Where Performance Max Wasted Spend Hides

Performance Max spends in two places, and only one of them leaves a trace

A Performance Max campaign buys across Search, Shopping, Display, YouTube, Discover and Gmail from a single budget. Some of that inventory starts with somebody typing something. The rest starts with Google deciding you should appear next to a video or in a feed. Only the first kind produces a search term.

That single fact decides everything about what you can audit. If a query existed, it may show up in campaign_search_term_view, the same reporting resource that serves Search and Shopping campaigns. If no query existed, there is nothing to report, nothing to negate and nothing a negative keyword tool can touch. Display, Video and App campaigns report no search terms at all, and the non-search inventory inside Performance Max behaves the same way.

We are not going to tell you what percentage of your Performance Max budget goes to each side. Google does not publish that split, it varies wildly by account, and every confident number you have read on the subject came from somebody else's campaign. What you can do is establish a floor for your own account, which is more useful than anyone's average.

How to measure your own search-versus-everything-else floor

Pull the search terms for one Performance Max campaign over the last 30 days. Sum the clicks. Compare that to the campaign's total clicks over the same 30 days. The gap is everything Google did not report: non-search inventory plus search terms that fell below Google's reporting threshold.

That gives you a floor, not a split. You now know that at least this many clicks came from a readable query, and that the rest is some mixture of two things you cannot separate. It is a blunt number and it is yours.

In the accounts we audit, that floor tends to sit far higher for retail Performance Max campaigns with a product feed than for lead-gen campaigns with no feed at all — the feed pulls the campaign toward Shopping inventory, and Shopping inventory answers queries. Beyond that direction of travel we will not give you a figure, because a range wide enough to be honest would be too wide to be useful.

Do the exercise per campaign, not per account. Two Performance Max campaigns in the same account, one with a feed and one without, will behave like two different products.

The readable half: what search terms in Performance Max actually show

Performance Max search terms are real search terms. They are not the "search categories" bucket you see in the campaign Insights tab, which groups queries into themes like "running shoes" and refuses to break them down. The reporting resource gives you individual terms with clicks, cost and conversions attached.

What it does not give you is all of them. Google withholds terms below a volume threshold for privacy reasons, and that withholding applies across Search, Shopping and Performance Max alike. The long tail — the one-click oddities that never repeat — is exactly the part that disappears. We wrote the full version of this in Google Ads Search Terms Report, and What It Hides, and the short version is that the terms you can read are the ones with enough volume to matter, which is fine, and the ones you cannot read are individually tiny but collectively not, which is not fine and cannot be fixed.

So when you audit Performance Max search terms, you are auditing a real but incomplete slice. Work it anyway. Incomplete and actionable beats complete and imaginary.

A typical week of readable waste in a plumbing account looks like this:

• plumber salary uk — somebody researching a career
• plumbing apprenticeship near me — same
• how to unclog a kitchen sink without a plunger — somebody who has decided not to call anyone
• plumbers merchants wholesale — trade buyer, not a homeowner
• is pimlico plumbers expensive — competitor research, maybe worth keeping, maybe not

Four of those five are money you will never get back. The fifth is a judgement call we covered in Competitor Searches in Google Ads, Block or Keep?.

In retail the pattern is different but just as readable:

• sony wh-1000xm4 ear pad replacement when you sell the headphones, not the pads
• used dyson v11 gumtree — second-hand intent
• nike air max fake vs real — authentication research
• free shipping code [your brand] — a discount hunter who was already coming
• [your product] repair manual pdf — an existing owner with a problem

None of these are mistakes by Google in the sense of being unrelated to your products. They are all about your products. They are just not about buying one. That distinction is the whole job, and it is why keyword-level relevance checks miss so much of it — more on why the auction produces these in Why Google Ads Shows Irrelevant Search Terms.

The opaque half: no query, no negative keyword, no fix

When your ad runs before a YouTube video or inside a Gmail promotions tab, there is no search term. There never was one. Adding free as a negative keyword does not stop a Display impression, because no word was typed for it to match against.

This is the part every "stop wasting money in PMax" article walks around instead of saying plainly. Negative keywords are a search-side tool. Inside Performance Max they govern the search-side inventory and nothing else. If your campaign is spending heavily on non-search placements and converting badly, no list of 10,000 negatives will change that number.

What does move it is structural: separate campaigns so budgets cannot bleed from one intent to another, feed segmentation so weak products are not subsidised by strong ones, and asset quality, because poor assets get shown where poor assets belong. Those levers are real. They are also not what we do, and pretending otherwise would be dishonest.

The practical consequence: judge Performance Max on two separate questions. "Is the search side buying the right queries?" is answerable and fixable tonight. "Is the non-search side worth what it costs?" is answerable only by changing structure and watching totals move, which takes weeks and a lot more nerve.

Why the standard Performance Max advice stops short

The advice you will find elsewhere is not wrong. It is just consistently missing the same two pieces.

