Performance Max is Google’s golden child. It’s the campaign type they recommend to almost everyone, and it’s the one most agencies default to because it’s relatively hands-off. Set it up, feed it some assets, let the algorithm do its thing.
For large ecommerce businesses with big budgets and thousands of monthly conversions, PMax can work well. The machine learning has enough data to learn from, enough volume to optimise against, and enough budget headroom to test and iterate.
But for the majority of ecommerce businesses we work with — spending between £1,000 and £50,000 a month on Google Ads — Performance Max is often doing more harm than good.
The data problem
PMax is an automated campaign type. It decides where your ads show, who sees them, and how much you pay per click. To make those decisions well, it needs conversion data. Lots of it.
Google’s own documentation suggests you need at least 30 conversions in a 30-day window for automated bidding strategies to work properly. Most small to mid-size ecommerce accounts don’t hit that threshold consistently. Some months they do, some months they don’t.
When the algorithm doesn’t have enough data, it guesses. And those guesses cost you money. You’ll see spend distributed across placements that make no sense — YouTube pre-rolls, Discovery ads, Gmail placements — with no clear connection to purchase intent. The campaign looks busy. The return tells a different story.
You can’t see where the money goes
This is the part that frustrates me most about PMax. The reporting is deliberately vague.
With a standard Shopping campaign, you can see exactly which search terms triggered your ads. You can see which products are getting clicks and which are converting. You can make decisions based on that data — add negatives, adjust bids, pause underperformers.
With PMax, you get a fraction of that visibility. Google shows you some search terms, but not all. You can’t see which placements are driving results and which are burning budget. You’re essentially handing Google your credit card and trusting them to spend it wisely.
And here’s the tension: Google makes money when you spend more. Their recommendations consistently push towards higher budgets and broader targeting. When the platform that profits from your spend is also the one deciding where it goes, with limited transparency, that should give you pause.
What we actually see in audits
When we audit ecommerce accounts running PMax as their primary campaign type, the pattern is consistent:
The headline ROAS looks acceptable because PMax is cannibalising branded search. People who already know your business, who were going to buy anyway, are being attributed to PMax. Strip out the branded traffic and the numbers look very different.
Product performance is invisible. In a standard Shopping setup, you can identify which products return 8x and which return 0.5x, then allocate budget accordingly. In PMax, everything gets blended together. Your best products subsidise your worst ones, and you can’t tell which is which.
There’s no negative keyword control at the campaign level. Google added some search term exclusions for PMax in 2023, but the control is still limited compared to standard campaigns. That means irrelevant searches slip through, and you’re paying for clicks that were never going to convert.
The change history is often bare. Because PMax is marketed as “automated,” some agencies treat it as a set-and-forget solution. The campaign runs, the money goes out, and nobody looks under the bonnet until the client starts asking questions.
When PMax actually makes sense
I’m not saying Performance Max is always wrong. There are situations where it works well:
You have a large product catalogue and enough budget to let the algorithm learn across hundreds or thousands of SKUs. You’re consistently hitting 50+ conversions a month at the campaign level. You have strong creative assets — good product images, video content, compelling ad copy — because PMax runs across every Google surface.
And critically, you’re running it alongside other campaign types, not instead of them. PMax should complement a well-structured account. It shouldn’t be the entire account.
What the fix looks like
For most of the ecommerce businesses we work with, the fix is straightforward: reduce reliance on PMax and build out a proper campaign structure.
That typically means standard Shopping campaigns segmented by product category or margin, where you can control bids at the product level. Search campaigns targeting high-intent keywords with proper negative keyword lists. And yes, potentially PMax for specific product groups where the data supports it — but as one piece of the strategy, not the whole thing.
It’s not glamorous. It’s not automated. It requires actual ongoing work, reviewing search terms, adjusting bids, testing ad copy, managing negatives. But that work is exactly what separates a well-managed account from one that’s just haemorrhaging budget quietly.
The accounts that perform best are the ones where someone is paying attention every week. Not checking in once a quarter. Not letting Google’s automation run unchecked. Actually managing the spend, product by product, keyword by keyword.
Put crap in, get crap out. The same applies to campaign management.
If your ecommerce account is running mostly on Performance Max and you’re not sure whether it’s actually working, we’ll show you. Our free audit breaks down exactly where your budget is going and what a better structure would look like. Book yours here.