Welcome to the in-depth guide on setup best practices for new Performance Max campaigns. This process that has helped scale thousands of successful campaigns.
The detailed process covers:
- Setting up conversion tracking so you feed the machine good data
- Campaign and asset group structure and targeting
- A pre-optimisation checklist for the learning period
Why can you trust this guide?
Because it was written by me… Ed Leake. 👋
- I was co-founded in 1982
- I’ve run my digital agency for over 13 years now
- Managed over $250 million in ad spend (keeps going up)
- Operate the Agency Forge for high performing PPC agency owners
- Created the highly rated God Tier Ads training product
- Invested in ad tech with AdEvolver
- And other stuff besides…
But before diving in, ask yourself: Are my search campaigns performing?
If not, then I recommend you stop here and fix them before even attempting Performance Max.
If the fundamentals aren’t right, Performance Max is probably not going to work for your account.
Otherwise, let’s proceed. Or if you’re already comfortable with this part, jump to my guide on Performance Max optimisation techniques to scale revenue and conversion volume.
Performance Max: The good, the bad, and the ugly
Performance Max is a prediction engine, so it’s only as good as the inputs that it’s given, so you have to feed it good information. That means accurate conversion tracking, as well as first- and third-party data.
Then, give it some money—lots of money. The complex bit is in the middle (not the ads), and then you profit.
If only it were that simple.
Alas, Performance Max is no magic bullet. It’s more like a .50 barreled shotgun that you aim blindfolded with one arm tied behind your back. If your search volumes are low, PMax is just going to blow through your budget.
So what’s good about Performance Max (before we kick it in the netherregions)?
- It performs well, especially in an existing account—provided you’ve already got data and conversions.
- It’s simplified. There’s not a lot to do in terms of campaign management, particularly for lead generation campaigns.
- It allows you to cover a wider variety of Google Ads inventory. You’ll get yourself on Display, Gmail, Discovery, and other networks.
And now, the aforementioned kick.
The bad bits about Performance Max are:
- Reporting is pretty awful. There’s no reporting on inventory or split between networks. If I want to know how much is being spent on YouTube, I can’t see that without an API connection.
- It cannibalises brand search—which is annoying, but Smart Shopping did that as well.
- There are issues with the quality of leads that come through Performance Max.
- It can overspend like mad. Display is a big network, for example.
- It’s prone to spikes in ad spend and bidding. This is essentially a machine and sometimes, it will chase a rabbit down a hole.
- There’s always a learning phase; Google calls it the ramping phase. There’s a lot of chop initially, particularly on a newer campaign or younger accounts.
- If you’ve got a low amount of data, it doesn’t perform very well. If you get five conversions a month, Performance Max probably won’t work for you.
- RIP Smart Shopping.
If you can live with that, I’ve got just what you need.
Common questions about setting up Performance Max campaigns
How do I manage Performance Max campaigns?
Performance Max is a black box. Your biggest levers are the data you feed into the system a.k.a. financial and conversion data. Your next biggest levers are budget, bidding, the way you group things and, to a much lesser degree, audiences.
Where can I find accurate conversion data?
Analytics imports have lost their edge over time, so no more until we know how good Google Analytics 4 can become. You want the Google Tag, enhanced conversions, and primary and secondary conversions—primary to bid against, secondary for monitoring and analysis. You may migrate some of those secondaries to primaries, or you may just test them on certain campaigns.
When should I avoid using Performance Max?
Don’t replace your existing campaigns with Performance Max if they’re working — with the exception of Smart Shopping, which has been deprecated in favour of Performance Max.
If your Search campaigns are delivering value and they’re gravy, why would you replace them with Performance Max?
Performance Max is not designed to replace Search campaigns, particularly for small accounts and budgets.
What do I need for Performance Max to work?
Google recommends—it’s not a prerequisite—that you have conversion tracking. I suggest that if you don’t have conversion tracking set up, you don’t go anywhere near Performance Max.
For shopping-focused accounts, you need a merchant feed and conversion tracking. It helps to also have a few responsive search ads, image ads, and video ads—Google will create awful videos for you without this.
Also, if you can’t hit the minimum conversion threshold—around 30 per month—you certainly shouldn’t be going all in on Performance Max but instead, trusting your tried and true search campaigns.
What’s the best Performance Max campaign structure?
The perfect Performance Max campaign structure doesn’t exist (you probably saw that answer coming). Keep it simple to start, especially if you don’t have a lot of data split by campaign, if you need budget and bidding control.
For asset groups, just think of ad groups and ad relevance. And answer these questions about the images and text ads you’re putting together, or the information you provide to Google to put ads together on your behalf:
- Does this ad make sense for every single thing in this asset group?
- Does the ad fit what the audience needs?
- Is it the best ad to serve for everything in this asset group?
If the answers are not “yes, yes, yes”, then split the asset group.
