Working with PMax is hard and easy at the same time. Setting it up is easy; there aren’t a lot of things you can do. You can either go with product feed or without it.
And if it starts to perform, most of the heavy lifting is done by the algorithm. So, again, less work for you.
However, I see a lot of mistakes being made when creating audience signals. So let’s review the common mistakes people make and how they cost them a lot of money.
Search Themes
I’ve written about Search themes before, and I would recommend reading it. But basically, Search themes, even though they look like keywords, are just signals. You can add up to 50.
But this is where people confuse them with keywords. It looks like you’re adding keywords because you type them. But this is not the case. They don’t work the same way they do in Search campaigns.
These are just recommendations, and Google may or may not consider them.
Also, don’t forget that once you add Search themes, PMax will start showing ads on Search, meaning it will compete with your Search campaigns, if you have any. So what sometimes happens is that your Search campaign performance drops, and you don’t want that.
If you add Search themes, closely monitor your Search campaign performance and Search terms report in PMax.
Not using your own data
This is probably the biggest mistake you can make with PMax. It’s an algorithm that trains on data, so your best bet to make it work is to supply that data.
- your past purchasers
- qualified leads
- New customers
- Returning customers
- …and so on
This should be your first and only audience until you see PMax starting to perform. Then you can test out broader/different audience signals.
Adding your data is as simple as just uploading your email list to Google Ads. You don’t even need integrations or some smart solutions. Just export your past purchasers and create an Audience in Google Ads by uploading those emails. Done. You might have to update that list form time to time, but it doesn’t take long.
Sure, any third-party tools that offer robust integration save a lot of time. But usually they are paid.
Adding too many audience signals
The more audience you add, the broader the signal. Here, these are in-market audience, and it is tempting to add all that apply to your business.
But that just makes the signal broader, which usually means lower intent and high cost. You don’t need that at the start. Focus on one audience that has the best chances of converting. If you sell SEO services, start with that, expand later.
Hyper-targeting
Many advertisers try to “force” narrow audiences → they starve PMax of scale. Remember that Signals are starting points, not walls. Overly restrictive lists just slow learning.
Even Google explicitly warns against relying on hyper-granular segments.
I’ve seen advertisers load a campaign with 10 tiny custom lists → campaign never exits learning. Meanwhile, broader but meaningful signals let the algorithm explore and expand profitably.
PMax is about direction, not restriction. Over-targeting kills efficiency.
There you have it. If you can avoid these mistakes, I promise you, you will save a ton of money, that you can use for scale later on.
Blogging gives me a chance to share my extensive experience with Google Ads. I hope you will find my posts useful. I try to write once a week, and you’re welcome to join my newsletter. Or we can connect on LinkedIn.
