Google Ads’ Performance Max was supposed to make paid search easier. One campaign. All of Google’s inventory. Machine learning doing the heavy lifting. For a founder already stretched across operations, team, and product, it sounded like exactly what was needed.
I, on the other hand, was a huge sceptic when Google first rolled out the ‘new’ way of working… I still was until mid-2025. A big believer in visibility and control of data, I saw pMax as a way for Google to build walls around the key insights my customers needed to truly optimise based on real customer data. It didn’t sit well with me. But… I’m now all onboard the pMax train. Let me explain to you why.
Brands launched their PMax campaigns. They connected their product feed. They uploaded some creative assets. They set a target ROAS and let Google get on with it.
Six months later, the campaign is spending. The ROAS looks acceptable. But growth has stalled. New customer acquisition feels flat. Brand search is consuming budget that should be working harder elsewhere. And nobody really knows what the machine is actually doing.
PMax doesn’t fail loudly. It quietly underperforms while looking just convincing enough that you don’t stop it.
That’s the trap most brands fall into. Not a catastrophic mistake. Just a slow leak of potential that compounds over time. And it riled us Google Ads veterans.
Why Is Handing Decision-Making to Pmax A Flawed Strategy?
The fundamental misunderstanding most brands have with PMax is this: they believe handing it over to Google is the strategy.
It isn’t. It’s an abdication of strategy.
Performance Max is an execution engine. It needs signal, structure, and intent from you before it can do anything meaningful. Without those inputs, it defaults to the path of least resistance, which is usually harvesting existing demand rather than generating new demand.
Let’s deep dive what this looks like in practice. PMax, when under-configured, gravitates toward branded keywords, retargeting audiences, and shopping placements where conversion probability is already high. These are the placements where you were almost certainly going to convert anyway. You’re paying Google to mop up demand that was already yours.
| The question most brands ask: ‘Is our PMax campaign performing?’ The question they should be asking: ‘What is our PMax campaign actually doing, and is it doing what we need it to do?’ |
Those are fundamentally different questions. The first invites you to look at aggregate ROAS and feel reassured. The second forces you to open the machine and inspect its behaviour.
What’s Really Happening Inside an Under‑Configured PMax Campaign?
When a brand launches PMax without proper configuration, Google’s algorithm does what any profit-maximising machine does: it finds the easiest conversions available.
Brand Cannibalism: Is PMax Bidding on Your Own Brand Terms and Stealing Credit?
The most common issue I see is brand cannibalism. PMax runs across all Google channels, which includes Search. Without a brand exclusion strategy in place, PMax will bid on your own brand terms and take credit for conversions from customers who were already searching for you by name.
You haven’t acquired a new customer. You’ve just paid Google a commission on a sale you were going to make anyway. Your reported ROAS looks great. Your actual new customer acquisition is doing nothing. And for large brands… these costs add up (significantly).
Weak Asset Groups: Are Generic Creatives Limiting PMax Performance Across Channels?
Asset groups are how PMax serves creative across channels including Display, YouTube, Gmail, and Discover. Most brands upload a handful of product images, a logo, a few headlines, and call it done.
The machine then assembles these assets into ad combinations and tests them at scale. But if your assets all say the same thing, feature the same product angles, or speak to the same customer moment, the machine has very little to work with. It isn’t creative. It’s combinatorial. Garbage in, generic ads out.
Strong asset groups are built around specific customer intent. Not ‘here’s our product.’ But ‘here’s what changes for someone like you when you buy this product.’ That means understanding the jobs your product is being hired to do, and building creative that reflects those distinct jobs.
| The Optimisation Principle Your PMax campaign is only as intelligent as the inputs you give it. Asset quality, audience signals, feed data, and conversion event setup are the variables inside your control. The machine learns from what you teach it. |
Poor Product Feed Quality: Are Unoptimised Titles and Labels Crippling PMax Results?
PMax draws heavily from your product feed for Shopping placements. If your feed titles are pulled directly from your product database with no optimisation, if your descriptions are manufacturer copy, if your custom labels are empty, then the machine is operating with imprecise data.
Every product in your feed is a potential placement. The feed is where the machine decides which products to show, to whom, and at what position. A weak feed is a weak campaign, regardless of how much budget you’re putting behind it.
Conversion Event Quality: Are You Optimising PMax for Micro‑Conversions Instead of Revenue?
