Google Performance Max Advertising Is No Longer a Black Box

Performance Max Stopped Being A Black Box And Nobody Noticed

Ask most business owners about Performance Max, and you’ll get some version of the same answer.

It’s a black box. Google spends your money and won’t tell you where. You have to trust it.

That was a fair description once. It isn’t any more, and the gap between the reputation and the reality is now wide enough to cost people money.

Over the last couple of years, Google has opened Performance Max up considerably. Search terms. Channel-level spend. Placements. Negative keywords. Brand exclusions. Device and demographic controls.

Each landed as a separate announcement, several months apart. Read one at a time, each looked like a small thing.

Read together, they mean the campaign type you were told to leave alone is now steerable. Most accounts are still leaving it alone.

I want to be careful not to overstate this. One real blind spot remains, and I’ll come to it. But “you can’t see anything” is no longer true, and repeating it is costing businesses the chance to fix obvious waste.

WHAT YOU CAN NOW SEE

Start with the one that changed the most.

The search terms report

Performance Max has a search terms report. You can select the campaign and see the queries that triggered your ads on Google Search.

This is the single biggest change, because search terms are where waste is most visible and most fixable.

It’s available to all advertisers, not just large accounts. Even accounts spending around fifty dollars a day get the detail that used to be hidden entirely.

There’s a caveat worth knowing. Only terms meeting a volume threshold appear, so very low-volume queries are still omitted. You’re seeing the meaningful bulk, not literally everything.

That’s a limitation rather than a dealbreaker. The terms wasting real money are usually the ones with enough volume to show up.

Channel-level reporting

This answers the question people actually ask, which is Where is my money going.

Channel-level reporting splits performance across Search, Display, YouTube, Discover and Maps.

It matters because Performance Max decides that split for you. If a campaign you think of as a Search campaign is quietly putting most of its budget into Display, you’d want to know, and now you can.

TLC

Placement reporting

You can see the specific sites and apps your ads ran on, segmented by network. An update in February 2026 extended this to Search Partner placements.

If you’ve read our post on Display waste, the same logic applies here. The rubbish inventory doesn’t avoid Performance Max because the campaign type is clever.

Audience and budget reporting

Expanded audience reporting and budget reporting arrived alongside the rest. Less headline-grabbing, useful when you’re trying to work out which signal is doing the work.

WHAT IT STILL HIDES

Now the honest part, because this is where most articles on the subject go quiet.

Performance Max does not report the queries that triggered your ads on Search Partner Network inventory.

Search Partners are sites other than Google that show Google search results. Your ads can appear there, and you can see the placements, but you can’t see what people typed to trigger them.

Why that matters

It means a portion of your spend is genuinely unexaminable. You can’t tell whether those impressions came from queries aligned with your business or from matches you’d never have chosen.

You can see that money went somewhere. You can’t audit whether it should have.

For a small account, this may be a minor slice. For a larger one, it can be meaningful. Either way, be aware it’s the part of the campaign you cannot inspect.

What to do about it

Two practical responses.

Use the placement reporting you do have, and exclude Search Partner placements that look wrong even without knowing the queries behind them.

And weight your judgement accordingly. When you’re comparing Performance Max against a Search campaign you can fully audit, remember you’re comparing something you can see all of against something you can see most of.

Performance Max is far more transparent than it was, and it is not fully transparent. Both things are true.

THE CONTROLS NOBODY SWITCHED ON

Visibility without control is just frustration. This is the half that actually changes outcomes.

Negative keywords, up to 10,000 of them

You can add negative keywords at campaign level in Performance Max, exactly as you would in Search or Shopping. They stop your ads appearing on unwanted queries across Search and Shopping inventory.

The campaign-level limit is 10,000 keywords, which is far more than any small business will need.

This is the direct pair to the search terms report. You look, you find the rubbish, you exclude it. That loop was impossible not long ago, and it’s routine now.

Brand exclusions

Brand exclusions stop the campaign from serving against named brands.

The common use is keeping a prospecting campaign away from your own brand searches. Without it, Performance Max will happily take credit for people who were already searching for you by name, which flatters the reporting and tells you nothing about new demand.

They apply at campaign level or at account level through the shared library, and asset groups inherit them from the parent campaign.

Account-level placement exclusions

The same account-level exclusion lists that cover Display apply here. Build once, apply broadly.

If you sell to businesses, app inventory is the first thing to look at, and it’s the same fix as the Display post covers.