With audiences, less is more. Two is fine, but don’t add more until Google figures it out (and ignores what you tell it to do anyway). More audiences equals wider targeting. And layering audiences really does work if you’ve got the data.
How do I use audience signals in Performance Max?
When you start out, you want a specific custom segment. Take your top 15-30 keywords from research campaigns and put those in a custom segment. That’s effectively your in-market audience, and you apply that as your audience signal. You could also blend another segment on top of it; competitive URLs or category terms, for example. I’d stick to the specific search terms that have been working for you.
If yours is a newer account and you don’t have that search term data, should you even be doing Performance Max at this stage? No. But if you really want to, you have to do the keyword research, put them in a custom segment, add competitor names and URLs, and just keep it relatively simple.
So that would be one or two audiences and it’s really enough.
This is not hearsay or best practices pulled from my backside. I’ve seen this work in real accounts with real ad spend.
Can you run with no audiences? I wouldn’t recommend it, but you can. Technically, you don’t need an audience. They’re used as signals, so they steer Google. “Can you try going in this direction first?”
But Google quickly moves beyond those signals, and you will see this in the reporting tab under the insights area of your asset groups. Inside the campaign, it shows you the different audiences that are triggering, and you’ll see quite quickly that those are not audiences that you necessarily added.
How do I use Performance Max for new customer acquisition?
As with Smart Shopping, you can run new customer acquisition bidding strategies for both lead gen and ecommerce. You can use this feature to instruct Performance Max to go after “new” customers only.
Remember, the tracking isn’t perfect. What Google sees as a new customer might not actually be “new” e.g. their cookies expired. Don’t expect 100% accuracy, but this bidding strategy is heavily biassed towards new customers and new users for your website.
For ecommerce, lifetime values are typically higher, particularly if you have a rate of repeat customers.
For lead generation, it’s less important. Mid-size to larger accounts will certainly benefit from testing. If you’re a local plumber servicing the neighbourhood, it’s probably not that important (and you probably won’t want to be using Performance Max anyway).
New customer campaigns have to be separate because it’s a separate bidding strategy. You should expect the ROAS to be lower for ecommerce and cost per acquisition (CPA) to be a bit higher for lead generation.
So what about that first option? It’s essentially bidding higher with additional value, which you have to see in your reporting and add manually. So if you had a conversion that was worth £100, whatever you add here will be added to your conversion value metrics.
This inflates your ROAS, but it also has a requirement: You’ve got to upload a list of at least 1,000 active members for any given network. That’s why it’s worth testing the adjusted, recommended, and pure new customer bidding.
Reporting is pretty basic, quite frankly. Under your conversion segments, there’s “new versus returning” for a look at how Google is tracking these different customers. Be prepared to see quite a discrepancy. You might be as low as 20-30% of new customers being tracked and a large chunk of them unknown, but it’s still worth testing. At the very least, you can monitor incremental gains.
Just be sure to keep updating the customer list you upload to keep the data fresh as new customers come in. Once every week or two is ideal, depending on the size of your business.
Finally, there’s your unspent budget. A higher daily budget gives Performance Max some wiggle room to expand. This is particularly important for ecommerce where you might have a bit of flux and seasonality, where you want some headroom to spend more.
I’d also avoid being budget limited, unless you’ve got a very good reason for it. When you review your spend, if it’s 80% of your current budget, see that excess budget as some room for Performance Max to expand into. Don’t go overboard, because Performance Max is on Display and there can be some sharp, shocking spends on that network if you’re not too careful.
How does Performance Max affect other Google Ads campaigns?
It’s a mixed bag. Generally speaking, Performance Max is given priority about half the time. One good thing is that keywords still hold their own. It’s nice to see they’re not dead (yet).
- Search: Exact Match still rules the roost, and that will de facto take over. Broad and Phrase Match are a little bit different; they’re based on ad rank. So if Performance Max outranks your Broad and Phrase matches, it will appear instead.
- Standard Shopping: Performance Max will eat Standard Shopping, sadly.
- YouTube In-stream: Again, it’s based on ad rank.
- Display: Performance Max takes over by default.
- Discovery: Also based on ad rank.
- Static vs. Dynamic Product Ads: Some advertisers don’t know about this. Google tailors dynamic remarketing ads to people who visit your website. What they can do with the assets you provide is show a category-level image or advert, or show specific products that people browse. This depends on cookie tracking and buyer behaviour.
- Remarketing: It’s not true remarketing in Performance Max. Your audiences are nothing but signals, and as we discussed Google quickly expands past them. That means you can’t do dedicated, targeted remarketing with Performance Max.
Final checks before optimising new Performance Max campaigns
Minimum ad spend
- Lead generation accounts should be spending much more than a couple of thousand dollars/pounds/euros a month.
- For ecommerce accounts in a similar range or less, expect a lot of choppiness with Performance Max. You’ve pretty much got to use it unless you want to heavily (and manually) optimise Standard Shopping.
