Performance Max Channel Reporting in 2026: Where the Budget Actually Goes
The standing complaint about Performance Max has always been the same: you hand over one budget and have no idea whether it went to Search, YouTube, Display or Gmail. Network reporting returned a single blended value, so every conversation about reallocating money was a conversation about guesses. In 2026 that changed: Performance Max channel reporting shows where the budget actually lands.
With Google Ads API v23, released 28 January 2026, that gap is closed. Performance Max performance now breaks out by channel — Search, Search Partners, YouTube, Display, Discover, Gmail and Maps. This guide covers what became visible, where to pull it, which conclusions the data supports, which it does not, and what you should actually change once you can see the split.
What changed in Performance Max channel reporting
The ad network type segment used to return a blended value for Performance Max — effectively “mixed.” It now returns the specific serving channel. Two additional flags shipped alongside it:
- Product data flag — whether the impression used Merchant Center product data. This separates the shopping half of PMax from everything else.
- Video flag — whether the impression involved a video asset, isolating video inventory from static within a single campaign.
The breakdown is available at campaign, asset group and asset level. One important constraint: asset group granularity is API-only and will not appear in the web interface. Historical data reaches back to 1 June 2025, so full year-over-year comparison is not possible yet.
| Before | Now |
|---|---|
| One blended network value per campaign | Seven channels reported separately |
| No way to separate shopping from display | Product data flag |
| Video and static blended together | Video flag |
| Campaign level only | Campaign, asset group and asset (last two via API) |
Where to pull the data
Option 1 — the Google Ads interface
Campaign-level channel breakdown is available in Performance Max campaign reporting. Open the campaign, go to the statistics view and segment by network, or open the channel distribution view. That is enough to answer “where did the money go” and see the overall shape of spend.
Option 2 — report editor and custom columns
For a recurring view, build a report segmented by channel with spend, conversions, conversion value, CPA and ROAS. Individual channel metrics can be surfaced directly in the campaign list through custom columns — the mechanics are in our guide to custom columns and the report editor.
Option 3 — API and BigQuery
Asset-group detail exists only in the API. If you already run a daily export, add the channel segment to your existing PMax report and the rest is ordinary warehouse work — see exporting Google Ads to BigQuery. A warning for anyone with automated reporting: on the newer API version, the network field stops returning the old blended value. Code that hardcodes handling for that value will silently drop the breakdown. Check your mapping before the quarterly report runs, not after.
Reading the split without drawing false conclusions
This is where discipline matters. Having per-channel numbers invites conclusions the data does not support.
Trap 1: “Search delivers all the ROAS, so cut the rest”
The familiar picture: Search shows most conversions on a modest share of spend, while Display and Discover show the opposite. The urge to cut the “inefficient” channels is understandable, but last-click attribution systematically credits the bottom of the funnel. Video and display often do upper-funnel work that Search gets paid for. Before cutting, look at conversion paths and assisted impressions. At meaningful budget levels the honest answer comes from marketing mix modelling with Meridian or a geo experiment, not from a channel report.
Trap 2: “We can manage channels now”
Reporting is visibility, not a lever. There is still no control that says “send 40% to YouTube.” The indirect levers are few: asset composition (remove video and video inventory drops, though Google may still auto-generate video), the presence and quality of a product feed, placement and brand exclusions, and URL expansion settings. Each shifts probability; none assigns a share.
Trap 3: “Cheap channel equals good channel”
Discover and Gmail nearly always show low CPM and attractive last-click CPA on brand and remarketing audiences. That does not make them sources of new demand. Separate the channel that brings new customers from the channel that cheaply harvests people who were already going to buy. Customer type labelling helps keep those apart — see customer type labelling and lifecycle goals.
Trap 4: “PMax Search is a different thing from my Search campaign”
It is not. If PMax spends heavily in the Search channel while a Search campaign runs on the same queries, you are very likely paying twice for one piece of demand. That is exactly the situation covered in campaign cannibalization between PMax, Search and Shopping. Channel reporting turns that argument from theory into evidence.