Device and demographic controls

Performance Max now supports device targeting across mobile, computer and tablet, plus demographic controls.

Use these sparingly. They’re genuine controls, and narrowing an automated campaign too aggressively is a good way to starve it of the data it needs.

I’d reach for negatives and exclusions long before I reached for these.

ASSET GROUPS, AND THE LIMITS THAT CHANGED

Two limits moved, and both are worth knowing if you build the campaigns yourself.

Videos per asset group rose from five to fifteen. That matters because Performance Max serves on YouTube, and more video means more ways to compete for that inventory rather than relying on auto-generated clips.

Search themes per asset group rose from twenty-five to fifty. Search themes are how you tell the campaign what people search for when your own data is thin.

The practical advice on search themes hasn’t changed with the limit. More is not better. Fifty vague themes will teach the campaign less than fifteen accurate ones.

Take them from the same place as everything else worth having: the terms that already produced enquiries in your Search campaigns.

HOW TO READ THE SEARCH TERMS REPORT PROPERLY

Opening the report is the easy part. Knowing what you’re looking at takes a few minutes more.

Sort by cost, not by conversions. You’re hunting for money spent, not results earned.

Then sort the terms into four piles. They need different responses.

Your own brand name

This is usually the first surprise, and it’s the most expensive kind of good news.

People searching your company name were already coming to you. When Performance Max serves against those searches, it takes credit for demand it did not create.

Your cost per conversion looks excellent as a result. It’s measuring the easiest conversions in your account.

A negative keyword won’t fix this. Brand exclusions will. Switch them on, and the campaign goes back to doing the job you hired it for.

Adjacent categories you don’t serve

Repairs when you sell new. Domestic when you serve commercial. Hire when you sell outright.

These are the terms that look relevant to an algorithm and are worthless to you. They’re the biggest single source of recoverable waste in most accounts.

Straight into the negative keyword list.

Research and job searches

People writing dissertations, comparing options with no budget, or looking for work in your industry.

Job-related searches are worth checking specifically, because plenty of industry terms appear in recruitment adverts. If your trade name doubles as a job title, you’re probably paying for applicants.

Genuine terms performing badly

The last pile is the tricky one, and it’s where people over-correct.

A term that fits your business but hasn’t converted may simply not have had enough volume yet. Excluding it on thin data shrinks the campaign for no reason.

Give real terms a fair chance. Reserve the negatives for terms that were never going to work, not terms that haven’t worked yet.

WHY GOOGLE OPENED IT UP

Worth a short digression, because it tells you something about where this is heading.

Performance Max launched with very little reporting. Advertisers complained continuously, and the complaint was reasonable. People were being asked to hand over budget and targeting decisions with no way to check the outcome.

That’s a hard sell to anyone spending their own money rather than a media budget.

Competitive pressure played a part too. Microsoft Advertising has been publishing its own transparency improvements for its equivalent campaign type, and transparency became something the platforms compete on rather than resist.

The result is a steady stream of releases: search terms, channel reporting, placement reporting by network, negative keywords, brand exclusions, device and demographic controls.

What that suggests

Two things follow, and both are practical.

The first is that “you can’t see that in Performance Max” has been wrong repeatedly over the last two years. If somebody tells you a control doesn’t exist, it’s worth checking rather than accepting.

The second is that the reporting is likely to keep improving. The Search Partner query gap is the obvious remaining candidate, and I’d expect pressure on it to continue.

None of which helps you today. But it does mean an annual look at what’s newly available is time well spent, because the answer keeps changing.

TLC

PERFORMANCE MAX AND YOUR SEARCH CAMPAIGN

These two campaign types share an account, and they interact, which catches people out.

The overlap question

Performance Max serves on Google Search. So does your Search campaign. Both can be eligible for the same query.

Broadly, a Search campaign with a keyword that matches the query closely will tend to take precedence over Performance Max. But you shouldn’t rely on assumptions here, because the behaviour is more complicated than any simple rule suggests.

What matters practically is that you can now see the overlap. Compare the search terms report from Performance Max against the one from your Search campaigns, and you’ll see where they’re competing for the same ground.

The brand problem again

This is the overlap that costs the most.

If your Search campaign has a brand campaign, and Performance Max is also serving on your brand searches, you’re running two campaigns against the same free demand.

Brand exclusions on Performance Max fix it. Your brand campaign keeps those searches at the low cost brand terms usually carry, and Performance Max goes back to prospecting.

