The trials and tribulations of Google Performance-Max

Having recently taken over an account which was heavily investing in Performance Max I thought I’d share some of the perils of this campaign type. But also to avoid being another Google bashing article from yours truly, I will also share some of the new(er) features. These may be ones that you might not yet be taking advantage of but could help you to regain control, minimise wasted spend or at least be a stepping stone in attempting to understand what is happening behind the scenes.

What is Performance Max?

First things first, in case you’re new to the PPC game or have kept your PPC account away from this campaign type (for good reason), it’s worth clarifying what Performance Max is.

In Google’s own words:

The biggest shift in thinking when you’re planning to use a Performance Max (or P-Max) campaign type is that you must focus on targeting users, not keywords. Google is attempting to use all its ad inventory and multi-media touchpoints to find the right people for your campaign goals rather than solely relying on their hero search product. However we can beg to differ on how ‘good’ the inventory outside of search really is…

And note, it’s meant to complement, not compete with your keyword-based search campaigns (more on this later).

The Performance Max reality check

If you were to read all the literature from Google around P-Max you’d probably think it was a no brainer - targeting users not keywords, complementing your existing search campaigns not cannibalising them and getting brand exposure far and wide across all of Google’s products. However, the reality is far from rosy.

Issue 1. P-Max shouldn’t cannibalise your search campaigns, but it will

The official line is that “Performance Max complements existing search campaigns and respects your keyword targeting. If the user’s query is identical to an exact match keyword, the search campaign will be prioritised over Performance Max.”

Unfortunately over (at least) the past 10 years Google has done a very good job at making the need for hosting lots of exact match terms in an account redundant. Exact is NOT really exact anymore - there are close match variants and similar meanings matching as well as misspelling coverage that have all removed the need for exhaustive keyword lists. However, the only thing PMAX respects is IDENTICALS in your keyword lists.

It goes without saying that standard search activity should always take priority over any P-Max strategy. Your top search terms will nearly always achieve more cost effective clicks and conversions than any multi-channel activity. The management of a standard search campaign will also have shown that you don’t need to add misspellings, plurals or similar terms into your campaigns to achieve the desired outcomes (which has been a blessing at times for PPC managers).

However, if these keywords aren’t added, then running a P-Max campaign to complement your search activity runs into problems. It won’t delve into your search term matches to prevent itself from winning the auction. So you’ll find that more and more of your search performance is swallowed up in the P-Max land grab and your ability to control your budget and performance starts to dissipate.

Big shout out to Brad Geddes and his team at AdAlysis for conducting some overlap/cannibalisation research between P-Max and search and confirming what we’ve witnessed in our clients’ accounts over the years. The study can be found in this Search Engine Land article.

So what can you do?

It’s definitely worth reviewing your top search terms from your P-Max activity and if you don’t have these terms as keywords in your main search campaigns it’s worth adding them.

Similarly you can start to add some negative keywords to your P-Max activity (like brand terms and top commercial terms) to ensure that P-Max can’t show and your predetermined search activity can.

Issue 2: Google’s content network inventory is notoriously awful

If anyone has run video-only or display-only campaigns within Google Ads they’ll know the quality of media placements is suspect. Usually the first thing we do is exclude mobile apps as they’re riddled with ‘happy clickers’ who inadvertently click on ads within them in error.

Even in standard search activity, if you’ve got the search partner network enabled you’ll be sure to be picking up loads of invalid clicks and spam traffic.

But all of this traffic is fair game in P-Max and helps Google monetise inventory most advertisers would choose to avoid buying. Low CTRs and conversion rates should be clear signals on these channels that they’re not the right environments for your brand.

There is a handy placement performance report available within the reports editor. Unfortunately it only offers impression stats, but this is better than nothing to see where your ads might be being shown within the Performance Max ecosystem.

This is an example from a client which shows how awful some of the placements are, but also reveals how erratic the ad delivery across these partners can be. I was interested to see how many ad impressions have started to be shown across taboolanews.com which have shot up in 2025, but also the varied impression delivery from within Google’s own Owned and Operated inventory. There doesn’t appear to be a pattern, which I would have expected if this inventory was either being optimised in or out of my campaign performance.

Seems like Google is just dumping its inventory on advertisers

So what can you do?

Luckily you can exclude placement types and individual URLs!

To do this, head to the content suitability section in the campaign settings and look for the excluded placements section.

Issue 3: Performance Max doesn’t discriminate by demographics even when you have a clear target demographic

This has been a big bugbear for me mainly because most of my clients have an ideal target customer and core demographic. In the travel industry this often goes as far as to distinguish touring for the over 50s or gap year travel for the under 30s, and yet in P-Max, ads could be shown just as frequently for out of market groups.

The reason for this is that when setting up P-Max campaigns you can only choose audiences as signals, not specific targets. Whilst P-Max activity should be set-up to optimise towards a goal which should act as a quality signifier, we all know that tracking signals have their own flaws (this will bring me to my next point soon enough).

So if you have a free-for-all P-Max campaign check the distribution of your ad impressions and spend by age and see how much you’re potentially wasting chasing the wrong audience. If the overall cost efficiency of your P-Max campaign is okay then this may be a wastage you have to bear, but it would be wise to check in case you draw a different conclusion on the reliability of this ROI calculation.