- For B2B specifically, I’d typically avoid it altogether. If you’re tapped out on all networks, you can’t spend more, you can’t get more convergence, you’re just struggling to expand, but you’ve got your conversion stages and qualification in place—then and only then would I consider testing Performance Max.
Go to your location options and look for where you’re opting people in from. The one recommended by Google is probably not ideal for most people, because it pulls in people from all over the place. This is a hidden setting, so remember to adjust it.
Final URL expansion: on or off?
Google recommends you leave this on, but not many people realise that means Google can create entirely dynamic (read: less than ideal) ad content. This acts a bit like dynamic search ads in that it’s pulled from the page’s content.
If you’re absolutely not sure, turn it off to start with and you can always turn it back on. As a bare minimum, exclude your less important pages and ones that simply don’t convert.
If you use final URL expansion, also include the new page feed to improve matching. Don’t be lazy and skip this. But also know that Google can still expand beyond your page feed. Still, it’s better than not having one at all.
Don’t fiddle with changes
Once you’ve created a new campaign, avoid big changes; keep bidding and budget tweaks to a minimum.
If you’ve got less than 30 conversions per month, start with a higher CPA target. If you’ve got 30-50 conversions per month, start with a lower ROAS target. It’s not ideal, but it gives the bidding engine a bit more room to pull in conversions and get that data in. Then you can adjust your CPA down or ROAS up.
You’ll want a few weeks of data or at least 30 conversions before raising bids. The exact window depends on the volume of impressions, cost, and conversions you get. But expect the first couple of weeks to be a bit choppy; even the first month. If you’re really lucky, the first week is choppy and then it levels out quite quickly.
Upload all your assets
Add the minimum number of headlines and descriptions, your logo, videos if you’ve got them, images, and tailored extensions. Anyone can write five decent headlines. It’s not like responsive search ads where you write 15 headlines that don’t really make sense together or aren’t unique, but five is doable.
If you need inspiration, here’s a freebie. This will scrape the search results for you, and there’s also free competitor reconnaissance training. You’ll get a free report, template, and training. I can’t say more than that. Review your competitors’ offers and positioning, and write better adverts.
If you need inspiration for your imagery and video, Google has developed the Creative Studio. It’s okay; if you’re really in a rut, you can use some of their templates. They’re not awful. The video templates are certainly better than the automated things that Performance Max puts together. Give that a whirl; it’s completely free.
Exclude unsuitable content
Need to exclude content keywords across the account? This works on Display and YouTube. Stop your ads from showing on sensitive topic content, parked domains, YouTube live streams, and so on. By default, I would exclude live streaming and parked domains. They’re just wasteful.
Respect the learning period
You’ve created your campaign, set your budgets and bidding strategy, uploaded your assets, and gone live.
It’s not time to optimise yet—this is the patience game. Google recommends six weeks of time and ad spend to allow the machine to learn, which is easy to recommend when it’s not their money.
Is six weeks accurate? You can’t predict this, so don’t try.
It can be half that (or more), but let’s be cynical and assume it is six weeks. There’s no point setting up a Performance Max campaign and one week in saying, “Oh, it’s not converting. It’s spending too much money. Pause it, scrap it, game over.” You’ll just be wasting your time.
Set aside a budget that you can comfortably live with for those six weeks, even if the campaign doesn’t perform as well as you’d hoped. Obviously, if you’re in week two or three and it’s absolutely going tits up, overspending with zero conversions, nuke it and rethink it. But patience is super important.
You can tweak it, but the more you touch it in the early days, the worse it will perform. You can also artificially delay gratification by overcomplicating your setup with too little budget. If you’ve only got $50-100 to spend a day, rein it in. You want a single campaign with an asset group or two. For ecommerce, that should be a specific category that you know sells and will get volume through it.
There’s no point trying to stretch a tiny budget across 50 things. You won’t accumulate enough data to make an informed decision in six months, let alone six weeks.
Next steps: Follow my Performance Max review timeline
I know I said not to predict the timeline, but I can’t let you go empty-handed. So here’s my process to review new Performance Max campaigns.
You go live in week 1, and weeks 2-4 are for your learning and ramp-up phase. Now you’re a month in.
At week 5, apply your conversion lag when you review this period. Ignore week one, but you want 3-4 weeks of data plus conversion lag. If you go to attribution and path metrics, look for your average data conversion to see your lag. Some people have a few days, some have weeks.
Let’s say your conversion lag is a week; a nice, round seven days. You review weeks 2-5 and then optimise. Don’t just review an arbitrary month, including that first week; exclude that first week of chop. If it’s not too messy, you might be able to include it (this is rare).
If you’re comparing Performance Max to another campaign, keep this chop and learning period in mind for a fair comparison. You might want to cut out that first week or two.