How to split them sensibly

The structure I’d suggest for a small account is simple.

Search campaigns take the terms you know work and want to control tightly. Performance Max takes the wider hunt, with brand excluded and a decent negative list.

Then judge them separately. Comparing an audited Search campaign against a Performance Max campaign with Search Partner queries you can’t inspect isn’t quite a fair fight, and it’s worth remembering when the numbers look close.

THE TWENTY-MINUTE PERFORMANCE MAX AUDIT

Here’s the whole thing, in the order I’d do it.

  • Open the search terms report for your Performance Max campaigns, 90 days, sorted by cost. This is the highest-value screen in the campaign, and most accounts have never opened it.
  • Read the top fifty terms and mark the ones that could never become a customer. Job seekers, students, DIY searches, other people’s products.
  • Add those as campaign-level negative keywords. You have room for 10,000, so there is no reason to be sparing.
  • Switch on brand exclusions if this campaign is meant to find new customers. Otherwise, it will keep claiming credit for people already searching your name.
  • Open channel-level reporting and check the split across Search, Display, YouTube, Discover and Maps. If most of the budget sits somewhere you did not expect, that is the finding.
  • Open placement reporting and exclude app inventory if you sell to businesses. Do it at account level, so it applies everywhere.
  • Check your search themes. Cut the vague ones and replace them with terms that produced real enquiries in Search.
  • Note what share of spend went to Search Partners, and accept that you cannot audit the queries behind it. Factor that into how much you trust the campaign’s own numbers.

That’s it. Nothing in that list requires a specialist, and the first three steps alone usually pay for the afternoon.

A WORKED EXAMPLE, START TO FINISH

A hypothetical business makes the shape clearer. Take a specialist equipment supplier on twelve hundred pounds a month, with most of it in one Performance Max campaign.

The campaign reports a healthy cost per conversion. Nobody has questioned it, because the number looks good and Performance Max is supposed to be left alone.

They open the search terms report for the first time.

Near the top is their own company name, taking a meaningful slice of spend. Those people were already looking for them, so the campaign is being credited for demand it did not create.

Below that, a run of terms about repairs and spare parts. They don’t do repairs. Somebody searching for a replacement part is not a buyer of new equipment.

Then a cluster of job-related searches, because their industry term also appears in job adverts.

They switch on brand exclusions, which removes the flattering own-brand conversions. The reported cost per conversion gets worse immediately, and that’s the point. It’s now measuring new demand rather than existing demand.

They add about forty negative keywords covering repairs, parts and jobs.

Then channel reporting, which shows a larger share going to Display than anyone assumed. Placement reporting shows a familiar pattern of app inventory, so they exclude apps at the account level.

A month later, the campaign spends the same and produces fewer, better conversions. The dashboard number looks worse than it did in month one.

The bank balance disagrees with the dashboard, and the bank balance is right. That’s the trade you make when you stop counting people who were already coming to you.

WHAT HAPPENS TO YOUR NUMBERS AFTERWARDS

This is the part that unsettles people, so it’s worth saying in advance.

Your reported performance will get worse. Sometimes a lot worse. That is the audit working, not failing.

Here’s why, in order.

Brand exclusions remove your easiest conversions

People searching your company name convert at a far higher rate than strangers. They’d made their decision before the ad appeared.

Take them out of the campaign and the average drops immediately. Cost per conversion rises, conversion rate falls, and the dashboard looks worse than it did last month.

Nothing got worse. You stopped counting the people who were already coming.

Negatives cut volume before they cut waste

Adding forty negatives reduces impressions straight away. Conversions may dip for a week or two while the campaign adjusts to a smaller pool.

Give it a fortnight before concluding. Automated campaigns need a little time whenever the inputs change.

The number that should improve

Qualified enquiries per pound, judged by you rather than by the platform.

Count the enquiries that were genuinely worth having over the month before and the month after. That’s the comparison that matters, and it’s the one the dashboard can’t do for you.

If the reported cost per conversion doubled but the number of real enquiries held steady on the same spend, you improved the account. If both fell together, you cut too hard and should relax the negatives.

Write down the before numbers first. Almost nobody does, and without them this comparison is impossible to make honestly.

WHEN PERFORMANCE MAX IS THE WRONG CHOICE

Being fair about this matters, because the campaign type gets sold hard.

It’s a poor fit when you have almost no conversion data. Performance Max learns from conversions, so an account with two a month gives it very little to work with.

