GOOGLE ADS PMAX CAMPAIGN ASSET GROUP SPEND REPORT: THE VIEW GOOGLE DOESN'T GIVE YOU BY DEFAULT

A PMax campaign has one budget and, in most accounts I inherit, four or five asset groups sharing it. Nobody decided that split. Nobody approved it. The algorithm made it, quietly, every day for months, and the interface opens on a summary card view that shows you thumbnails and Ad Strength instead of the one number that would tell you what actually happened to your money.

That's the gap. Not that the data doesn't exist it does, and it's free but that the default view is built around creative quality, and the question I have is never about creative quality first. My question is: where did the budget go, and did anyone intend it to go there?

I've opened accounts where four asset groups existed and one had taken 89% of the spend for a quarter. The other three were decorative. The client believed they were running four themes. They were running one theme with three expensive placeholders, and nothing in the interface had ever said so out loud.

This is the structural cost of the campaign type. In Search you get an ad account full of levers you can read search queries, prune with negative keywords, shift budget between ad groups by hand. PMax removed the levers and kept the budget. What it left you is a report, and the google ads pmax campaign asset group spend report is the closest thing to a budget lever you still have. So it's my first stop on any PMax audit, before assets, before signals, before I look at a single conversion number.

How I pull the PMax campaign asset group spend report

The whole thing is three clicks and one toggle most people never touch. Google documents the path plainly: go to Asset groups under the Campaigns menu, select the Performance Max campaign by name, and switch from summary view to table view using the table icon in the top right. Then use the Columns button to add metrics Google Ads Help on asset group reporting. That toggle is the entire secret. Summary is a creative gallery. Table is a spend report. Same data, same page, and the difference between them is whether you can see cost at all.

The columns I add, in this order

Cost first, always. Then impressions, clicks, conversions, conversion value. I add Ad Strength last and I keep it last on purpose, because if it sits on the left I start reading the table as a creative review instead of a budget review, and those are different jobs done on different days.

What I want is one row per asset group with a cost number on it. That's it. Everything else in this report is context for that number. Google's own recommended metrics for asset group performance lean conversion-first conversions, conversion value, conversion rate, cost per conversion and that's sound advice for evaluating creative. It's the wrong order for evaluating allocation, because conversion metrics tell you how well the money worked and say nothing about why that much money went there.

The column Google won't give you

There is no "share of campaign spend" column. So I download the table and add one myself: each asset group's cost divided by the campaign total. Percentages, not currency. It takes thirty seconds in a sheet and it changes what you see completely €4,100 and €480 read as two numbers, 84% and 10% read as a decision. If you'd rather not do it by hand every month, the same pull automates cleanly through a scheduled export into Sheets, and once it's there the percentage column is a formula that maintains itself.

The date range matters more than the columns

Last 7 days tells you almost nothing here a single asset group can dominate a week for reasons that evaporate. I pull 30 days minimum, and when I'm auditing I pull 90 and segment by month. Distribution that's stable across three months is a structural fact. Distribution that swings month to month is usually seasonality, or a feed that went in and out of stock, and I chase that instead.

When I go to the API instead

The interface is enough for one campaign. It stops being enough at fifteen, and for account-wide work I pull the asset_group resource through the API and build the share column in the query itself. Two caveats I've learned the hard way: PMax has no ad group entity at all, so anything you've built on ad group reporting returns nothing here and you have to rewrite it against asset groups. And API figures won't match the interface to the cent different processing stages, different dedupe timing. For distribution work the gap is irrelevant, because a share of 84% is 84% either way. Don't reconcile them line by line; you'll lose a morning proving something that was never designed to tie out.

What the spend distribution actually tells me

Here's the inference this report supports, and the one it doesn't. It does not tell you which asset group is best. It tells you which asset group the algorithm found easiest. Those get confused constantly, and the confusion is expensive in both directions.

The 80/20 that isn't a compliment

When one asset group holds most of the spend, my first hypothesis is never "it's the winner." It's that it has the widest addressable inventory the biggest product set, the loosest signal, the most generic landing page. Breadth wins budget in PMax. Quality wins budget too, but breadth wins it faster, and the cost column alone can't tell the two apart.

The tell is efficiency at scale. If the dominant group holds 84% of spend and its ROAS matches or beats the small ones, that's a real winner and the structure works. If it holds 84% and its ROAS is the worst in the table, breadth ate the budget and the algorithm did exactly what you asked while producing exactly what you didn't want.

Why the audience signal shows up here

An audience signal doesn't restrict who sees the ads it's a starting hint, and PMax serves outside it whenever it finds likely converters elsewhere. But a signal built on a broad customer list gives the model somewhere obvious to go on day one, and a signal built on a narrow custom segment gives it almost nothing. That asymmetry compounds. The group with the wide signal accumulates data faster, converts faster, and gets fed more budget, which produces more data. Six weeks later it holds most of the spend, and the report shows you a winner that was really just a head start.

The starved group

An asset group under roughly 5% of spend is, in practice, not running. It has too little data to be evaluated, too little volume to matter, and it's occupying structure and attention for nothing. But before I conclude anything I check whether starvation is the cause or the symptom thin assets, a narrow product filter, or a signal so tight it can't find anyone will all produce the same near-zero row, and each has a different fix. Ad Strength usually answers that in one glance, which is the one time that column earns its place.