Decisions the breakdown actually supports
| What the report shows | Sensible response | What not to do |
|---|---|---|
| Heavy Search spend alongside an active Search campaign | Resolve the overlap: brand exclusions, intent split, query review | Quietly raise budgets on both |
| Most impressions in Display with weak conversions | Check placement exclusions, creative quality, audience signals | Declare PMax “broken” and rebuild the campaign |
| Video inventory consuming budget on auto-generated video | Upload real video or rework the asset mix | Hope the automation produces a good video on its own |
| Almost all volume is product-data impressions | Work the feed: titles, images, custom labels | Adjust targets without touching the feed |
| Gmail and Discover show cheap conversions | Verify whether it is remarketing; separate new from existing customers | Scale a “cheap” channel as a new-demand source |
A practical move for retail accounts: overlay the channel split on your feed segmentation. It is common to discover that the same channel performs very differently across margin tiers, which turns “cut the channel” into “reallocate the catalogue.” Segmentation mechanics are in custom labels in your product feed.
A 40-minute PMax channel audit
A repeatable sequence that ends with a list of actions rather than a list of observations.
Minutes 1–10: build the distribution
Take a period of at least 30 days plus your conversion window. For each channel, write down three shares: spend, conversions, conversion value. Not CPA, not ROAS — shares, because they expose imbalance instantly. A channel taking 40% of spend and returning 10% of value is visible without a calculator.
Minutes 10–20: separate product from non-product
Apply the product data flag. If you run a feed, the product half almost certainly behaves differently from the rest. The levers then diverge: the product side is fixed through feed and assortment, the non-product side through assets, signals and landing pages. Blending both into one conclusion is how teams end up doing nothing useful.
Minutes 20–30: check the Search overlap
Compare PMax’s share of spend in the Search channel with what your Search campaigns spend on the same products. If PMax is spending materially in Search next to a campaign covering the same queries, the job is not “turn one off” — it is to separate intent: give brand to the controlled campaign and leave broad demand to the automation, or the reverse, depending on where your traffic quality control is stronger.
Minutes 30–40: look at video and write down actions
The video flag tells you whether you are buying video inventory and on which creative. A common finding: auto-generated video assets built from statics consume a visible share of impressions and convert poorly. The fix is not to remove video — it will be assembled anyway — but to upload something worth showing.
What you should end up with is not “PMax performs badly” but three or four concrete items: rewrite feed titles for a specific product group, separate brand demand, replace the auto-generated video, add exclusions for a specific placement category.
What a “normal” distribution looks like
The first question everyone asks is how much should go to YouTube — and there is no universal answer. The mix depends on three things.
| Factor | Effect on the distribution |
|---|---|
| Product feed present | With a feed, product impressions dominate and display shrinks |
| Asset composition | Plenty of good video means more video inventory; statics only means the system generates video itself |
| Brand awareness and search demand volume | With weak brand demand the Search channel has little to capture, so budget flows to reach inventory |
Which is why other people’s benchmark shares are conversation material, not management targets. The only meaningful comparison is your campaign against itself a month ago. A sharp shift in shares with no changes on your side almost always means an input changed: part of the feed dropped out, assets went through review, demand volume moved.
What to do about exclusions
Channel visibility usually exposes placement quality first. The existing tools still apply: placement exclusion lists, content and topic exclusions, and brand lists and brand exclusions for controlling brand traffic inside broad match and PMax. Broader inventory hygiene is covered in placement exclusions and brand safety.
The rule of thumb: exclusions fix “showing in the wrong place,” not “showing to the wrong people.” If a channel delivers traffic on relevant inventory that still does not convert, the problem is usually audience signals and landing experience, not the placement list. For the Search side of PMax, control is more precise through search themes in Performance Max.
What the channel report does not show
Know the limits before you build conclusions on the data.
- No audience breakdown inside a channel. You can see YouTube spent a given amount; you cannot see which audience signal drove it.
- No conventional assist view. A channel that prepares a conversion but never gets the last click looks weak in this split.
- No query detail for the Search channel. Spend share is visible; actual queries live in search terms and search themes reporting, and even there only partially.
- History starts in June 2025. Full-depth year-over-year comparison is not yet possible.
- Asset group level is API-only. Without an export pipeline, that granularity is closed to you.
Which gives you a working rule: the channel report answers “where did the money go,” not “why did it go there.” The second answer lives in the feed, the assets, the audience signals and URL expansion settings — and those are what you change.
When PMax eats your brand search
The most common finding from a first report: a meaningful share of spend in the Search channel while a separate brand campaign already covers that demand. Not a reason to panic, but not a reason to leave it either.
- Size it. PMax spend in Search plus brand campaign spend, against the total brand demand available. If you are covering the demand several times over, you are paying twice for the same impressions.
- Decide where brand demand should be served. A controlled campaign gives better control of copy, landing page and targets. PMax gives broader coverage across adjacent formats. Pick based on which matters more to you.
