Fix Your Performance Max Campaign That Is Wasting Spend

Your Performance Max campaign looks fine on paper. But the settings might be draining your budget in five different ways right now.

You do not always need a new bidding strategy or a fresh asset group. Sometimes the fix is just a proper foundation. Chris walks through every PMAX setting, top to bottom, and shows exactly what breaks when you get each one wrong.

The CSS setting that costs you 20% of every bid

Near the top of your campaign settings you will see a field called CSS - comparison shopping service. By default it says Google Shopping CSS. Most advertisers leave it there without knowing what it costs them.

Here is the story. Years ago the European antitrust commission fined Google billions for abusing its market dominance in Google Shopping. Part of the ruling meant Google had to pay a 20% markup on every click made through its own CSS. Google's response was simple: they did not pay it themselves. They passed it on to advertisers.

In practice that means a €1 bid through Google CSS has an effective bid of just 80 cents. Google takes the 20% off the top. You never see it. You pay the full €1 CPC but you only bid 80 cents in the auction.

Any CSS that is not Google's own is unaffected by that ruling. Chris uses the ZenoX Media shopping CSS on accounts. There are others like Channable and Product Hero. They all work the same way. The switch itself takes two minutes inside the campaign settings.

This one applies to European advertisers. If you are running ads in Europe and using Google CSS, you are paying that markup right now. See the full Performance Max best practices guide for more on how the auction works.

The conversion goal mistake that fakes your results

Conversion goals sounds boring. It is one of the most dangerous settings in the account.

Chris sees this too often: an advertiser sets up tracking and does not configure which actions are primary. So the campaign tracks purchases, add-to-carts, begin checkouts, and page views - all as conversions. The dashboard shows 300 conversions in a day. The advertiser wonders why they are not getting 300 sales.

They are not. They are tracking 300 different micro-events and calling all of them "conversions." The campaign then uses that data to decide who to bid on. It optimises for people who look at a page, not people who buy. You are feeding the algorithm junk and wondering why the results feel off.

Chris says this is way more common than you would think. The advertiser sees something tracking, assumes it is working, and moves on. It is not working. It is just counting everything that moves. Fix this before you touch anything else.

The customer acquisition setting that burns money on people who already know you

This setting matters more as your store grows. For small stores just getting started, Chris says do not overthink it - leave it alone.

But for bigger brands, this is one of the most costly mistakes he sees. PMAX is aggressive in the lower funnel. It chases people who are already looking to buy. And if you have an existing customer base, a lot of those people are already looking for your brand specifically.

Chris has seen a PMAX campaign spending €2,000 a day where 80% of the conversions came from existing customers - people who were already searching for that brand by name.

I have seen PMAX campaigns spending 2K a day and 80% of the conversions are from existing customers already looking for that brand. The ROAS looks amazing. Of course it does - you are targeting warm traffic.

The ROAS on those campaigns looks great. Warm traffic converts well, the CPC is low, and everyone's happy. But you are spending fifty, sixty thousand a month to reach people who were going to buy anyway. You are not buying new customers. You are paying Google to show your ad to your own audience.

The fix is in the customer acquisition settings. You can set campaigns to bid for new customers only. For bigger brands with an existing customer base, separating branded and non-branded traffic is not optional - it is one of the highest-leverage changes you can make. This applies across Meta, TikTok, and Google, but PMAX makes it especially bad because it is so aggressive at chasing existing demand.

The location setting that sends your ads to the wrong country

This one is sneaky. The default location targeting option is "presence or interest." That sounds fine. But it means your Germany campaign can show to someone in Indonesia, the Philippines, or South Africa - as long as Google thinks they have shown interest in Germany.

That Shopify traffic from random countries? This is often why. Even if it is just 1-3% of clicks, it adds up. And it almost never converts.

The fix is to pick the second option: "presence - people in or regularly in your target location." That keeps your ads to people who actually live there or spend regular time there. It is one checkbox but it changes where your money goes.

Chris also adds the local language plus English to language targeting. A lot of people run their browser in English even when they live in Germany. If you leave English out, you miss them.

