ArticleMaximizing Performance Max
If you’re unfamiliar with Performance Max (“PMax”) it is (essentially) a Dr. Frankenstein’s monster of Google inventory, combining search, display, YouTube, shopping, discover and local placements into a single, fully-automated campaign type. The advantage is pretty simple: simplicity and ease-of-use. PMax can work brilliantly well for smaller advertisers, and (if properly structured and managed) can help larger advertisers unlock new audiences and scale.
The challenge (or the negative) is that simplicity cuts both ways: PMax removes many of the traditional control levers available to marketers – from granular keyword targeting, to specific audience targeting, to robust exclusions.
PMax is a campaign type designed and intended to be a “set-it-and-forget-it” – but after looking at thousands of these campaigns, I can safely say: that’s a recipe for disaster. Yes, PMax is more complex and nuanced to manage than other campaign types, but that doesn’t mean it can’t be managed or optimized; it just means that you need to think differently about how you approach accounts with PMax in them vs. those without.
While there are (literally) 1,000s of guides out there about PMax, I’ve found most of them are woefully deficient in four core areas: setup, structure, creative and management – so that’s what I’m going to cover in this week’s issue.
Account Setup:
More than any other campaign type, PMax requires exceptional conversion tracking to be in-place to be effective. For eCommerce brands, this means having both enhanced conversions and offline/re-stated conversions in place (such that if a user goes on to buy another product, or return all or part of a purchase, the value of the original sale is updated to reflect the now-current value). For lead generation brands, optimizing off of form submits with PMax tends to be a mistake (and lead to a deluge of spam/low-quality submissions); instead, focus on MQL/SQL type leads using (you guessed it) offline conversions.
Lack of a robust data feedback loop is the #1 reason why PMax campaigns over-spend and under-perform. If you don’t have this in place, don’t bother setting up PMax; instead, spend your time and energy getting your data into a good place.
Once you have solid data flowing into your account, the next step is to upload your existing customers/clients/leads. This is what enables the “Bid For New Customers” setting (available in all Google campaign types) – a key piece of the puzzle for keeping PMax focused on what it should be doing (acquiring you net-new customers) and away from what it should not (serving remarketing ads to your existing customers until they buy or die).
Make no mistake: data is THE optimization lever for Performance Max. If you don’t have it, it will run amok across your account (more on this later).
Structure
Most advertisers running Performance Max fall into one of three categories:
- PMax-Only – these tend to be exceedingly small advertisers (<10 SKUs or a local service business, for example). The entire Google Ads account consists of a single PMax campaign with one (maybe two) Asset Groups.
- PMax + Others – this is overwhelmingly the most common structure I see: there’s an existing account (either eCommerce or Lead Gen) with multiple other campaigns (search, shopping, YouTube) running. For many of these accounts, PMax was added to the existing mix. Many of these accounts have noticed significant declines in the performance of their other campaigns, while PMax appears to be performing quite well – often leading to discussions around sunsetting other campaigns and/or adding more PMax to the picture.
- Multiple PMax + Others – I see this most often in larger/more complex eCommerce accounts or multi-vertical lead gen. Accounts running multiple “legacy” campaign types (search, shopping, YouTube, display) along with multiple PMax campaigns tend to be larger, but they often see the same types of challenges as the PMax + Others: PMax appears to be performing better than everything else, but (unlike the above), scaling this up is much more difficult.
To be blunt: if you’re in Category 1 above (PMax only), you probably don’t need to be worried about structure at this stage. You’ll get more value from PMax by optimizing your creative + (potentially) adding a few other asset groups.
While there’s no one “ideal” structure, I’ve found that the following framework tends to work best across brands:
This structure operates on a few, foundational principles:
- Despite PMax being automated, it remains exceedingly advantageous to be able to control tCPA or tROAS at a more granular level than “everything we do/sell”
- We want to be able to prioritize how budget flows into PMax to avoid a suboptimal/lower-priority campaign out-spending a high-priority campaign
- There’s value in differentiated creative + distinct audience signals.
Obviously, if you have a more complex business (i.e. one with 10,000 SKUs across a dozen product categories, or 13 different services each with radically different customer values and profit margins), it may be necessary to expand on this further; for 95% of brands, the above is sufficiently robust.
In the above, Campaign #1 is the highest-priority. It should be given the most budget and the least constraints (i.e. highest tCPA or lowest tROAS) possible. Since the products/services being advertised by Campaign #1 are both your best sellers (in terms of volume) and most profitable (in terms of contribution margin), you can afford to be more aggressive.
