Performance Max Isn’t a Guessing Game Anymore. Here’s How to Actually Steer It in 2026.

For three years, the standard advice on Performance Max was “feed it good signals and hope.” Upload assets, write search themes, wait to see what the black box gave back. If it didn’t work, you had no real way to find out why.

That advice is out of date. Between mid-2025 and this fall, Google shipped nearly every piece of reporting and control advertisers had been asking for since PMax launched. It arrived in pieces through help-center updates and API release notes, so it never landed as one announcement, and most teams are still managing PMax the way they did in 2023.

Our position: PMax is not transparent, and probably never will be, but it is controllable. This matters now, because September is when most Q4 budgets lock in. 

What changed, and what most advertisers missed:

• Channel performance reporting across seven channels: Search, Display, YouTube, Discover, Maps, Gmail, and Search partners, with data back to June 6, 2025.
• Search terms reporting for all PMax campaigns.
• Asset-level stats across all campaigns, including assets Google generated through final URL expansion, which you can now remove.
• Negative keywords at 10,000 per campaign, up from 100, plus lists you can apply across campaigns.
• Fifty search themes per asset group, up from 25.
• Channel data in the Google Ads API (v23) at campaign, asset group, and asset level. Asset-group channel data is API-only, which is why almost nobody has seen it.
• Coming through 2026: a budget report projecting end-of-month spend, age and gender audience reporting, network-segmented placements, and first-party audience exclusions.

What most advertisers missed isn’t the announcements. It’s the why. Plenty of teams opened the channel report, saw they still couldn’t shift budget away from Display, and closed it. The channel report was never a control. It’s a diagnostic that tells you which of your other levers is misfiring. Heavy YouTube spend against a video Google assembled for you isn’t a YouTube problem; it’s a creative problem you can finally see.

The five levers, ranked by impact

1. What you’re bidding toward. Nothing else comes close. PMax spends against the conversion actions you mark as primary, at the values you assign. If a low-intent action like a button click is primary, that’s what you’ll get more of. Audit this first.
2. Campaign and asset group structure. Structure decides what gets its own budget, its own signals, and its own readable line in a report.
3. Exclusions. Negatives, brand exclusions, and soon first-party audience exclusions. At a 100-keyword cap this was a token lever. At 10,000, it’s a real one.
4. Creative inputs. PMax uses what you give it and generates what you don’t. Asset-level stats plus channel data finally let you connect weak assets to wasted spend.
5. Search themes and audience signals. Real, but the weakest of the five. These are hints, not targeting controls. Teams that spend an afternoon on search themes and ten minutes on conversion values have it backwards.

Why more asset groups usually beats more assets

The instinct is to build one asset group and stuff it with every headline and image you have. That made sense when asset-level reporting didn’t exist and volume was the only blind lever. Now, extra assets past the point of coverage mostly compete for the same impressions and make your reporting harder to read.

Separate groups are the better trade: each gets its own creative set, audience signal, 50 search themes, and reporting row. Split by service line, product category, margin tier, or geography when the offer truly differs. You can run up to 100 asset groups per campaign, so the ceiling isn’t the constraint – conversion volume is. Don’t create a segment you couldn’t evaluate at the end of a month. Always keep brand and non-brand separate, or your budget will drift toward the cheap and easy brand conversions.

Negatives and brand exclusions that work

PMax negative keywords apply to Search and Shopping inventory only. They do nothing to Display, YouTube, Discover, Gmail, or Maps. Account-level negatives apply automatically across that same inventory, so put universal exclusions there: “free,” “jobs,” “careers,” “salary,” “reviews,” and terms that read as research rather than intent. Build campaign lists from the search terms report, sorted by cost with zero conversions.

Don’t try to fill all 10,000 slots. Over-excluding starves the bid strategy, and you’ll feel it within a week or two.

For brand traffic, use brand exclusions instead of a pile of negatives. They automatically cover misspellings and sub-brands. Two caveats: PMax supports exclusions but not inclusions, which are Search-only, and you can let Shopping ads keep serving on excluded brand terms, which is usually right for retail.

The Q4 checklist: before October 15

1. Verify primary conversion actions and their values.
2. Pull the channel report for the last 90 days before planning budgets.
3. Build negative lists from the search terms report.
4. Set brand exclusions and decide who owns branded queries.
5. Audit asset groups against your Q4 offers and promotions.
6. Remove off-brand assets created by final URL expansion.
7. Upload real video. If you don’t, Google will build one, and it will show.
8. Set budget headroom before peak weeks.
9. Freeze structural changes after mid-October so learning periods don’t restart during your best weeks.

You still can’t set channel budgets. The goal isn’t to control PMax, it’s to make sure every input it reads from is one you chose on purpose. If you’d rather have someone else make that pass before budgets lock, let’s talk.

Performance Max FAQs

Can you actually control Performance Max?

You can’t see everything inside Performance Max, but you can control what it reads from. Between 2025 and 2026, Google added channel reporting, search terms data, and up to 10,000 negative keywords per campaign. The campaign still runs on the conversion actions, structure, exclusions, and creative you feed it, so those inputs are your real controls.

Do Performance Max negative keywords apply to every channel?

No. PMax negative keywords only apply to Search and Shopping inventory. They do nothing on Display, YouTube, Discover, Gmail, or Maps. To cut waste on those channels, focus on creative quality, asset group structure, and brand exclusions instead.

Should you build more asset groups or add more assets?

More asset groups usually wins. Past the point of solid coverage, extra assets in one group compete for the same impressions and make your reporting harder to read. Separate groups each get their own creative set, audience signal, search themes, and reporting row, so split by service line, product category, margin tier, or geography when the offer truly differs.

What should you optimize first in Performance Max?

Your conversion actions and their values. PMax spends toward whatever you mark as primary, so a low-intent action like a button click will get you more low-intent results. Audit this before you touch search themes or audience signals.

Can you set channel budgets in Performance Max?

Not yet. You still can’t tell PMax to spend a set amount on Search versus Display or YouTube. The new channel report is a diagnostic, not a budget control, so use it to spot which of your other levers is misfiring rather than to shift spend directly.