What you'll learn in this article
- Why the channel distribution table is the only section I work from
- What the google ads performance max channel reporting update november 2025 actually changed
- The one inference about a surface split that everyone gets backwards
- How channel data feeds exclusions and why the lever is never where the spend is
For three years the honest answer to "where is my PMax budget going?" was a shrug with a placement report attached. The two questions clients actually ask about Performance Max campaigns is it eating my brand traffic, is it dumping money into Display were unanswerable from inside the interface. That's what people meant by black box: not that it was mysterious, but that the money left the room and never filed a report.
That gap is now closed, more narrowly than the announcements suggest. You get impressions, clicks, conversions and cost per surface, across Search, YouTube, Display, Gmail, Discover and Maps plus Search partners. You get no control over any of them. That asymmetry is the whole story.
How I pull the channel performance report
In your Google Ads account: Campaigns menu, Insights and reports, Channel performance. Google documents the structure plainly the channel performance page has three sections, a performance summary at campaign level, a Channels-to-Goals chart, and the channel distribution table, plus Download and Segment on the table toolbar Google Ads Help on the channel performance report. One constraint before you promise anyone a trend line: the date picker only goes back to June 2025. No year-over-year, and none coming until the data ages into it.
I skip the chart and go to the table
The Channels-to-Goals visualisation is a persuasion object, not a working document. Cost lives per row in the distribution table, and I add a share-of-spend column myself: Google gives currency, my question is proportion. Same omission, same fix as the PMax asset report.
Where Shopping actually is
The most common confusion I field: there is no Shopping row. Shopping isn't one of the channels in Performance Max it's a format. Ads built on product feeds sit behind the "Ads using product data" segment, spanning Search, YouTube Search, Maps and Search partners plus dynamic remarketing ads on Display. Skip it and you'll credit Search with 60% of the budget when much of it was search-shop intent clicking product ads.
The status column earns its keep
Before interpreting a single percentage I read the status column and diagnostics: eligible, eligible with limits, not eligible. It names the blocker outright missing video asset, Merchant Center problem, disabled final URL expansion. That's the difference between "the algorithm chose not to spend here" and "it couldn't": identical in the cost column, opposite actions.
The November 2025 update, and what it changed for me
The report shipped in stages announced at Google Marketing Live in May, open beta over the summer, then bulk reporting and segmentation in September. The google ads performance max channel reporting update november 2025 is the one that mattered: general availability across all PMax campaigns, Search partners added as its own line in the visualisation and the table, and API access.
General availability is the boring change that matters most
Beta-gated reporting is worthless for auditing accounts I didn't build a report I can only get sometimes isn't a process, it's a coincidence. From November, performance max channel reporting became the first thing I open.
Search partners as its own row
The change with real teeth, because partners inventory was previously folded invisibly into a Search story clients had decided they liked. It performs differently from google.com, and in Search campaigns you can switch it off. In PMax you can't. The row gives you a number to be unhappy about and no lever useful anyway, because it explains gaps against what a standard Search campaign delivers on the same terms.
API access, and what still isn't there
The interface stops scaling around a dozen campaigns. Two API caveats: PMax has no ad group entity, so anything built against ad group resources returns nothing, and figures won't tie to the interface to the cent noise for share-of-spend work, a wasted morning if you reconcile it. Still missing: placement detail, search terms, and any per-channel control. Reporting improved dramatically in 2025; the levers didn't move an inch.
What a surface split actually tells me
The inference this report supports, and the one it doesn't. It does not tell you which surface is best. It tells you which surface the algorithm found cheapest at the margin, at that moment, given your target.
Marginal versus average ROI
The sentence I'd tattoo on the report, and Google says it themselves: PMax bids in real time on marginal ROI the return on the next euro not average ROI. A channel with a worse average return can correctly take more budget, because the next euro there is still cheaper than the next euro elsewhere. Read the table at average and every conclusion is backwards: the channel with the best ROAS is often the one that ran out of cheap inventory, which is why it's small.
Why Display is usually less of a scandal than it looks
Three checks before I call it waste. Is it remarketing segment by ads using product data and much of that row is often your own feed retargeting your own visitors. Conversion delay channels are hit unevenly by lag, so a window including the last two weeks flatters Search and punishes upper-funnel surfaces. Attribution on last-click, a surface that opens the journey always looks like a passenger.
The zero row, and the one that's urgent
Eligible with no issues means the algorithm found better returns elsewhere and there's nothing to fix; limited means a missing input, and the fix is upstream. "PMax won't run on YouTube" turns out to be "this campaign has no video assets" more often than any other diagnosis. The real emergency is different: meaningful spend, near-zero conversions, sustained a month, on one surface. That's breakage, and it's invisible at campaign level because the healthy surfaces average it away.
The Search question everyone actually has
Nine times out of ten, "where does my spend go" is really "is it buying my brand terms." This report can't answer that the surface says nothing about the query. If you're chasing it, the report you want is about overlap between PMax and Search.
How I use channel data for exclusions
The ceiling first: you cannot exclude a channel. No checkbox for Display, no cap on YouTube, no Search partners toggle. Everything below is indirect influence, and Google is upfront that most of it reduces reach.
The exclusions that genuinely exist
Account-level placement exclusions work, and they're where I start when a Display or YouTube number is a brand problem rather than a performance one. Content suitability settings do real work on the YouTube side. Account-level negative keywords now apply to PMax, the closest thing to a Search-side lever you'll get. Short list, entirely defensive: none of it moves budget where you want, it only stops it going where you don't.
Why I almost never exclude on a bad ROAS
Because of marginal ROI, and because the campaign optimises as one unit. Excluding a surface for a poor average return removes inventory the model used to hit your target cheaply, and the budget doesn't relocate to your favourite channel it competes harder for what's left, at worse prices, or it stops. I've watched an account exclude a large Display slice and land on the same CPA with 30% less volume. Nothing improved; the table just looked nicer.
Brand safety is a different and legitimate argument it needs no ROI justification and I'll take the reach hit without arguing. What I push back on is dressing an efficiency opinion up as a safety one.
What I do instead, in order
Fix the diagnostics first every limited status is inventory you're paying for and not getting. Then feed the surfaces you want: text assets and search themes for Search, video for YouTube, which can't serve without one. Assets are the closest thing to a channel-mix lever you have when you optimize your campaigns, which is why asset group structure shapes where money lands. Only then the structural option: if a surface needs its own budget and target, it doesn't belong inside PMax.
My cadence, and what I write down
Monthly. Structural questions don't move in seven days weekly pulls made me worse at this, seeing noise and calling it a trend. And always excluding the last two weeks, because conversion lag hits surfaces unevenly.
Per surface I log four things: share of spend, ROAS, conversion count, and one sentence explaining the share. The sentence is the artifact. "Display is 34% because two thirds of it is dynamic remarketing on our own feed" is a structure working as designed. "Display is 34% and nobody knows why" is a project. Twelve rows is a year of history and the only way to tell drift from design on the slower loop of my optimization routine.
The honest summary: google ads performance max channel reporting turned a sealed box into a glass one you still can't reach into. You can see the split, explain it, and influence it only through assets, signals, diagnostics and exclusions that cost reach. Most months you'll conclude the algorithm was right. The value is entirely in the months you don't.