Performance Max management that steers the black box instead of feeding it.
Performance Max hands Google one budget and near-total control of where it goes: Search, Shopping, YouTube, Display, Gmail, Discover and Maps. Left alone it will quietly spend on brand terms and cheap placements, then take the credit. We manage it as something to be steered, with clean feeds, deliberate audience signals, hard exclusions and conversion data worth optimising to, all judged against contribution margin.
Performance Max management is the work of controlling Google's most automated campaign type, which serves a single budget across Search, Shopping, YouTube, Display, Gmail, Discover and Maps with very little native reporting. The campaign itself is easy to launch and hard to run well, because the levers that matter are indirect: the quality of your product feed, the conversion data and values you feed the bidding, the audience signals you seed it with, and the brand and placement exclusions you insist on. We manage all of that, use scripts to see the channel and search-term detail Google hides, run it alongside dedicated Search rather than letting it swallow both, and measure it on margin rather than the flattering blended return the platform reports back to itself.
Performance Max, in full.
Performance Max is Google's answer to a world it wants fully automated: one campaign, one budget, and machine control over which of Google's surfaces your ads appear on. In exchange for that control you get reach across Search, Shopping, YouTube, Display, Gmail, Discover and Maps from a single setup, and for the right advertiser that breadth is genuinely useful. The catch is that the same automation that makes it powerful also makes it opaque. Native reporting tells you very little about where the money went, what search terms triggered it, or which placements it bought, and Google is not in a hurry to tell you.
That opacity is the whole management challenge. A Performance Max campaign is trivial to launch and easy to make look successful, because it will gravitate towards the cheapest conversions it can find, which are very often your own brand searches and people already in a remarketing pool. It will then report a healthy blended return built substantially on demand you already owned. The job of a Performance Max manager is not to tweak bids, because you cannot, but to constrain and feed the system so its incentives line up with yours: to make it go and find new, profitable demand instead of quietly harvesting the demand that was coming anyway.
Doing that well means working on the inputs the campaign actually responds to. A clean, complete product feed for retail; conversion tracking that reports real values rather than raw counts; offline conversion import so a lead that closes weeks later is scored correctly; audience signals that point the algorithm at the right starting population; and a discipline of exclusions, brand settings, account-level negatives and placement housekeeping that stops the budget leaking into rubbish. On top of that we run scripts that surface the channel split, the search-term categories and the placements Google's own interface will not show you, so decisions are made on evidence rather than faith.
We run Performance Max as one instrument in a paid programme that answers to contribution margin, not as a set-and-forget box that promises to do everything. The four of us have spent roughly forty years between us on paid and analytics, and there are no juniors on your account. On a channel this automated that is the point: the automation will happily optimise to the wrong goal at speed, and it takes someone senior to notice, prove it, and change the inputs so it optimises to yours instead.
Performance Max, broken down.
Product feed quality and structure
For retail, the feed is the campaign. Titles, attributes, images, GTINs and custom labels decide what PMax can sell and how profitably. We audit and restructure the feed, segment by margin with custom labels, and fix the errors that silently exclude products, because no amount of bidding cleverness rescues a poor feed.
Explore →Conversion values and offline import
Automated bidding optimises to whatever you tell it a conversion is worth. We set real values, deduplicate goals, and import offline conversions so a sale that closes later, or a lead that turns into revenue, is scored properly. Feed it counts and it chases volume; feed it values and it chases margin.
Audience signals and asset groups
Audience signals do not restrict PMax, they seed it, and the seeding matters. We build asset groups around real product or service themes, attach considered signals from your first-party data and intent audiences, and give each group creative worth serving rather than one generic pool spread across everything.
Brand exclusions and account-level negatives
The single biggest source of overstated PMax performance is uncontrolled brand traffic. We apply brand exclusions, account-level negative keywords and campaign-level negatives so the campaign is not paid to intercept people already searching for you, and so its reported return reflects incremental demand.
Channel and placement visibility via scripts
Google's interface hides the channel split, search-term categories and placement detail that tell you what PMax is really doing. We run scripts and reporting that surface it, so we can see whether spend is going to genuine prospecting or to cheap Display and video impressions that pad the numbers.
Search cannibalisation and campaign structure
Performance Max and standard Search campaigns compete for the same auctions, and PMax often wins the easy, high-intent ones. We structure the account so dedicated Search keeps the queries it should own, and PMax is pushed towards the discovery it is actually good at.
Explore →The way we run Performance Max.
Model the economics and the goal
We start from what a customer or order is worth and what you can afford to pay to win one, then define the conversion goal and values PMax will optimise to. If the goal is wrong, the automation will pursue the wrong thing efficiently, so this comes first.
Fix the feed and the tracking
Before scale, we get the foundations right: a clean, well-structured product feed for retail, accurate conversion tracking with real values, offline conversion import where sales close later, and Consent Mode so collection is lawful and complete. PMax built on weak data optimises to a lie.
Build asset groups and signals
We structure asset groups around genuine themes rather than one catch-all, write and supply strong assets, and seed each with considered audience signals from your first-party data and intent audiences. The aim is to point the algorithm at the right starting population, not to hope it guesses.
Set the guardrails
Brand exclusions, account-level and campaign negatives, placement and content exclusions, and sensible campaign structure go in before spend scales, so the budget cannot quietly drain into brand harvesting, junk placements or queries a dedicated Search campaign should own.
Launch, watch the black box, and read the scripts
Campaigns go live with script-based reporting from day one, so we can see the channel split, search-term categories and placements Google hides. We watch how the automation behaves in its learning period and intervene on the inputs, since there are no bids to touch.