It’s a poor fit when your conversion tracking is unreliable. An automated campaign optimising towards a badly configured conversion will chase the wrong thing efficiently, which is worse than doing nothing.

It’s a poor fit when you need tight control over messaging in a regulated sector, because you’re handing over a lot of decisions about combination and placement.

And it’s the wrong first move if your Search campaigns still lose impression share on your best terms. Capture the demand you can see before automating the search for more.

Where it works well is a business with steady conversion volume, reliable tracking, and enough budget to let it learn. In that setting, and now with the controls above switched on, it earns its place.

THE MISTAKES THAT WASTE THE MOST MONEY

  • Believing the black box story and never opening the search terms report. It exists now, and it’s the most valuable screen in the campaign.
  • Leaving brand exclusions off in a prospecting campaign, so it claims credit for people already searching your name.
  • Adding no negative keywords at all when the campaign accepts up to 10,000.
  • Never checking channel reporting, so you don’t notice most of the budget went to Display.
  • Ignoring placements, on the assumption that automated campaigns avoid rubbish inventory. They don’t.
  • Piling fifty vague search themes into an asset group because the limit went up.
  • Judging Performance Max against a Search campaign without allowing for the Search Partner queries you cannot see.
  • Running it with two conversions a month and expecting the automation to find a pattern.

TLC

FREQUENTLY ASKED QUESTIONS

Can you see search terms in Performance Max?

Yes. There’s a search terms report showing the queries that triggered your ads on Google Search.

It’s available to all advertisers, including small accounts spending around fifty dollars a day.

Only terms above a volume threshold appear, so the very long tail is still omitted. The terms costing you real money generally have enough volume to show.

How do I add negative keywords to Performance Max?

At campaign level, in the same way as Search and Shopping. They prevent your ads from appearing on unwanted queries across Search and Shopping inventory.

The limit is 10,000 per campaign, so there’s no need to ration them.

Work from your own search terms report rather than a generic list. Your account’s rubbish is specific to your account.

What are brand exclusions in Performance Max?

They stop the campaign from serving against named brands.

The main use is keeping a prospecting campaign off your own brand searches, so it isn’t credited for people who were already coming to you.

They apply at the campaign or account level, and asset groups inherit them from the parent campaign.

Expect your reported cost per conversion to get worse when you switch them on. That’s the number becoming honest, not the campaign getting worse.

How do I see where Performance Max is spending my budget?

Channel-level reporting splits performance across Search, Display, YouTube, Discover and Maps.

Placement reporting, segmented by network, shows the actual sites and apps. Since February 2026, that extends to Search Partner placements.

Between the two, you can usually work out whether the budget is going where you assumed.

Is Performance Max still a black box?

Much less than it was, and not entirely open either.

Search terms, channel spend, placements and audiences are all reportable. Negative keywords and brand exclusions give you real steering.

The genuine remaining gap is that queries triggering ads on Search Partner inventory are still not reported. That slice stays unexaminable.

How many search themes can an asset group have?

Fifty, up from twenty-five. Videos per asset group also went from five to fifteen.

Higher limits aren’t an instruction to fill them. Fifteen accurate themes beat fifty vague ones.

Should I use Performance Max or a Search campaign?

Search first if you can still grow it. It’s cheaper to capture demand than to automate the hunt for it.

Performance Max earns its place once Search is capped, your conversion tracking is solid, and you have enough conversion volume for the automation to learn from.

WHERE TO START THIS WEEK

Open the search terms report on your Performance Max campaign. Ninety days, sorted by cost.

If you’ve never looked, that single screen is likely to pay for the time on its own. Most accounts find their own brand name, a category they don’t serve, and a run of searches from people who were never going to buy.

Then do the three things that follow from it. Add the obvious negatives, switch on brand exclusions if the campaign is meant to find new customers, and check the channel split.

Expect your reported numbers to look worse afterwards. Removing your own brand searches from a prospecting campaign always makes the dashboard less flattering and the decisions better.

The campaign type earned its black box reputation fairly. It just doesn’t deserve it any more, and the businesses still repeating it are the ones leaving the controls switched off.

Did You Enjoy This Blog Post?

I hope you enjoyed this blog post, and thank you so much for being here. We also upload videos to our YouTube channel every weekday. Please subscribe so you are one of the first to be notified.

If you enjoyed this blog, you may also like:

Tools We Use & Love!
Scroll to Top