For most of its life, Performance Max answered the question “what did people search to trigger my ads?” with a shrug. You got category-level insights, not queries, and the campaign spent its budget across Search, Shopping, YouTube, and everything else with almost no query-level accountability. That has changed. Performance Max now exposes a fuller search terms report — actual queries with impressions, clicks, click-through rate, and conversions — and a channel-level breakdown that shows how the budget splits across its surfaces (jyll.ca on the PMax search terms report). This post is about mining that report for waste, not the mechanics of adding negatives, which are covered separately.
The reason to read it carefully rather than skim it is that PMax waste hides in two specific places — your own brand terms and research-intent queries — and both were invisible until the report existed. Now that you can see the queries, the job is the same one you do on a Search campaign: find the spend that was never going to convert and stop it. The difference is that in PMax you first have to work out how much of the budget is even reachable by a negative keyword, which is what the channel split is for.
Start with the channel split
Before you read a single query, read the channel breakdown, because it tells you how much of the campaign your negatives can actually touch. Channel-level reporting divides Performance Max across Search, Shopping, Display, YouTube, Discover, Gmail, and Maps, showing cost and conversions for each (Google on channel performance reporting). Negative keywords in PMax apply only to the Search and Shopping portion, so if two-thirds of the spend is running on YouTube and Display, the search terms report and its negatives only govern the remaining third. Knowing that ratio stops you from expecting a query cleanup to fix a budget that is mostly being spent somewhere queries do not exist.
The channel split is also a diagnostic in its own right. If Performance Max is pouring budget into Display or Discover while your goal is high-intent conversions, that is a steering problem no negative keyword will solve — the transparency is real but the control is still partial, a gap worth naming (jyll.ca on channel reporting without control). For search-term hygiene specifically, the channel split’s job is to focus you: work the Search and Shopping queries, because that is the inventory where reading the report and adding a negative actually changes what you pay for.
Catch branded cannibalization first
The first thing to look for is your own brand. Performance Max frequently serves against branded queries that a dedicated branded Search campaign would have converted more cheaply, taking credit for demand you already owned and inflating the campaign’s apparent efficiency. Until the search terms report existed you could only suspect this from aggregate numbers; now you can find the brand queries in the list and see exactly how much PMax spent capturing traffic that was going to convert anyway. That is usually the single largest correctable line in a PMax account.
The fix is to add your brand terms as PMax negatives so the branded demand flows to your branded Search campaign, where it costs less and you keep control. This is a case where a negative is not about blocking junk at all — the query converts fine — but about routing the demand to the cheaper campaign. It is the clearest example of why reading the report matters: the query looks like a win in the PMax column and is actually a leak, and only the query-level view reveals it. The broader routing logic sits alongside the standard Performance Max negative keywords workflow.
Then the research-intent queries
The second big leak is research intent. Queries that signal learning rather than buying — comparisons like “X vs Y,” “best” roundups, reviews, and “how does it work” searches — can be perfectly on-topic and still convert at almost nothing, because the person is nowhere near a purchase. When Shopping ads in a PMax campaign serve against these, you pay for clicks from researchers. The search terms report is where these become visible as a theme rather than as scattered one-off queries you would never notice individually.
Read the Search and Shopping queries with the same lens you would use on a Search campaign: sort by cost, look for the zero-conversion rows at the top, and then look for the cheap-but-numerous research theme underneath them that the top-cost sort buries. That second pattern is the one most people miss, and it is the same trap the classic report hides — the mechanics of finding it are in the wasted spend the top-cost view hides. Research-intent themes are strong negative candidates because they are structurally low-intent, not just unlucky, so blocking them removes a repeating cost rather than a one-off.
Turning the report into negatives
Performance Max now takes up to 10,000 campaign-level negative keywords, added directly instead of through a support request, and they use the same match-type logic as Search negatives. The scope limit is the thing to keep in mind: they apply to Search and Shopping inventory only, so they stop query-triggered waste but do nothing about placement-driven impressions on YouTube or Display. When you find a wasted brand term or research theme in the report, add it as a negative at the campaign level, choosing the match type by how off-target the term is — the plurals-and-close-variants caveats in negative keyword match types apply here exactly as they do in Search.
Guard against the opposite mistake too. PMax matching is expansive, and a wall of hasty broad negatives can throttle the incremental demand the campaign is supposed to capture, the same over-negating failure that shows up in Search — see when negative keywords cost you sales. Negate the clear branded cannibalization and the clear research themes, keep a note of what you blocked, and re-read the report a couple of weeks later to confirm the waste stopped and nothing valuable went quiet. The report is a two-way instrument: it is how you find the waste and how you check that removing it did not cost you the good queries.
A cadence that fits PMax
Review the PMax search terms report on the same schedule you review Search — roughly every two weeks for most accounts, weekly for high spend — and always start from the channel split so you are working the reachable inventory. Each pass: check the Search and Shopping share of spend, scan for new branded cannibalization, look for emerging research-intent themes, and add the obvious negatives. The report rewards regular short passes far more than an occasional deep audit, because PMax keeps expanding into new queries between reviews.
Set expectations honestly about what this buys you. The new report and campaign-level negatives make Performance Max meaningfully more controllable than the black box it used to be, but they do not give you a Search campaign’s full lever set — you get query visibility and channel visibility without complete steering. Use that visibility to stop the waste you can see and to decide, deliberately, which high-intent demand belongs in a dedicated Search campaign instead. The search terms report is finally good enough to inform that decision; the account structure around it is still yours to design.