PMax cannibalisation, and how to actually measure it

Performance Max is the most aggressive credit thief in the Google Ads ecosystem. It will happily bid on your branded terms, intercept your direct traffic, and claim conversions that would have happened anyway through organic search or repeat customers. Then it reports a great ROAS and you scale spend on it. The longer this goes on, the more your account becomes dependent on a campaign that is not actually generating incremental revenue. It is just rebranding existing revenue as paid revenue. This is not paranoia. It is the documented behaviour of how PMax works. Google's auction system gives PMax wide latitude to bid wherever it thinks conversions will happen. Brand terms are conversion machines. So PMax bids on them. The conversion fires. PMax gets the credit. The brand campaign next door, which would have got the same conversion at half the cost, gets quietly starved of impressions. The measurement problem is that Google does not give you a clean way to see this. The PMax search terms report shows you a sample of queries, but it is heavily curated, and brand queries are often hidden or aggregated. Google has a vested interest in making PMax look good. They are not going to volunteer the data that makes it look bad. Here is how I actually measure cannibalisation, in order of effort. The first signal: pull the Performance Max campaign's search insights and compare to your Brand Search campaign's volume. If Brand Search impressions have dropped over the last 90 days while PMax has scaled, your brand search is being eaten. The lift in PMax conversions is partially or entirely transferred from brand. Not new revenue. The second signal: incrementality test. This is the only definitive measurement and most agencies do not do it. Pause PMax for two weeks in one geographic region, keep it running in another. Compare total conversions in both regions. If the paused region's conversions only drop slightly, PMax was mostly cannibalising. If conversions drop substantially, PMax is genuinely incremental. Most accounts I have run this test on show 30 to 50 percent of PMax conversions are not incremental. The third signal: brand exclusions. Add your brand terms as PMax brand exclusions in the campaign settings. This stops PMax from bidding on them. Watch what happens to PMax reported conversions over the next two weeks. If they barely move, your PMax was not really driving brand search anyway. If they crater, PMax was relying heavily on brand traffic to inflate its numbers. The fourth signal: assisted conversion paths in GA4. Look at how many conversions had a brand search interaction in the path before a PMax click. If brand search is touching the path before PMax in 60 percent of conversions, PMax is mostly closing deals that brand search opened. The credit is split, but PMax is taking the full conversion in Google Ads. There is also a structural signal worth watching: if your account has multiple PMax campaigns and your brand search campaign, and you check the impression overlap in the auction insights, you will see them competing with each other. PMax is bidding against your own brand campaign. You are paying twice for the same auction. What to do about it depends on the situation. For most accounts I run, I add brand exclusions to PMax as default. The brand campaign keeps the brand traffic at much lower cost per click. PMax has to earn its conversions on non brand traffic only. Reported ROAS on PMax usually drops by 30 to 50 percent, which makes the campaign look worse, but actual incremental revenue stays the same or improves. The account is finally telling the truth. Some clients struggle with this. The PMax dashboard looks worse. The reported ROAS on the campaign drops. They feel like something has gone wrong. The honest framing is that nothing went wrong. The campaign was previously claiming credit it had not earned. Now it is being measured fairly. Total revenue is the same. Total spend is similar or lower. The math is cleaner. If you have not measured PMax cannibalisation on your account, run the brand exclusions test this month. The result will tell you whether your PMax is actually doing work or just claiming credit.