For most of its life, Performance Max answered the question "where is my money going?" with a shrug. Spend reported as a single "Cross-network" blob, and you were left to guess how much of it landed on Search, Shopping, YouTube, Display, Discover, Gmail or Maps. During the 2026 transparency rollout that changed: Google added channel-level reporting to Performance Max, exposing the per-channel split at campaign, asset-group and asset level. One agency summed up the shift as Google finally showing what is behind Performance Max after years of the black box.
That split is the single most useful input to a Performance Max spend audit, because it tells you which levers can actually move the number. This post is not about the reporting mechanic itself — the API change that splits the Network field covers that, including the filters it silently breaks. Here the job is the audit: how to read the channel mix, how to judge it on efficiency rather than raw cost, and a decision tree for what each channel's share of the budget demands — because a lot of the time the answer is not "add negatives."
What channel-level reporting now shows
The channel view breaks a Performance Max campaign into the specific surfaces it served on, rather than collapsing them into one line. Depending on how Google has rolled the control into your interface, you will find it as a segment or grouping on the campaign's reporting or insights view, and the same breakdown is available at asset-group and asset level so you can see not just which channel took the spend but which creative drove it there. The channels you will see named are the familiar set — Search, Search Partners, Shopping, YouTube, Display, Discover, Gmail and Maps — each now reporting its own cost, impressions, clicks and conversions instead of hiding inside an aggregate.
Treat the exact menu location as a moving target. Google iterates this surface, labels shift, and a control that lives under "insights" one month moves to the report editor the next. If you pull data programmatically, the split arrives as separate rows per channel rather than one Cross-network row, which is a schema change your connectors and saved reports have to expect — the reason the companion post on the Network-field split matters. The important thing for the audit is simply that the data exists now: you can finally see the shape of a campaign's spend instead of inferring it.
Why spend distribution is the first question in a PMax audit
Spend distribution comes first because it bounds every other decision you are about to make. Performance Max negative keywords apply only to Search and Shopping inventory; they do nothing to YouTube, Display, Discover, Gmail or Maps. So before you spend an hour mining the search terms report and building negatives, the channel mix tells you the ceiling on what that hour can achieve. If a campaign runs 80% of its cost on display and video, query hygiene can move at most the other 20%, and the real waste — if there is any — is somewhere your negatives cannot reach. That is the coverage gap on non-search channels made measurable.
Reading distribution first also stops you misdiagnosing a campaign. A Performance Max campaign that looks like it is wasting money on irrelevant queries might simply be spending most of its budget on Display placements that were never query-driven in the first place. Without the channel split you would scroll the search terms report, find little to cut, and conclude the campaign is clean — while the actual leak sits on a surface the report barely describes. The split turns "this campaign feels inefficient" into a specific, located problem, which is the whole point of an audit.
Reading the channel mix: cost share next to efficiency
Open the channel breakdown and read two things together, never one alone: how the cost splits across channels, and how conversions and cost per conversion (or ROAS) split alongside it. The cost column tells you where the money goes; the conversion columns tell you whether that is a problem. A channel taking a large share of spend is only a finding if its efficiency is poor — few or no conversions, or a cost per conversion well above the campaign's blended figure. A channel taking a large share of spend and earning the best return is working exactly as intended, and cutting it would be self-harm dressed up as hygiene.
Two cautions keep this read honest. First, channels and the older network groupings do not map one-to-one, so if you are comparing against historical data reconcile the two schemes deliberately rather than assuming a clean join. Second, attribution inside Performance Max spreads credit across surfaces, so a video or display impression may have assisted a conversion that the last-click view hands to Search. Do not treat a low last-click conversion count on an upper-funnel channel as proof it is dead weight — look at assisted and view-through signals before you write it off. The goal is a fair read of marginal contribution, not a quick verdict from one column.
A decision tree: what each channel's share demands
Once you can see the mix, the audit becomes a routing decision — which tool fits which channel. When Search and Shopping carry most of the productive spend, you are in query-hygiene territory: mine the search terms report and add channel-level negatives, fence off cannibalised brand and converting terms, and tighten the list on the inventory your negatives can actually police. This is where an hour of search-term work pays, because the inventory is driven by what people type, and what people type is exactly what a negative keyword blocks.
When the cost concentrates on Display, YouTube, Discover or Gmail, negatives are the wrong tool and reaching for them just burns time. These surfaces are steered, not queried: the levers are asset-group structure, audience signals, placement and content exclusions where the campaign type allows them, and brand exclusions for irrelevant brand traffic. If a channel is both large and unproductive and no steering control reins it in, the honest move is structural — pull the high-intent query demand into a dedicated Search campaign that takes auction priority, or reconsider whether a single Performance Max campaign should own this mix at all. The 2026 steering controls are what you reach for here, not the negative list.
Judge on marginal return, not spend share
The trap in any distribution audit is treating the biggest bar as the problem. Size is not the verdict; marginal return is. The question to ask of each channel is not "how much does it spend?" but "what would I lose if I could pull the next dollar out of it?" A channel carrying half the budget at a strong ROAS is not waste, however large it looks; a channel carrying a tenth of the budget at zero conversions is waste, however small. Rank channels by efficiency and you will often find the real leak is not the headline number at all but a quieter surface that never earned its keep.
This matters more in Performance Max than in a hand-built account because you cannot simply pause a channel the way you would pause a campaign. You influence the mix indirectly, through signals and exclusions, and the algorithm pushes back toward whatever it thinks will convert. So the realistic output of this step is not "turn off Display" but a prioritised list: which channels earn their spend, which are candidates for steering pressure, and which justify a structural change. Feed that into your wasted-spend scorecard so a bad channel trips a threshold instead of hiding inside a campaign average.
Make it a standing audit, not a one-off
The channel mix is not static. Google reallocates Performance Max spend as it learns, your assets and audience signals change, and a budget increase can tip a campaign onto cheaper, lower-intent inventory overnight. A single read goes stale, so fold this into the cadence you already run — monthly for most accounts, tighter for high-spend ones, and always just after a campaign ramps or its budget moves. Log the per-channel share and efficiency each time so you are reading a trend, not a snapshot, and a channel swelling month over month shows up before it has eaten a reporting cycle of budget.
Attach it to your broader search hygiene audit so the channel read and the search-term read happen in one pass and inform each other: the distribution tells you how much the query work can move, and the query work cleans the slice the distribution says is reachable. Done together they stop you doing the common, satisfying, useless thing — polishing a negative list on a campaign whose money was never on Search to begin with.
Related reading
For the reporting mechanic behind this data and the filters it silently breaks, see the Sept 21 Network-field split. For why negatives only reach part of a PMax campaign, see the non-search coverage gap, and for turning per-channel findings into thresholds, see the wasted-spend metrics scorecard.