The trap I'd warn about loudest

Google is explicit that asset groups with higher CPA or lower ROAS still contribute to campaign goals, and recommends against removing them on that basis alone. I'd extend it: the campaign is optimized as one unit toward one target, so a group carrying expensive marginal volume can be doing precisely its job. The efficient small group looks better because it's small. Kill it and you've killed the easy wins that were subsidizing the rest. That's marginal versus average ROI, and reading this report at average is how good accounts get worse. I hold campaign performance as the number that matters and asset group performance as the number that explains it invert that order and you'll optimize a table until the campaign gets worse.

Spend without conversions

A group with meaningful spend and near-zero conversions is the only row I treat as urgent. It's usually a broken final URL, a product set that's out of stock, or a theme that never matched intent. This is the finding this report exists to surface, and it's invisible at the campaign level because the healthy groups average it away.

What this report will never contain

It's worth being blunt about the ceiling, because people keep hoping. This is a cost table, not a diagnosis. It won't tell you which channel the spend went to, so if you want the Search-versus-YouTube-versus-Display picture you need the channel report, not this one. It won't give you placement reports at asset group level either brand safety work lives somewhere else entirely. And there's no search terms reporting inside it: PMax surfaces search themes and query insights through a separate report, aggregated in a way that would be unrecognisable to anyone used to Search campaigns, where you read the raw query and act on it the same afternoon.

How I use asset group spend reporting to restructure

Reading is the easy half. The reason I run google ads pmax campaign asset group spend reporting on a schedule is that it feeds exactly three decisions, and I try hard not to let it feed a fourth.

Decision one: split the campaign

When a group is both starved and profitable small share of spend, best ROAS in the table, consistently across 90 days that's not a restructure inside the campaign. Nothing you do to asset groups gives it more budget, because they don't have budgets. Only a campaign has one. So it becomes its own campaign with its own budget and its own target, and the moment it does you're managing budgets rather than hoping. That's the whole trade in how PMax asset groups are structured: grouping by theme buys you creative relevance and costs you spend control.

The caveat I always attach: a split restarts learning and fragments conversion volume across two budgets. If the profitable group is small because the segment is genuinely small, splitting gives it a budget it can never spend. I split when the ceiling is budget, not demand and the spend table alone won't tell you which, so read impression share next to it.

Decision two: merge the decoration

Three groups under 5% each, none with enough data to evaluate, all with overlapping themes that's not four asset groups, it's one asset group and a filing system. I merge them, pool the assets, and get one group with enough volume to produce a readable signal. Fewer, fatter asset groups beat more, thinner ones almost every time in accounts I manage.

Decision three: fix the inputs, not the structure

This is the most common outcome and the least satisfying one. Skewed spend is frequently a feed problem wearing a structure costume one asset group points at the product set with complete titles and images, the others point at the products the feed half-populated. Restructuring won't fix it; the same skew reappears in the new shape within a month. The feed is the targeting layer in PMax, so if the spend distribution mirrors the feed quality distribution, that's where the work is.

Anyone who ran search and shopping side by side before PMax existed will recognise the pattern instantly, because it's the old Shopping lesson with a new name on it: in a feed-driven campaign, the feed is the campaign. Shopping ads never let you pretend otherwise there were no headlines to hide behind. PMax gives you headlines, so people hide behind them, and then wonder why the reorganised asset groups produced the same table.

What I don't do

I don't pause an asset group for a bad CPA in a 30-day window. I don't chase Ad Strength as an outcome. And I don't restructure on a single pull I want the distribution to hold across two consecutive months before I touch anything, because PMax reallocates on its own and half the skews I've panicked about corrected themselves while I was writing the recommendation.

My cadence, and what I write down

Monthly, not weekly. This report answers a structural question and structural questions don't change in seven days. Weekly pulls made me a worse manager: I saw noise, I called it a trend, and I restructured a campaign that would have fixed itself. Monthly, the shape is real. It slots into the same rhythm as the rest of my optimization routine just on a slower loop than the things that genuinely move week to week.

The four numbers I log

Per asset group, per month: share of campaign spend, ROAS, conversion count, and one sentence explaining the share. That sentence is the whole artifact. "Group A holds 84% because it covers the full catalogue while B–D are seasonal subsets" is a structure that works as designed. "Group A holds 84% and nobody knows why" is a project.

Twelve of those rows is a year of distribution history, and it's the only thing that tells you whether a skew is drift or design. The sentence matters more than the numbers: in six months you'll remember none of the context, and the percentages alone will let you conclude something confident and wrong.

Where this sits in an audit

First. Before I view assets, before signals, before the conversion review. I want to know how the money split before I have any opinion about why, because opinions formed at campaign level are almost always about the dominant asset group wearing the campaign's name. When someone says "PMax is working," they mean one asset group is working. When they say "PMax is a black box," they usually mean they've never switched from summary view to table.

The honest summary: the google ads pmax campaign asset group spend report isn't hidden, it's just not the default, and defaults decide what people look at. One toggle, one cost column, one percentage you calculate yourself, and a campaign that was a black box becomes a budget allocation you can argue with. Most of the time you'll conclude the algorithm was right. The value is in the times you don't.