Let’s Talk About Performance Max

For those unaware of what this even is, it's the newest campaign type on Google Ads. With Performance Max, Google essentially takes a 'shotgun' approach by putting (almost) all of their campaign types into a single ad unit and dynamically serving it across all of their platforms and placements. The way that you could think about it is similar to Facebook's 'Auto Placements' if that also came with regular ads, dynamic carousel ads, lead ads, messenger ads, etc, all in one ad unit.

So what is it? Google explains it in this video:

Yes, it's very cool.

While this tool is very interesting conceptually, it may come as bad news to many that Google intends to replace Smart Shopping and Local campaigns with Performance Max altogether later this year. This reminds me a lot of the restrictive tactics Facebook has used, such as entirely removing Ad Set Budgets for a period of time, which received a lot of pushback from users. The less tools at our disposal, the worse.

In some previous testing of mine, it appeared that Performance Max would entirely cannibalize all deliverability from a Smart Shopping campaign running along side it, which tells me that Performance Max is truly meant to replace Smart Shopping. In the event that you only want Shopping Ads, this can be considered a con, in that to run Performance Max you also have to run Display, Search, etc.

Again, hopefully they put a bit more thought into this before pushing any major changes through. For now, all you need to know is one thing:


Yes, you read it right. When I first heard about Performance Max, it was from a few friends who run agencies and had been early adopters to it. After they tested it on a few client accounts, they spread the word of great results to me, where I then decided to test it for myself. Skeptical of the ad format at first, I built it out and let it run for two weeks. The results did not disappoint.

Let's take a look at the first 14 days of the test:

  • 6.33x ROAS

  • $39.27 CPA

  • $0.53 CPC

Not pictured here, an average CPM of $4.21. I can't remember the last time I saw metrics this low on Facebook. Ah, this brings me back...

First, I'd like to address the fact that, yes, the AOV on this brand is probably a bit higher than normal, and therefore a $40 CPA is more sustainable. Remember, this is just the first, initial build of the campaign without any major changes. There are a ton of ways to optimize these Performance Max campaigns, and I feel very confident that doing so could get that number down even further.

Second, something that I would like to make a note of is that this started very strong in its first ~4 days and started tapering off from there. Not into the realm of 'poor' performance, but definitely down from the peaks that it reached early on. It seems that Google is getting very good at optimizing for your ideal customer pretty quickly, which is great to see.

Third, for this test I did not add any videos to the Asset Group to make use of the YouTube placement. YouTube is an entirely separate beast, and I wanted to first test the Campaign Type to see how it performs compared to similar campaigns that I would normally run.

Now, let's take a look at the following 14 day period, for a full month look-back window. This changes things slightly:

Right around the middle of this chart I increased budget by ~50% without making any major changes to the ads. I wanted to see how it would respond to budget changes without disrupting the test too much. Here are my key takeaways:

  • For this campaign I set my bidding strategy to have a goal ROAS of 400%, of which after a full month it produced a 470% ROAS. Not only did it reach my goal, it exceeded it.

  • When expanding this look-back window to include the budget increase, we can see that ROAS did take a hit. It seems that the system was performing more profitably at lower budgets, which tells me that it seems to be having a hard time unlocking new audiences. This potentially raises concerns for the campaign type around scalability, which doesn't appear to be a problem for Smart Shopping at the moment.

  • CPC remained the same. Interestingly, this doesn't appear to have changed at all. Naturally, if performance has dropped but CPCs remain the same, this tells me there is something up with the website, however there were no major CRO changes during this time. That in mind, this might be illustrating that while Google was able to find people likely to engage with your ad at the same rate, the rate at which it can find people who are likely to convert has declined.

  • It may appear as though I increased budget by 100%, but I actually only increased it by ~50%. I think that Google was holding back some of its budget in the first two weeks to flight out later in the month. This is why we are seeing the bulk of the budget deployed in the latter half of the month. Nothing too surprising here, though, as this is pretty typical for Google.

  • While CPA did increase, a $40-$50 CPA isn't half bad, especially with no major optimizations. I can't wait to see what happens when we put some elbow grease into this.

With the performance change, there are two sides of this coin for me. On one side, when increasing budgets it is normal to see CPA increase, and with a 50% increase in budget we saw only a 20% increase in CPA. However, on the other side of this coin, in the broader scope of things these budgets really aren't that high, and an increase of budget this small (monetarily, not in terms of a percentage) causing a 20% increase in CPA may not be sustainable for most.

That said, again, the budget increase came with no major campaign optimizations, so there may be room to get those numbers down as you scale budgets.

This requires more testing on my end, but regardless of the CPA increase we were still able to drive profitable traffic to our client's website. Performance Max, in my books, is performing wonderfully so far.


Well, I'm glad you asked. Or maybe you didn't, but I'm going to tell you anyway.

Performance Max is still relatively new, so I'm not sure if the 'gold standard' setup exists just yet, but this is how I built my campaign. This may change over time, and your build may differ, but hopefully this can serve as a rough guideline.


To build your Performance Max campaign (the same way I built mine), just follow these steps:


  • Create a new campaign, select Sales, and choose your Conversion Goal. I have a few Conversion Goals that populate at this point, so I reduced this to just Purchases for sake of testing

  • Select Performance Max from your options of Campaign Types, then select the Merchant Center account and geographical region you would like to use


Like Smart Shopping, Performance Max doesn't have a traditional Ad Group, so this next stage is just going to be determining more of your campaign settings before moving on to the Ad itself.

  • Choose your daily budget

  • For bidding, you can choose whatever you like to normally use, but for this campaign I used 'Conversion Value' and let Google use the 'Maximize Conversion Value' bid strategy

  • I also set a target return on ad spend at 400% - This is optional

  • Select the country you would like to advertise in


The following is based on how I built the Asset Group for my test, and is not based on best practices. Adjust as needed for your own build.

  • Choose the Listing Group you would like to use from your Merchant Center - I just used All Products for this test

  • 4x Headlines - I used the brand name as one, pinned to the first position (you can add up to 5 headlines total)

  • 2x Long Headlines (you can add up to 5 long headlines total)

  • 2x Descriptions (you can add up to 4 descriptions total)

  • 4x Landscape Images (you can add up to 15 images total)

  • 4x Square Images (matching the landscape images, just a different crop)

  • Videos - I did not use any videos or the YouTube placement for this test (you can add up to 5 videos total)

  • 1x Square Logo Image (you can add up to 5 logos total)

  • Choose your Call To Action - I updated this to only use Shop Now

  • You can replace the default account business name here if your public name is different from your legal name

  • Add an Audience Signal - This is optional, but for this test I used 'Past Purchasers' as my signal

  • On the next screen, you can include whatever Extensions you want. In this case, I used a handful of Sitelinks, Callouts, Structured Snippets, and added a Call Extension as well

That's it! You'll find that it's actually quite straightforward once you start building it, so long as you have all of the assets required ready to go. In terms of asset development, this campaign is a bit more intensive than others, but that's just due to the nature of the Campaign Type.

Good luck launching your Performance Max campaign! If you launch a test after reading this, let me know how it goes! The more data we have on this new Campaign Type, the better off we will all be in finding success with it.

Join the conversation

or to participate.