Campaign #2 contains your next-best products/services. This might be (for instance) seasonal product sets with sky-high contribution margins, decent-selling services with solid (if unspectacular) contribution margins, and/or super-popular items/services with decent contribution margins. In order to prioritize this campaign below Campaign #1, you’ll either reduce the tCPA or increase the tROAS. Campaign #3 contains the rest of the stuff you’re advertising – super profitable niche items, loss leaders, etc. And, just as you did with Campaign #2, you’ll increase the tROAS or reduce the tCPA in order to ensure that it is prioritized below both Campaign #1 and Campaign #2.
Equally important is how you handle the asset groups – there’s a temptation to throw as much stuff into each asset group target as you can (your data, user signals, search terms, interests/behaviors). Resist that urge.
Instead, think carefully about each Asset Group’s audience: Who are they? What do they want? What pain points do they have? Why are they coming to your brand vs. another? What is compelling them to act now? (Yes, this is why audience research is absolutely essential).
Once you have clarity around the audience, judiciously add both your data (i.e. a segmented list of purchasers of just that item/service) and only the most relevant search terms or interests. Less is more. Google’s objective is to be able to serve as many ads to as many people as possible, which is why they push for your Audience Signals to be as broad as possible; your goal should be to serve the right ads to the right audience as much as possible.
One of the most common issues I see in Audience Signals is the same signals or data points being used across many different asset groups – this just muddies the waters and prevents both you and Google from gaining a clean, clear signal:
- Audience Lists – always use the most specific and relevant audience possible, not simply an “all site visitors” or “all customers” list (unless you don’t have enough customers to create multiple, segmented lists).
- Search Terms – focus on the most specific keywords or search terms possible. Avoid branded keywords. Avoid competitor names (unless you’re running a competitor audience signal).
- Interests + Behaviors – I typically avoid “Affinity” audiences like the plague: they’re exceedingly broad, and often aren’t reflective of true, in-market intentions. As with above, pick only the 1-5 best and most relevant. Google may complain that your audience isn’t robust enough, but remember: your goals + Google’s goals aren’t the same.
Exclusions
Finally, we come to the most important part of your PMax Structure: the exclusions. At present, for most accounts, there are only four ways to exclude things from a PMax Campaign:
- An Account-Level Exclusion (Manual or Google-Provided) – PMax will honor account-level negative KWs, as well as account-level placement exclusions.
- A Negative KW List Applied to the PMax Campaign by a Google Rep – If you want to specifically exclude specific topics, placements, keywords, etc. from either your account or a specific PMax campaign, there’s a “Performance Max Campaign Modification Request Form” available here.
- A Brand Exclusion – Finally, there are brand exclusions. These can be added in the “Campaign Settings” section of the PMax campaign, under “show more” (yeah, Google doesn’t make it easy). According to Google, these should exclude all common misspellings of your brand, but that’s hit-or-miss. There is ZERO reason to include your brand in your PMax campaigns – at best, these searches being included will artificially inflate the impact of PMax while siphoning these queries away from better-converting, lower-cost branded campaigns. At worst, this will result in lower-quality ads appearing for your branded searches, potentially costing you in the form of higher CPCs + fewer conversions. Keep PMax away from your brand.
My preferred way to handle exclusions for Performance Max is threefold: (1) always add your brand exclusion lists. (2) Use the Campaign Modification Request Form to have Google add a “PMax Negatives” list to your PMax campaign. This will allow you to continually update your exclusions based on the search terms + insights you see in PMax reporting. (3) Use the Placement Exclusions to cut out all app traffic (it’s overwhelmingly garbage), as well as spammy, low-quality and/or poorly-performing placements.
Aside – if you’re not sure how to exclude placements or apps. Kirk Williams at Zato Marketing has put together step-by-step guides on how to view your PMax placements, Exclude Placements from PMax and Exclude Apps from PMax. If you’re looking to get even more insight into PMax performance (helps with management), check out this incredible script from Mike Rhoades. This provides insight into: (a) inventory types (what is your breakdown between search, display, shopping, video); (b) which search categories (basically search queries) your ads are showing for and (c) a “product matrix” that assigns all products (if eCommerce) into one of 6 buckets based on performance.
Creative
The consensus for PMax Creative has always been to add as many assets as possible into each Asset Group. This has resulted in many, many accounts re-using the same, generic creatives across multiple asset groups.
My recommendation: if you are going to invest in creating an asset group, then invest in creating a set of assets tailor-made for that audience signal and unique to that asset group. This is no different than how I typically recommend approaching Meta: the creative is part of your targeting. If you have specific targeting but generic creatives, you’ll end up with a semi-specific audience that performs moderately well. If, instead, you have a specific audience signal AND hyper-specific creative that speaks directly to that target audience, you’ll have a hyper-specific audience that performs exceptionally well.