What can you do?

This is fairly hot off the press, but the new Google Ads Editor version 2.9 will have the ability to add age exclusions, or negative age criteria, to Performance Max campaigns at the campaign level. (Hooray!) However, when we downloaded the latest version it was still throwing an error, so until this is sorted and made universally available, get in touch with your Google Account Manager and ask to be added to the beta.

Issue 4: If your conversion signals can’t be trusted, then P-Max can’t be trusted

Remember, P-Max is a goal-based campaign type meaning that it needs to work towards a target metric. But how reliable are your goal tracking metrics?

P-Max is brilliant for ecommerce because sales are tracked online and the value of a sale is also provided which can be used to give value-based targets as opposed to count-based ones. This helps to push impressions and clicks towards high value customers and it helps to bid appropriately for inventory based on the likely reward.

However, if you’re not an ecommerce business and you track enquiry forms, phone calls or simply clicks to call, your tracking set-up might not be so reliable. P-Max was certainly built with rose tinted glasses on, for a world where spam didn’t exist and where quality signals were guaranteed.

Unfortunately we’ve seen how P-Max can spiral out of control in the wrong direction when it learns bad habits from bad conversion signals.

In one instance P-Max learned very bad habits from some of the conversion signals (and the absence of a quality signifier) and chased a lot of spam leads. The spend bloated, the conversions were off the chart, but low and behold that wasn’t the reality in our customer’s database. P-Max would have thought it was doing really well so continued to chase these spam clicks and bot actions. When validating performance we had other views and swiftly ceased using it!

In another instance a client had extremely high invalid click rates, high bounce rates and obscene click to call counts from this campaign type. This data also couldn’t be filtered out or prevented since, again, the system thought it was doing what it was meant to be doing - generating calls. When faced with this risk it seems hard to justify blindly spending precious budget on P-Max.

In these scenarios there will have been ‘good clicks’ bought and genuine leads achieved, but when most advertisers have tight budgets, knowing that a lot of it is deployed in wasteful areas is hard to stomach. I genuinely believe the only real winner of P-Max adoption is Google!

What should you do?

If you have questionable conversion tracking tread lightly with P-Max. You may need to hold off venturing this far until it’s more reliable.

Or if you’re fairly confident it’s usable, push for as many quality conversion signals as possible - upgrade your tCPA bidding strategy to a tROAS one where possible.

But above all else, verify where possible, that the data that is being counted against P-Max activity is a level of merit you’re comfortable with. This may require extra tracking and reporting processes to dive deep into user journeys but if you’re to scale this campaign type you’d be wise to make sure it’s built on firm foundations.

Issue 5: Lack of insights

I know this is probably just a technical hiccup with AI Overviews but it’s pretty comical that the only time I’ve seen it happen is on this search… Here’s a breakdown… ‘oh wait, there’s nothing to say’

P-Max truly is a black-box on how budget is deployed across search, display, video and other Google environments and I genuinely don’t believe we’ll ever receive transparency from Google or a level of control we’ve experienced with search activity. It’s a great money spinner for them.

Google doesn’t want you to control how your brand is shown, where it’s shown or how often, so it makes it pretty hard to understand what it’s doing on your behalf. It places its emphasis on machine intelligence and it learning what’s right, but how often has the machine got things wrong? And how important is human knowledge to your marketing tactics and brand creative? I’d say in a world of AI there is still a lot to be said for human intelligence!

If you’re keen to dig a little deeper, there are clever people like Mike Rhodes & Tobias Hink who have built scripts to unearth some of the performance splits by platform, but these mainly suit ecommerce advertisers.

What should you do?

Move over P-Max because where there is some hope for insights is within the Demand Generation campaign type. From this month, if you’re using this campaign type then you might start to learn a lot more about where your ads are running from within the user interface itself and how they’re performing! This is meant to come for P-Max in the future as well, but currently only available to some accounts and very limited information at present about when.

To learn where your ads are showing and how you money is being spent by Google you can segment your Demand Generation campaigns by network.

The Demand Generation campaign type is very similar to P-Max, which is why I raise it here, and could be a better solution for lead generation advertisers.

Most brands jump into Performance Max because this was the first newer automated campaign type outside of standard search. It however places a lot of emphasis on shopping placements. So Demand Generation was later introduced to focus on lead generation and offer advertisers a way to reach users before they search, building engagement and nurturing potential customers across Google properties.

Be warned, where previously Demand Generation didn’t have access to Google’s display network (as per the venn diagram below from LinkedIn member Raj Singh), from this month, campaigns will have access to this inventory. We would advise opting out of this as the quality of placements is notoriously poor (as previously mentioned)!

In summary

If you hadn’t guessed already, I’m not P-Max’s biggest fan. As explained via the 5 key issues I’ve identified in this article hopefully you can appreciate why that might be. However, even I will admit that there are slight changes afoot that offer some hope for advertisers, who should be able to start controlling some core elements and be in the driving seat when it comes to budget deployment and efficiency.

Don’t let Google’s black box trickery fool you - seek the insights, use the tools available and invest wisely.