PMax brand exclusions: exclude brand terms, or let them run and measure the overlap?

Brand exclusions in PMax look like the clean fix for a store with strong brand demand, but they're messier than they seem. Excluding brand terms can still leak brand queries through shopping and display placements, and you lose sight of how much revenue was brand demand to begin with. Leaving brand in has its own problem: PMax takes credit for sales you'd probably get anyway, so ROAS looks better than the incremental value really is.

My leaning is to keep brand traffic in for a quarter, pull the brand vs non-brand split from the search query data, and only exclude once you know what share of conversions would have happened without the ad. How are others handling this when brand search volume is high?

The query report tells you where the brand clicks landed, but not whether those sales actually depended on the ad. PMax will claim any brand click it touches, so its brand ROAS will always look strong.

The test that answers it is total brand conversions, paid and organic combined, pulled from GA4 or your CRM. Compare a period with brand excluded against one with it running. Seasonality distorts before-and-after comparisons, so a geo holdout is more reliable if you have the volume to hold out a region.

If total brand conversions barely move when you pause brand, exclude it and shift the budget to non-brand. If they drop noticeably, the ad is doing real work, and the ROAS number was closer to honest than the skeptics assume.

Agreed on the holdout. One thing that gets missed a lot: brand exclusions in PMax work as account-level negatives, so make sure the brand list isn't also catching your own generic product names. A model number people search without your brand is a good example. Those can quietly drop out of your non-brand reach.

First thing to check is what the brand list actually matches. Brand exclusions block the brand name and anything containing it, so a product line like "Northwind Trail 40" gets caught even when the shopper just wants a backpack and isn't looking for your store. If your brand is also a common word, the list can strip out a real chunk of non-brand reach. Filter the PMax search terms for queries that contain the brand term but aren't navigational. If that slice is large, the list is too broad, and your brand vs non-brand split will be off before the holdout even starts.

Agreed, and it leaks the other way too. A brand list only catches the spellings you entered, so "northwnd" or "north wind" slips through as non-brand and inflates that bucket. Scan the search terms for common misspellings before you trust the split.

Search Console is where I'd build the typo list. Organic brand queries show up there with the spellings people actually type, and the PMax search terms report hides a lot of low-volume queries, so the misspelling tail is undercounted there. Filter GSC for queries containing part of your brand and sort by impressions. Any variant with a few dozen impressions is a candidate for the exclusion list. Those people already find you on Google, so they're the same searchers PMax will be bidding on.

Check each one before you add it, though. Some misspellings are ordinary words or belong to another company, and excluding those would quietly cut non-brand reach, which is the problem Kai described.

Good call on GSC, but if the misspelling is also a real word or another company's name, the impressions will look just as big as for a true typo, so volume alone won't tell you which one you're dealing with. The quickest way to find out is to search the variant yourself. The results page will show you whether people typing it are after your brand or something else entirely.

The SERP check is the right test. I'd run the variant in a logged-out or incognito window and see whether the Shopping ads and the People also search for block are about your brand or some other product. If a misspelling pulls in another company's shopping results, that's a clear sign to keep it out. I'd also check the volume in Keyword Planner before committing, because a term with real volume that belongs to a competitor's brand is the one most likely to quietly eat your non-brand reach if you exclude it by mistake.

The Keyword Planner step doesn't hold up for me. Volume tells you how many people type a term, not whether excluding it costs you anything you wanted. A competitor's brand name with real volume is usually one you'd exclude anyway, since PMax clicks on rival names tend to convert poorly and Google's trademark rules make those ads a headache. The term that actually hurts is a generic word with lots of unrelated searches, and Keyword Planner shows that as high volume too, so the number can't tell the two cases apart.

For a local business, check location instead. The SERP test only means something if you run it from inside your service area, because local packs and ads shift by city. Say a dental clinic runs a misspelled brand query from its own town and sees its own listing, while the same query from another state shows a different company. Test from a few places where you actually take bookings, not just from the office.

I'd push back on excluding competitor names by default. People who search a rival's name are often comparison shopping and close to buying, and trademark rules mostly limit what your ad copy can say, not whether the term can trigger your ad. Pull up those queries in the search terms report and look at their conversion rate before you exclude anything on principle.