Asset optimization - the switch that breaks your feed-only setup

Chris is a big fan of feed-only PMAX campaigns. Force Google to show your ads on Shopping. Do not spread budget across Search, YouTube, Display, and Discovery if you have not intentionally set those up as separate campaigns.

There is a new beta feature called "channel performance" that now shows you the split - how much comes from Shopping versus other channels. When it is clean, you see most spend landing on Shopping. When the settings are wrong, you see a big chunk going to Search or Display with ads that are not even using your product feed.

The asset optimization settings are where this breaks. If you enable them, Google gets freedom to pull images from your landing page, generate text from your site, and create new assets. It then uses those assets to run display ads, search ads, and more - even if you set the asset group up as feed-only.

Google also likes to push recommendations that automatically turn these settings back on. Check in occasionally and make sure they have not switched themselves back.

For devices, Chris keeps it to computer and mobile phone. Tablets and TV screens have lower conversion rates for ecommerce. They work for awareness campaigns but not for a conversion-focused PMAX.

Excluding app categories to stop junk placements

The last one is in content suitability settings. Google loves to fill your budget with cheap display and app placements. The CPCs look fine. The conversion rate is not.

Chris excludes app categories at the account level as a default - all 140 of them. You go to excluded placements, add the app categories, save, and you are done. This matters even more once you start running Demand Gen or YouTube campaigns, but it is worth doing on every PMAX campaign too.

Bid strength increase

+25%

App categories to exclude

140

Existing-customer conversion share (worst case seen)

80%

This is what it looks like when everything is set up right: CSS switched, purchases as the only primary conversion, new customer acquisition on for bigger brands, location set to presence only, asset optimization fully off, and app categories excluded. That is the foundation. Build on that and the campaign has somewhere solid to stand.

Watch the full walkthrough:

If you want to go deeper on how PMAX fits into a full campaign structure, read the Performance Max vs Standard Shopping breakdown or the Performance Max for dropshipping guide. And if you want help getting this set up on your account, see how we work.

Frequently asked questions

What is CSS in Performance Max and why does it affect my bids?

CSS stands for comparison shopping service. By default, Performance Max campaigns use Google's own CSS. Due to a European antitrust ruling, Google passes a 20% markup cost to advertisers who use its CSS. Your listed CPC stays the same but your effective bid in the auction is only 80% of what you pay. Switching to a third-party CSS such as Channable, Product Hero, or ZenoX Media CSS removes that markup and makes your bid 25% stronger for the same spend. This only affects European advertisers.

What should I set as my primary conversion in a PMAX campaign?

Purchases only. In ecommerce, your primary conversion action should be purchases and nothing else. Add-to-carts, begin checkouts, and page views should be secondary or removed entirely. Tracking multiple event types as primary conversions makes the campaign optimise for micro-events instead of actual sales.

Why is my PMAX ROAS high but new customer growth low?

PMAX is very aggressive in the lower funnel and will chase existing customers who are already searching for your brand. If you have an established customer base, a large share of your conversions may be coming from warm traffic that would have bought anyway. Chris has seen campaigns spending 2,000 a day where 80% of conversions came from existing customers. Use the customer acquisition settings to bid for new customers only.

Which location targeting option should I use in Performance Max?

Choose "presence - people in or regularly in your target location." The default "presence or interest" allows your ads to show to people anywhere in the world who have shown interest in your target country - which is usually wasted budget. The presence-only option keeps spend on people who actually live or regularly spend time in your target market.

Will enabling asset optimization break my feed-only PMAX setup?

Yes. Even if your asset group is correctly set up as feed only, enabling asset optimization gives Google permission to pull images from your landing page and generate text and display ads from your site. This pushes spend to Search, Display, and other placements outside Shopping. Turn all asset optimization settings off and check periodically that Google recommendations have not silently re-enabled them.

Why should I exclude app categories in Performance Max?

Google will spend your budget on cheap app placements if you allow it. These placements often have low CPCs on the surface but very low conversion rates for ecommerce. Excluding all 140 app categories in content suitability settings stops that budget leak. Do it at the account level so it also covers Demand Gen and YouTube campaigns you add later.

Five settings. Fix them once. Stop the bleed.