"Add brand exclusions." Correct, and it takes ten minutes. It also only handles your own brand. The waste in a mature account is rarely brand — it is the thousand non-brand terms that sit somewhere between relevant and useless. Brand exclusions do nothing for plumber salary.

"Use search themes to steer it." Search themes are inputs, not constraints. They tell the system what you want. They do not tell it what to refuse. A campaign given a good theme will still buy adjacent queries it believes might convert.

"Split your campaigns." Good advice for feed segmentation and budget control. It also multiplies your negative keyword workload, because Performance Max does not accept shared negative keyword lists. Six Performance Max campaigns means six separate copies of the same list, maintained separately, forever. That is the detail that quietly kills the manual approach, and we worked through the arithmetic in Negative Keywords in Performance Max and the 10,000 Cap.

"We cut PMax spend by 80% and kept conversions." Those case studies are usually a structure story, not a negatives story. They are worth reading and they do not replace weekly search term work.

The piece none of them include: search term waste is not a project. It is a recurring condition. You clean a campaign on Monday, and by Friday it has found four new ways to spend money on people who wanted a salary figure. A one-off audit produces a good screenshot and a temporary result.

Deciding which Performance Max search terms were never going to convert

Relevance is the wrong test. plumbing apprenticeship is highly relevant to a plumbing business. It is also worthless to a plumbing business buying clicks. Run three questions instead.

Did this person want to buy, or to know?

Informational queries — how to, what is, vs, guide, template, pdf, calculator — usually want an answer, not a supplier. Some businesses genuinely close from these. Most do not, and the ones that do can tell from their own data because the term converted.

Are they a buyer at all?

Job seekers, students, journalists, wholesalers, existing customers looking for support. Salary and career terms are the cleanest example and the easiest to miss, because the head keyword is right there in the query — that whole family is catalogued in Job Seekers Clicking Google Ads? Block These Terms.

Are they buying something you do not sell?

Accessories when you sell the main unit. Second-hand when you sell new. Rental when you sell outright. B2B quantities when you sell to consumers. These are close enough to your business that the system keeps finding them, and far enough that they never convert.

Then apply the only hard rule worth having: if the term converted recently, leave it alone regardless of how ugly it looks. PPCleaner never blocks a term that converted in the last 90 days, because the strangest-looking query in the report is occasionally the one paying for the month. We would rather keep ten bad terms than kill one good one, and any rule that cannot state which way it errs is not a rule.

For the order in which to look at all of this when a campaign is bleeding, Google Ads Spending Money, No Conversions? Check This Order is the sequence we use.

Adding the negatives without breaking the campaign

Performance Max accepts up to 10,000 campaign-level negative keywords. That is a lot of room and it is also a real ceiling that active accounts reach. Three things decide how fast you get there.

Match type choice. Broad, phrase and exact are your three options, and they do not behave like their positive equivalents. A phrase negative on salary kills the entire salary family in one slot. Ten exact negatives kill ten specific queries and nothing else. Phrase negatives are the efficient use of the cap. Broad negatives behave unpredictably across match types and can block more than you intended — the mechanics are in Negative Keyword Match Types Don't Work Like Positives.

Duplication across campaigns. No shared lists means the same negative exists once per campaign. If you run five Performance Max campaigns, one salary phrase negative consumes five slots across the account and five separate edits every time you change it. Standard Shopping does accept shared lists, which is one of several ways the two behave differently — see Shopping Campaign Negative Keywords Work Differently.

Over-blocking. A single-word negative that looks safe rarely is. free blocks free delivery on orders over $50. cheap blocks cheap flights to your destination city in a travel account that wanted exactly that. The cost of a bad negative is invisible, because blocked traffic leaves no row in any report. That asymmetry — visible waste versus invisible loss — is why How Many Negative Keywords Are Too Many? argues for reversibility over volume.

Adding negatives does not reset Performance Max learning. The campaign keeps its conversion history. What changes is the pool of queries it is allowed to buy, which is the point.

What PPCleaner does inside Performance Max, and what it does not

We read campaign_search_term_view every night for your Search, Shopping and Performance Max campaigns. We judge which terms were never going to convert. We add them as campaign-level negatives on the Performance Max campaigns — individually, because shared lists are not available — and to the relevant lists on Search and Shopping. We never touch budgets, bids, ads or keywords. Every block is reversible in one click for 90 days, which matters more than any accuracy claim, because the only honest answer to "what if you get one wrong" is "undo it".

What we do not do: reach the non-search half. We cannot stop a Display placement, we cannot exclude a YouTube channel, and we cannot tell you how your Performance Max budget split between the two. We do not see the search terms Google withholds below its threshold, so the long tail stays where it is. And we do not restructure campaigns, which is the lever that actually moves the opaque half of the spend.

If you want the manual version first — and you should try it once, to see what your own account looks like — How to Find Wasted Ad Spend in Google Ads is the same process done by hand. Do it for three weeks. Then decide whether you want to do it for the next three hundred.