Open the campaign, find the channel performance report, and read the split.
Search, Shopping, Display, YouTube, Discover, Gmail, Maps. That breakdown did not exist for most of this campaign type’s life, and it is now the first thing to look at, because for a lead generation business the answer is frequently that half the money went to Display and YouTube while the conversions were credited on Search.
That is the whole complaint about Performance Max, stated as a number instead of a feeling.
The campaign type was never broken. It was unreadable, so nobody could tell a working one from a failing one, and 2026 is the year that stopped being true. Three controls arrived that change what you should actually do on Monday morning.
What changed, including something I told you last year
Channel level reporting. Where the budget went, by surface. Use it to sanity check that spend is landing where intent lives for your business.
Full negative keyword support at the campaign level. This is the one worth correcting properly, because when I wrote about Google Ads budget leaks the position was that Performance Max did not accept campaign-level negative keyword lists and account level was the only reliable route. That is no longer the case. You can now add negatives at the campaign level, and campaign-level negative lists exist for managing the same exclusions across several campaigns.
If you built account-level lists as a workaround, they still work. You now have a finer instrument as well.
Brand exclusions as a proper feature. Set in the campaign settings, managed as brand lists at the account level. Select your brand and it catches the name plus known misspellings, alternate names and product line names, which is more than a negative keyword list will do by hand.
Google’s own guidance is to use brand exclusions rather than ordinary negatives when the goal is to keep the campaign off your own branded queries.
The brand problem, briefly
PMax launches, conversions jump, everybody is pleased, and a chunk of that jump is people typing your company name who were coming anyway.
I covered the diagnosis and the fix in the budget leaks post, so I will not repeat it here beyond the one line that matters: switch your brand off and watch the numbers for two weeks. If total conversions barely move while cost per acquisition improves, you were paying for people who already knew you.
The reason this matters more in Performance Max than elsewhere is credit. Branded traffic converts easily, so it makes PMax look like your best campaign, which gets it more budget, which starves the prospecting you actually needed.
Asset groups are the part people get wrong
One asset group with everything in it is the most common structural mistake I see, and it is worse than it looks.
The asset group is how you tell the system what a coherent offer looks like. Put four unrelated services in one group and you have told it that a $200 repair job and a $40,000 installation are the same thing. It will optimize toward whichever produces conversions fastest, which is always the cheap one.
One theme per asset group. One service line, one product category, one audience with a shared problem. Each with its own headlines, images, video and search themes.
If you sell things at very different prices, split them. Not because it is tidier, but because a single conversion goal across mixed price points will quietly send your budget to the cheapest outcome you count.
Two other things worth doing while you are in there. Give it video, even a poor one, because it will generate something from your assets otherwise and you will like it less. And write the search themes as things people search rather than as categories you use internally.
The black box eats whatever you feed it
Everything above is settings. This part decides the outcome.
Performance Max optimizes toward the conversion actions you marked as primary. If your primary conversions include newsletter signups, phone number clicks that nobody answered, and a contact form that gets 60% spam, then that is the definition of success you handed it, and it will pursue it competently.
Two checks.
Open your conversion actions and confirm exactly which ones are set to primary. This takes ninety seconds and it is the single highest leverage screen in the account.
Then, if the values of your outcomes vary, send values rather than counts. A campaign told that every conversion is worth 1 will chase volume. A campaign told the actual gross profit will chase money. For anything where order sizes vary a lot this is the difference between a campaign that works and one that reports well.
There is also a volume floor underneath all of this. Automated bidding needs roughly 30 conversions in a 30 day window before it performs reliably, so a Performance Max campaign generating six conversions a month is not being badly managed, it is being asked to do something it cannot do.
When a plain Search campaign beats it outright
Say this plainly, because the platform will not.
Low conversion volume. Below the learning threshold, a standard Search campaign with keywords you chose and bids you control will beat an automated system with nothing to learn from.
Strong existing brand demand. If most of your searches are already for your name, PMax will find that traffic and bill you for it. A brand Search campaign at a controlled bid does the same job cheaply and honestly.
A narrow, high value service where you know the exact queries. Twenty terms that matter, a large value per customer, and no need for discovery. Exact match Search, tight negatives, done.
When you need to know what happened. Even with the new reporting, Search gives you a cleaner read on which query produced which sale. If you are still building your understanding of a market, buy the clarity.
Performance Max earns its place when you have real conversion volume, genuine breadth of inventory or services, and you want reach across surfaces you would not buy individually. Ecommerce with a product feed is where it is strongest by a distance.
Before lunch
Four things, in this order, about forty minutes.
Open the channel report and write down the percentage of spend going to Display and YouTube. If you sell services and that number is above half, you have found your problem.
Check whether brand exclusions are on. If they are not, turn them on and diary a check in two weeks rather than judging it after three days.
Open conversion actions and remove anything from primary that is not a real business outcome.
Look at your asset groups and count how many distinct offers are crammed into each one. If the answer is more than one, that is next week’s job.
My notes on performance marketing cover where this sits in an account, and the cost per click arithmetic decides whether any of this is worth doing before you touch a setting.
If you’d like a second read on your own account structure, drop me a line on email, WhatsApp or LinkedIn and we can have a quick chat. I’m contracted full time so this isn’t a pitch.