This one is foundational and often overlooked. PMax optimises toward the conversion events you tell it to optimise toward. If you’re including micro-conversions like add-to-cart or email sign-up in your conversion tracking alongside purchases, the machine may be optimising toward signals that don’t reflect actual revenue.
The result is a campaign that looks busy and conversion-rich, but that has drifted away from the actions that actually matter. Tighten your conversion event setup to prioritise purchase events and, where your data allows it, high-value purchases.
The machine is doing exactly what you told it to do. The problem is most brands never clearly told it what they actually needed.
The Reframe: Start Treating PMax as a Configurable System, Not a Set‑and‑Forget Campaign
Stop thinking of PMax as a campaign you launch and leave. Start thinking of it as a system you configure, feed, and continuously refine.
The brands I’m working with that are getting the most from Performance Max are the ones that treat it as an active component of their growth machine rather than a passive budget allocation. They’re making regular decisions about what signals to give Google, what products to prioritise, what audiences to exclude, and what creative to test.
Separating Brand Search From PMax: Should You Run a Dedicated Brand Campaign and Apply Brand Exclusions?
Run a dedicated brand campaign in standard Search alongside PMax. Use campaign-level brand exclusions in PMax to prevent it from harvesting your branded traffic. Yes, Google makes this more complicated than it should be. Do it anyway.
This one structural change often reveals the true performance of PMax almost immediately. When brand terms are removed from the equation, you get a far clearer picture of what the campaign is actually achieving in terms of new demand generation.
Start Building Asset Groups Around Customer Intent and Jobs‑to‑Be‑Done
Look at your customer research. Look at your post-purchase survey data, your reviews, your support tickets. What are the real reasons people buy your product? What problem are they solving? What change are they hoping to make?
Build separate asset groups for distinct customer jobs. A brand selling premium dog food might have one asset group for ‘transitioning from mainstream brands,’ another for ‘dog with dietary sensitivities,’ and another for ‘new puppy owner.’ Same product. Very different conversations.
| Customer insight is the highest-leverage input into your PMax campaign. The better you understand why people buy, the better you can configure the machine to find more of them. |
Use Audience Signals Proactively: Customer Lists, CRM Data, and High‑Intent Custom Audiences
PMax allows you to add audience signals: hints to help the algorithm find relevant people faster. Upload your customer lists. Create high-intent custom audiences based on search behaviour. Layer in data from your CRM.
These aren’t targeting constraints. They’re signals. The machine will use them as starting points and then expand from there. But starting with strong signals dramatically shortens the learning phase and improves early campaign performance.
Clean and Enrich Your Product Feed to Strengthen Shopping and PMax Performance
Conduct a proper feed audit. And I mean proper (not a ‘free’ add-on from an agency hunting down your account management). Optimise your title structure to lead with the attributes customers actually search for. Write descriptions that reflect buyer language, not product spec. Use custom labels to segment your feed by margin, bestseller status, or strategic priority, and use those labels to control where your budget concentrates.
A well-structured feed is a compounding asset. The improvements you make today will influence Shopping performance, PMax performance, and any future formats Google introduces that draw from feed data.
The Core Principle: PMax Rewards Better Inputs, Not Bigger Budgets
PMax is a powerful tool that rewards operators who understand how to configure it. It’s a dangerous tool for those who treat it as a set-and-forget solution.
The brands losing ground with PMax aren’t doing anything dramatically wrong. They’re just doing the minimum. They’ve handed the machine a blunt set of inputs and expected sharp outputs. That’s not how it works.
Optimisation here isn’t about finding the perfect bid strategy or the magic budget split. It’s about treating each input into the campaign, the feed, the assets, the signals, the conversion events, as a lever. Every lever you improve makes the whole system more effective.
And that compounds. A cleaner feed improves Shopping placements. Better asset groups improve Display and YouTube performance. Stronger audience signals shorten learning phases. Tighter conversion tracking means the machine is optimising toward what actually matters.
Three modest improvements to your PMax inputs will outperform a doubled budget every time. Better signal beats more spend.
The growth machine doesn’t need more fuel. It needs to be better configured to use what it already has.
| PMAX DIAGNOSTIC QUESTIONS • Have you applied brand exclusions to your PMax campaign? • Do your asset groups reflect distinct customer jobs, or are they all variations of the same product message? • What conversion events is PMax optimising toward, and are they your highest-value signals? • When did you last audit your product feed for title quality, description language, and custom label structure? • Are you using customer lists and CRM data as audience signals within the campaign? • Can you identify what percentage of your PMax conversions are coming from branded vs non-branded activity? |
Written By:
