Performance Max brand exclusions and how to measure brand cannibalization
Google has rolled out brand exclusions for Performance Max. Here is how to check whether PMax is taking your branded traffic, and how to limit it without starving growth.
Late last month Google began rolling out brand exclusions for Performance Max campaigns. If you run PMax next to a branded Search campaign, this is the first control that directly addresses the most common complaint about the format: that it takes credit for people who were already looking for you. Before you use it, though, find out whether you actually have a problem.
What Google released
Google Ads Liaison Ginny Marvin announced two brand controls on June 29, as reported by Search Engine Roundtable. Brand exclusions for PMax were rolling out to all advertisers. They let you choose brands that PMax should not serve for, including misspellings and foreign-language versions, across Search and Shopping inventory. You can exclude your own brand or third-party brands, and you add them under Settings, then Additional settings, then Brand exclusions. If a brand is missing from Google's list, you can request it.
The second control, brand restrictions for broad match Search keywords, works the other way round: ads serve only on searches that include selected brands or related products. At the time of the report it was still being launched and was not yet available to everyone.
Know the baseline rules of overlap
Google's documentation on how PMax interacts with other campaigns says a Search keyword that exactly matches a query is preferred over Performance Max. If no keyword matches exactly, the campaign with the highest Ad Rank serves. That means a well-funded branded exact-match Search campaign should win most of your brand queries on its own. When PMax still shows for brand terms, the usual causes are a Search campaign that is limited by budget, one with narrow targeting or low-quality ads, or queries that are close variants rather than exact matches.
Measure before you exclude
Do not assume cannibalization. Test for it with numbers you already have.
- Pick a before and after window. Use the weeks before PMax launched and the weeks since, ideally at least four each.
- Track total brand conversions, not each campaign alone. Add brand Search conversions and PMax conversions, then compare with the earlier total. If the sum is flat while PMax rose, PMax mostly moved credit around.
- Watch brand Search volume and impression share. A drop in brand Search impressions after PMax started, with no drop in demand, suggests PMax is winning those auctions.
- Look at cost. If blended cost per conversion on brand improved only because PMax absorbed cheap branded conversions, the apparent efficiency is not real growth.
Hypothetically, suppose brand Search delivered 1,000 conversions a month before PMax. After PMax, brand Search drops to 700 and PMax adds 400, for a total of 1,100. That is mostly redistribution with a small net gain, and PMax's reported cost per conversion would look artificially good.
Limit it, then verify the limit worked
If the numbers show overlap, take these steps in order:
- Fix the Search side first: enough budget, healthy ads, and exact match coverage of your brand terms.
- Add your brand names to the PMax brand exclusion list, including common misspellings.
- Repeat the same before-and-after comparison after two to four weeks.
The setting lives in each campaign's additional settings, so if you run several PMax campaigns, check that every one carries the list. Re-run the comparison for each campaign separately, since one with a feed and one without can behave differently around brand terms. Keep a dated note of when each exclusion went live so later swings in brand traffic can be explained.
Expect a trade-off. Excluding brand will lower PMax's reported conversions, and the campaign may look worse. That is the point: you are making it earn conversions from people who did not already know you. Resist the urge to read the drop as failure.
Also be careful with the opposite error. If your brand terms are already well covered and PMax contributes new customers elsewhere, aggressive exclusions add little. Exclude your own brand, measure, and only then consider competitor names.
The takeaway
Treat brand exclusions as a measurement step as much as a setting. Compare total brand results before and after, make the exclusion, then compare again. If the total does not move, you did the right thing; if it falls, you have learned how much PMax was really adding.