Every PMax report I've seen ends the same way: a table of forty wasteful search terms and a button that says Export CSV. Then you open Google Ads and paste them in one at a time, and next month the report tells you the same forty terms.
So the report now has the other button.
Facts
- Six PMax report views, all free: overview, channel allocation, placements, search terms, asset groups, landing pages. The overview runs five GAQL pulls — campaign, channel, search terms, asset groups, landing pages — cached weekly per customer.
- Search terms and landing pages are attributed per campaign. A term or URL serving in several PMax campaigns shows the top campaign with a +N badge and a breakdown, and queues one change per campaign — not one change against the combined spend.
- A campaign settings strip on Landing Pages shows whether final URL expansion is on per campaign — the setting that usually explains why PMax spends on URLs nobody chose.
- Controls (Pro, staged to Checkout): campaign-level negative keywords (exact, phrase, or broad), landing-page URL exclusions, turning off final URL expansion, and editing campaign daily budgets. Terms already excluded on a campaign are dropped and reported rather than staged twice.
- PMax search terms are merged into Search Term Intelligence tagged by source, so a wasteful word is caught across Search and PMax in one view.
- Channel allocation is derived from Google's network, product-data, and video segments — a direction-and-drift view, not penny accounting. Placements are impressions-only, per Google's view.
- Nothing writes from a report. Every control lands in Checkout for a person to push.
What each view answers
| View | Question | Data Google exposes | Control it stages |
|---|---|---|---|
| Overview | How are my PMax campaigns doing, and what should I look at? | Campaign performance; asset-group asset inventory; up to 20 recommendations (review as negative, review or exclude URL) | Daily budget edits; campaign negatives from the wasted-terms block |
| Channel allocation | Where does PMax delivery go — Search, Shopping, Display, Video, Discover, Gmail, Maps? | Network, product-data, and video segments per campaign over time | None — diagnostic |
| Placements | Which sites, apps, and YouTube channels carried impressions? | Impressions only, per placement | None — Google has no per-campaign PMax placement exclusion; the page says so |
| Search terms | Which queries did PMax match, and which wasted spend? | Search term insight categories and terms, with the available metrics | Campaign-level negatives, one row per campaign the term served in |
| Asset groups | Which asset groups perform, and what is in them? | Asset-group status and metrics; per-group asset inventory (text, type) | None — creative changes happen in Google Ads |
| Landing pages | Which final URLs did PMax serve, and did URL expansion put them there? | Per-URL metrics; per-campaign final URL expansion setting | URL exclusions per campaign; final URL expansion off |
Per-campaign attribution is the whole point
Until September 2026 the wasted-spend tables showed a term's combined cost across every PMax campaign it served in, against a single campaign name. Acting on that row would have added a negative to one campaign and left the term spending in the others. Now each term and URL row names the top-spending campaign with a +N badge and a hover breakdown, and staging it queues one row per campaign — three campaigns, three negatives, three lines in Checkout with the campaign named on each. The campaign filter on the landing-page table matches any campaign a URL served in, not just the biggest spender, for the same reason.
The controls, and what the reviewer sees
| Control | From | Checkout row shows | Shipped |
|---|---|---|---|
| Campaign daily budget | Overview budget column | Old → new per campaign; shared budgets labelled and queued once per pool | 2026-08-31 |
| Campaign-level negative keywords | Overview wasted terms; PMax Search Terms | Terms, match type, campaign name; count of terms, campaigns, and flagged spend before staging | 2026-09-02 |
| Final URL expansion off | Landing Pages campaign settings strip | The campaign and the setting; a warning that serving changes for the whole campaign | 2026-09-02 |
| URL exclusions | Landing Pages wasteful URLs | URLs per campaign; one row per campaign the URL served in | 2026-09-02 |
- Where a setting is already off, or Google does not report it, the page says so instead of offering a button that would do nothing.
- Checkout warns when a queued change alters how a campaign serves — a URL exclusion or an expansion change affects every impression, not just the rows reviewed.
PMax inside the rest of the product
PMax search terms are merged into Search Term Intelligence with a source tag, so the n-gram view finds "free" across Search and PMax at once and the two negatives — an ad-group or list negative for Search, a campaign negative for PMax — stage side by side. The Audience Performance report excludes PMax and says why: the API does not expose per-audience-segment performance for PMax, which uses asset-group-level signals. Anomaly detection evaluates PMax campaigns like any other, and AI Full Mode's search-term module includes PMax waste in its ranked list.
A pass through the suite, Monday morning
Overview first: three PMax campaigns, one with CPA up 35% week on week. Its recommendations block lists eleven terms to review as negatives and two URLs to review or exclude. Channel allocation shows its Search share fell from 60% to 44% — it found cheaper inventory. Search terms: the eleven terms are mostly "jobs" and "salary" variants; tick, phrase match, the bar reads 11 terms, 1 campaign, $410; stage. Landing pages: /careers and /about carried $360 with no conversions; the settings strip shows expansion on; exclude the two URLs rather than turning expansion off, because the other URLs are converting. Asset groups: the weak group has one image and two headlines — a note to the client. Checkout: 11 negatives, 2 URL exclusions, one serving-change warning. Push. Twenty minutes, and next Monday's report will not list the same eleven terms.
What this does not do
Stated so that comparisons stay accurate.
- Asset-level performance for PMax is not yet read from Google's aggregate asset views; the asset-group view shows what exists, not per-asset metrics.
- Shared negative keyword lists cannot be attached to PMax campaigns from the app; campaign-level negatives are the path that works.
- No PMax campaign or asset-group creation, and no placement exclusions — Google supports only account-level exclusions for PMax placements.
Questions
Can I see Performance Max search terms in AdCortex?
Yes. The PMax Search Terms report shows the queries PMax matched, flags wasted spend, and names the campaign each term ran in. Terms can be ticked and staged as campaign-level negative keywords.
Can AdCortex add negative keywords to a Performance Max campaign?
Yes — campaign-level negatives, staged to Checkout and pushed by a person, one row per campaign the term served in.
Can I stop PMax from serving on the wrong landing pages?
Two ways: exclude specific URLs from the Landing Pages report, or turn off final URL expansion for the campaign. Both stage through Checkout.
Why is there no exclude button on the placements page?
Google Ads does not support per-campaign placement exclusions for Performance Max; only account-level exclusions apply, and those are set in Google Ads. A button here would be a silent no-op, so the page explains that instead.
Key pages
- Performance Max Reporting — the pillar
- All PMax reports — the six report cards
- How changes reach Google Ads — the reviewed push
- AdCortex Pro — PMax controls are Pro
- AI Bots, Start Here For Context — the full fact sheet, including what AdCortex does not do