- Apply brand exclusions where it should not serve. This is the only direct lever; targets and budgets will not solve it.
- Measure after two weeks. Look at total brand spend and total brand conversions, not one campaign’s CPA. Frequently one campaign’s CPA worsens while the combined economics improve — that is the right outcome.
Also check whether URL expansion is pulling in low-quality search volume. With expansion on and landing pages chosen automatically, some of that search traffic may land on pages you would never have advertised.
Building it into routine work
Channel data is most valuable as a standing reporting layer, not as a one-off revelation. A workable cadence:
- Weekly — spend distribution by channel at campaign level; watch for sharp shifts in share.
- Every two weeks — CPA and ROAS by channel, adjusted for your conversion window; short windows lie, especially with long sales cycles.
- Monthly — asset-group breakdown via API, to see which group pulls which inventory.
- Before any change — snapshot the current distribution so that in two weeks you can tell your changes apart from auction movement.
Conversion lag deserves its own note: judging channels on yesterday’s numbers systematically underrates the upper funnel — the mechanics are in conversion windows and conversion lag.
Channel reporting does not make Performance Max a manual campaign. It makes it explainable — and that is enough to stop arguing with the automation blind and start changing the inputs that genuinely move it: feed, assets, signals and account structure.
How this changes the conversation with stakeholders
The practical value is often communication rather than optimisation. “Why is PMax more expensive than Search?” used to be answered with “different inventory.” Now you can show the split: this much spend went to the Search channel at this CPA, this much to video with this role in the funnel. A debate about trust becomes a discussion about structure.
A good report format is three numbers per channel: share of spend, share of conversions, share of value. From there it is immediately visible which channel overeats relative to contribution and which is undercredited by last click. Build the dashboard once and it works for months — the general approach is in our piece on Google Ads reporting without manual exports.
Where this is heading
The direction is clear: Google is handing back transparency where there used to be a black box, but not handing back control. Channel reporting, search themes, brand lists, exclusions — all of these are instruments of observation and constraint, not manual budget allocation. The way to work inside that logic is to make inputs as accurate as possible (feed, assets, signals, correctly valued conversions), make measurement as honest as possible — including first-party tag delivery through Google tag gateway — and keep checking that the automation has not started harvesting demand you already pay for somewhere else.
If your constraint is account access and launch capacity rather than analysis, that is a separate problem — see Google Ads agency accounts and the full PPC Rebels service list.
FAQ
Can I now turn off an individual channel in Performance Max?
No. Reporting shows distribution, but there is still no channel toggle in PMax. Influence is indirect: asset mix, feed, exclusions and URL expansion settings.
Why is there no asset group breakdown in the interface?
That level of detail is exposed through the API only. The web interface covers campaign level.
How far back does the historical data go?
Channel data is available from 1 June 2025 onward. Earlier periods cannot be split this way.
What does the product data flag mean?
It indicates the impression was built using Merchant Center product data, which lets you separate the shopping portion of a campaign from the rest.
Will the update break my automated reporting?
It can. The network field used to return a blended value and now returns specific ones. If your report code hardcodes the old value, the breakdown disappears without throwing an error. Verify the mapping before your next reporting cycle.
Which channel is usually the most expensive?
There is no universal answer — the mix depends heavily on vertical, feed presence and asset composition. Typical shares quoted from other people’s case studies are orientation, not a benchmark to manage against.
Should I split PMax into several campaigns to control channels?
Splitting for channel control rarely pays off: you lose learning volume and the channel mix still is not fixed. Split by catalogue, margin or geography instead.
How quickly will an asset change show up in the split?
Allow two weeks plus your conversion window. Earlier conclusions are almost always built on incomplete data.
What if almost all budget goes to Display?
Check creative quality and audience signals first, placement exclusions second. If conversions land within target, that may simply be normal for your vertical; if they do not, work the inputs rather than the budget.
Does this help settle cannibalization arguments with Search campaigns?
Yes, and that is arguably its most practical use. PMax’s share of spend in the Search channel is the first objective evidence of overlap you can put on the table.
Should I change bid strategy because of the new reporting?
Reporting alone is not a reason to change strategy. It becomes one if you discover conversion value is distributed differently than you assumed — in which case you fix the values, not the strategy.
How is this different from the placement report?
The placement report answers “on which sites and apps.” Channel reporting answers “in which Google product.” They are different layers and work best together.
