demand Gen - looks poor because it rarely captures the final click. Last-click attribution pushes investment toward bottom-of-funnel channels and starves upper-funnel channels that feed them.
The data-driven alternative
Google Ads defaults to data-driven attribution for accounts with sufficient conversion data. Switch to this in your conversion settings and compare how channel performance changes. Typically, branded search CPA goes up (it is getting less exclusive credit) and non-brand and discovery channels go down (they are getting more credit for their contribution). This gives a more realistic picture of where the value is being created.
Comparing models
In Google Ads, you can use the Model Comparison tool to see how different attribution models change your reported conversions and CPA across campaigns. Run a comparison between last-click and data-driven and look at where the biggest differences are. These differences tell you where last-click is most misleading in your specific account.
Setting expectations
When you switch attribution models, reported performance changes even though actual performance has not. Brief your clients and stakeholders in advance. A campaign that looked like it had a £40 CPA on last-click might report a £65 CPA on data-driven because it is now only getting partial credit for multi-touch conversions. That does not mean it got worse. It means it is now being measured more honestly.
Found this useful?
Start a conversation - no pitch, no pressure.