Invest the extra effort to tailor your assets – headlines, images, long descriptions, videos, etc. – to your Audience Signal.
Management + Optimization
Finally, let’s get into management + optimization of Performance Max. To get the obvious out of the way, the options available to you (the advertiser) to optimize PMax are radically different from the options available in other campaign types. That does NOT mean you can’t optimize it; it means that you need to think differently about how you go about it.
Keyword + Query Management
One of the biggest issues that I see in accounts running PMax is that PMax cannibalizes existing, high-performing campaigns even when it does not appear that it should. Google has published the following help information on how Search + Performance Max work together (you can view it here)
In theory, this sounds good – as long as you have an Exact Match KW in an active search campaign, you’re good to go. But that’s not exactly what this says:
- There’s a difference between “identical” and “exact” – in order for the EM search term to be given priority, it must be IDENTICAL (no plural, no other words, no misspellings)
- If the EM KW has “low search volume” status, it is considered “ineligible” (even if it actually serves) and will default to PMax
- If your search campaign uses Broad Match targeting (you know, like Google suggests), PMax will be given priority a disproportionate amount due to the “creative relevance” factor; the “performance” aspect of this is almost entirely disregarded (gotta love it).
The data I have mirrors what Brad Geddes found when reviewing hundreds of millions in spending across Adalysis’s account portfolio: when PMax and Search show for the same query, Search campaigns tend to have a higher CTR and CVR, and a lower CPC – yet PMax serves more.
So, as an advertiser, this is where you want to add more exact match KWs to your account. Reviewing your Search Terms Report (and the Search Themes from PMax) is absolutely critical. Similarly, when Google pops up that “Remove Redundant Keywords” recommendation, dismiss it with prejudice – remember: in order for the search campaigns to be given priority over PMax, the search term must IDENTICALLY match the EM keyword. Removing EM keywords reduces the probability of that happening, which INCREASES the probability of PMax serving.
The same is true for negative KWs – you want to relentlessly kick out the crap – bad topics, bad placements, apps, etc. This is why I recommend having Google Add multiple negative KW lists to your PMax campaign: (1) a generic “negatives” list and (2) a negatives list that covers other products/services/lines of business you want to sculpt out of a given PMax campaign. In theory, this should get easier in the coming months, as Google has announced that they are bringing negative keywords to PMax (and I’ve seen it in Beta in some accounts).
In a “mixed” account (i.e. one running PMax alongside one or more other campaign types), the goal of PMax should be one of audience expansion + keyword/search theme discovery (essentially functioning like a hybrid of smart shopping and old-school DSAs). As you identify converting keywords/search themes, “promote” them (via EM KW additions and/or new ad groups) into the relevant evergreen campaign, then (if possible) exclude them from PMax.
This forces PMax to continually hunt for new keywords/audiences/themes, pushing it far more into the niche term + emerging concept discovery vs. less efficient conversion cannibalization from known terms. The more you can push PMax to be a discovery engine, the healthier your account is likely to be; the more PMax is allowed to simply harvest conversions that would have otherwise been captured by your business-as-usual campaigns, the more your account will rely on PMax and be more difficult to scale. This applies to both lead gen and eCommerce.
Note: if you’re not sure how to access this information, here’s a step-by-step walkthrough from Fred Vallaeys at Optmyzr.
URL Expansion + URL Feeds
One other feature of PMax that I’ve seen run haywire of late is the “URL Expansion” – this (essentially) lets PMax ignore your ad destination/landing page in favor of…whatever it wants. This can have all sorts of adverse effects on performance (especially if Google decides to send people to an irrelevant or odd LP) that might be quite difficult to parse.
To exclude URLs from PMax, follow these instructions. My standard recommendation – if you’ve followed the other advice in this article – is to turn OFF automatic URL expansion. This will force PMax to send your traffic to the page(s) you designate within your campaign / asset group.
In some cases, you might want to give PMax a bit of rope, but not all of the rope – for instance, you might want to allow PMax to choose from several different landers, or to send people to either a collection page or a product page from that collection (but no others). In this case, a URL Feed is the way to go. 99% of advertisers don’t have these in place, but would benefit from them. If you’d like to know how to set them up, check out this resource.
If set up, structured and managed properly, PMax can be a remarkably powerful campaign type that effectively supports the rest of your account, uncovers new search trends and helps you unlock new audiences. The problem is that the PMax deck is stacked against advertisers – it isn’t easy to make PMax work, and the instructions/recommendations Google provides are often diametrically opposed to the advertiser’s best interests. So while I fully understand and appreciate how frustrating it can be, I hope some of these suggestions/ideas are helpful as you try to make PMax work in your account.
Cheers,
Sam