Test incrementally and cut against margin
Where budget justifies it we run experiments and brand holdouts to separate demand PMax created from demand it intercepted, then act on margin: shift values, refine signals, tighten exclusions and rebalance against Search. Weekly, honestly, not once a quarter.
How we approach Performance Max.
Control the inputs, not the placements
You cannot hand-pick where Performance Max shows, so the campaign is won or lost on its inputs. We put the work into the feed, the conversion values, the audience signals and the exclusions, because those are the only real levers. A well-fed PMax with disciplined signals beats hours spent wishing for placement controls that do not exist.
Feed it truth, not just conversions
Value-based bidding is only as good as the data behind it. We send back real conversion values, offline conversions where the sale closes later, and new-customer signals, so the algorithm optimises to margin and genuine acquisition rather than counting every cheap lead the same as a booked order.
Stop it eating your brand
Left to its own devices, PMax leans on brand search and remarketing because they convert cheaply, then reports a glowing return. We use brand exclusions, account-level negatives and holdouts to separate demand it created from demand it simply intercepted, so the number you see is the number that is true.
Senior hands on an opaque channel
PMax is the campaign type where a tidy dashboard hides the most, so it is exactly where junior autopilot costs you. One of four senior practitioners runs your account, reads the script-level detail Google omits, and tells you plainly what the automation is actually doing with your money.
What Performance Max puts on your desk.
Who runs your Performance Max, and why we refuse to take it on faith
Performance Max is the Google campaign type most designed to be trusted and least willing to show its working. That combination is exactly why it needs senior judgement rather than autopilot. Your account is run by one of four experienced practitioners who has managed PMax through its awkward early versions and its steady improvements, knows where it still overstates itself, and reads the script-level detail Google leaves out. We would rather show you a smaller, true number, built on incremental demand, than the larger blended figure the platform prefers to report.
Senior-led, no percentage on spend
The person building your feed, values and signals is the person you meet. Our fee does not rise just because Google spends more of your budget, so there is no incentive to let the automation scale past the point where the margin still works.
Measured on margin and incrementality
We use brand exclusions, experiments and holdouts to separate demand PMax created from demand it intercepted. If the campaign is mostly harvesting your own brand searches, we will show you the evidence and change the inputs, not defend the dashboard.
Honest about a black box
We will tell you plainly what Performance Max does and does not let you control, and where its reporting flatters itself. No pretending a script is a placement lever, and no promising a blended return we both know is inflated by demand you already owned.
Performance Max: common questions.
What does Performance Max management actually involve?
Performance Max management is mostly the work you cannot see in the campaign settings. Because you cannot pick placements or set keyword bids, the real levers are the product feed, the conversion values and offline data you feed the bidding, the audience signals seeding each asset group, and the exclusions that stop the budget leaking. On top of that we run scripts to reveal the channel, search-term and placement detail Google hides, keep PMax from cannibalising your Search campaigns, and measure the whole thing against contribution margin rather than the blended return the platform reports.
How is Performance Max different from a standard Search or Shopping campaign?
A standard Search or Shopping campaign gives you direct control over keywords, bids and, to a degree, placements, and clear reporting on where spend went. Performance Max trades almost all of that control for reach: one budget spread by Google's algorithm across Search, Shopping, YouTube, Display, Gmail, Discover and Maps, with far less visibility. It can find demand the others miss, but it needs much tighter management of its inputs and exclusions, because there are no bids or keywords to adjust when it goes wrong.
Does Performance Max just cannibalise my brand and Search campaigns?
It will if you let it. Left uncontrolled, PMax gravitates towards cheap, high-converting traffic, which is very often your own brand searches and existing remarketing audiences, then reports a strong return built on demand you already owned. The fix is deliberate: brand exclusions, account-level negatives, sensible campaign structure and holdout testing to prove what is genuinely incremental. Managed properly it complements dedicated Search rather than eating it, but that separation has to be engineered, not assumed.
How much does Performance Max management cost?
There is no single figure, because it depends on your account, whether you are retail or lead generation, and how much feed and tracking work the foundations need. Media is paid directly to Google with no percentage markup from us, and our management fee does not rise as your budget does. Senior-led management starts from £1,500 per month, and we scope honestly once we have seen the account and what it actually needs.
Can you see where Performance Max is spending my money?
More than the standard interface shows you. Google deliberately limits native reporting on channel split, search terms and placements within PMax, but scripts and the reporting we layer on top surface a great deal of it: which channels are taking spend, the categories of search terms triggering ads, and the placements running on Display and video. That visibility is central to how we manage it, because without it you are optimising a black box on faith.
Is Performance Max worth it for lead generation, or only ecommerce?
It can work for both, but the requirements differ. Ecommerce gets the most from PMax because a strong product feed gives the algorithm rich material to work with. For lead generation, the make-or-break factor is conversion quality: you must feed back real values and import offline conversions, so the automation optimises for leads that become revenue rather than whichever form-fill is cheapest. Without that, PMax will happily deliver a pile of low-quality leads and a flattering cost per conversion.
What are audience signals, and do they control who sees my ads?
Audience signals are suggestions, not restrictions. Unlike a targeted audience on Search or Meta, a Performance Max signal tells the algorithm where to start looking, but it remains free to serve beyond it. That is why the seeding matters and why it is easy to misread: strong first-party and intent signals give the automation a good starting population, but you still control the real boundaries through exclusions and conversion goals, not through the signal alone.
How do you measure whether Performance Max is actually working?
On contribution margin and incrementality, not the blended return the campaign reports. We separate brand from non-brand, run experiments and holdouts where the budget justifies them, and use script-level reporting to check spend is going to genuine prospecting rather than cheap impressions. The question we answer is whether PMax is winning profitable demand you would not have won anyway, which is a very different and much harder number than the one Google shows by default.