Performance Max reports the best return in most ecommerce accounts I audit. That is usually a reporting artifact rather than a result. Somewhere inside the campaign, Google is buying clicks on the store’s own name, and most of those orders were arriving regardless.
Proving it used to take a geo holdout and a patient client. It no longer does. Google now publishes individual search terms for Performance Max, supports negative keywords at both campaign and account level, and lets you exclude brands from specific inventory. The audit below uses all three and takes an afternoon.
What a brand leak is, and why the system creates one
None of this is Google acting in bad faith. Performance Max is a bidding system pointed at a target you set, and it takes the cheapest available path to that target. In a typical store, brand queries convert several times better than everything else and cost less per click. Give the system a return target and no instruction about brand, and your own name is the most efficient inventory it can find.
The cost shows up in the number you report. Campaign return becomes a blend of demand you created and demand you already had, and a blend is not something you can set a target against. Every decision downstream, budget increases included, gets made on a figure that is partly a measurement of your own brand equity.
Step 1: Pull the search terms
Go to Search terms in the Campaigns menu, then switch the dropdown to the Performance Max search terms report. You get individual terms with their landing pages and ad formats, and history reaching back to March 2023. Segment by ad format so Shopping ads and text ads sit in separate rows, because the remedy for each one is different.
Export it and classify every term. Use a regex on your brand token and its misspellings rather than working down the list by eye, because a four thousand row export will beat your patience long before it beats your judgment.
Five buckets are enough:
1. Pure brand. The brand name on its own, plus misspellings, plural forms and spacing variants.
2. Brand plus product. “brandname trail shoes” and the like. This is still brand demand, and the shopper has already chosen you.
3. Brand plus qualifier. Reviews, discount code, sizing, returns, login. Often existing customers.
4. Brand plus competitor. Genuinely contested, so keep it separate from the rest.
5. Nonbrand. Everything else.
One caveat belongs in your notes before you present any of this. The report covers Search and Shopping inventory, so it says nothing about Display, YouTube or Discover impressions. What you are calculating is a brand share of search-originated traffic, which is where the leak lives, not a brand share of the whole campaign.
Step 2: Size the leak in dollars and in orders
Most audits produce one number where they need two. Brand share of cost tells you what you are spending. Brand share of conversion value tells you what the campaign’s reported success is actually made of.
Then rebuild the campaign without brand. Strip brand cost and brand conversion value out, and calculate the return on what is left. That figure is what your nonbrand advertising is genuinely delivering, and it is the only version of the number that a target should ever be set against.
The gap between the two is usually where the conversation changes. A campaign reporting a comfortable blended return can be running well under break-even on nonbrand once the store’s own name comes out, and nobody in the account had a report that would have shown it.
Corroborate it outside the platform while you are in there. Pull clicks for the same brand queries out of Search Console and compare the trend against paid brand clicks. Total brand search volume sitting flat while paid brand clicks climb is about as clear as this evidence gets.
Step 3: Decide whether the leak is worth plugging
Not all brand spend is waste, and an audit that recommends cutting it unconditionally is a bad audit. Four situations make paying for your own name defensible:
- Competitors or affiliates bid on your brand, so you lose the top slot the moment you stop.
- Resellers and marketplace listings outrank your own product pages, which means the click you decline lands on a page you do not control.
- The brand name doubles as a category word, so the traffic is genuinely mixed and cannot be cleanly separated.
- You hold the top organic result but the Shopping unit sits above it, so the visible top of the page is paid whatever you do.
The point of the audit is not to stop paying for branded searches. It is to pay for it deliberately, at a price you chose, in a line item you can read, instead of accidentally at whatever Performance Max decides your name is worth this week.
Step 4: Carve it out with the right lever
Three controls exist and they cover different inventory, which is why reaching for the wrong one produces results nobody can explain.
- Negative keywords on the Performance Max campaign: These apply to Search and Shopping inventory only. Precise, list-based, and available at account level too when you want brand out of everything at once.
- Brand exclusion lists: These reach Search, Shopping and YouTube search inventory, and they carry a checkbox that allows Shopping ads to keep serving on searches mentioning the excluded brands. For a retailer that checkbox is the setting worth knowing, because it pulls your text ads off brand queries while keeping your products in the shopping unit.
- A dedicated brand campaign: Exact and phrase match, its own budget, its own ceiling. The spend does not vanish, it becomes visible and priced by you.
Sequence matters more than the choice. Build and launch the brand campaign first, confirm it is serving, then apply the exclusion. Do it the other way around and you hand the top of your own results page to whoever was bidding second, which is a memorable way to learn the lesson.
How you structure the rest of the account governs how much of the recovered budget does any work, and the full setup I use for Google Ads Performance Max for Shopify stores is written up separately.
Step 5: Measure the right thing for long enough
Campaign return will fall once the carve-out lands. That is the expected outcome rather than a failure, and it is the moment most brand carve-outs get quietly reversed by someone reading a campaign dashboard.
So settle the measurement before you touch a setting. Read total account spend, total revenue, total orders and new-customer orders, over a fixed window of at least four weeks, and write the window down so nobody shortens it on day nine.
Two outcomes, and they are easy to tell apart. Total revenue holding while total spend drops means the brand spend was buying orders you already had. Total orders falling by roughly the count of brand conversions you removed means those orders needed the ad, and the answer is to reinstate the brand deliberately in its own campaign rather than back inside Performance Max.
The rigorous version of all this is a geo holdout, and PPC Hero has already made the incrementality case against running Performance Max on brand. What I have described is the account-level version, the one you can run this week without designing a test.
What the audit usually turns up
Across the accounts I look at, brand share of Performance Max cost tends to be smaller than the client feared, and brand share of conversion value much larger than they expected. That combination is exactly what makes the leak hard to see. The spending looks reasonable while the reported performance quietly does most of its work on demand the store already owned.
Google has spent the last two years handing back the visibility that made this argument unwinnable, and the run of Performance Max updates reads differently once you see it that way. The controls are all there now. Very few accounts have picked them up.
