Performance Max Audit: What to Check, What to Cut, and What to Keep

Google wants you to run Performance Max. Their Customer Success Managers will tell you to put 25 to 30 percent of your budget into it. Their dashboard will recommend it every time you log in.

What none of them will tell you is where that money actually goes once it lands inside Performance Max (PMax for short, Google's all-in-one campaign type). The placement breakdown is hidden by default. In most accounts at the $50K monthly spend tier, 30 to 50 percent of PMax budget is bleeding into Display and YouTube placements that were built for retargeting, not for finding new customers. The patterns repeat across $150M+ in managed ad spend.

This post is the 6-section audit I run on every PMax campaign that lands on my desk. What to check. Where to find it inside Google Ads. What good and bad look like at each step. It pairs with the full 8-stage Google Ads audit method, but goes deeper on the PMax-specific places where Google hides the data on purpose.

Google itself reports that businesses earn $8 in profit for every $1 spent on Google Ads on average. Performance Max is the campaign type Google pushes hardest right now, and it is also the one where that ROI claim most often fails to land. This audit explains why, and what to check inside the campaign to find out where the budget is actually going.

Why Performance Max is over-allocated in most accounts

The reason Google's Customer Success Managers (CSMs for short) push you to put 25 to 30 percent of your budget into Performance Max is not that PMax works. It is that their key result is reallocating your spend onto Google's newest products. CSM incentives and founder outcomes are not the same thing. They never were.

Here is the part the agencies skip. Performance Max distributes your budget across six placement types: Search, Shopping, Display, YouTube, Discovery, Gmail, and Maps. Three of those (Search, Shopping, parts of Discovery) carry real commercial intent. The other three (Display, YouTube, most of Gmail) are retargeting inventory. They were built to rehammer a product to someone who already saw it on your site.

That is not an opinion. Read Google's own documentation on Performance Max and the placement list is right there. The documentation says PMax finds the best mix on its own. What it does not say is that the best mix, for Google's revenue, is the mix that fills its lower-demand retargeting inventory.

Putting acquisition spend on retargeting placements is purely stupid. The retargeting impression hits someone who already knows the brand. It does not find new buyers. It just rebills the spend onto a different placement.

Performance Max hides which is which by design. The placement-level data lives behind three clicks. Most founders have never seen the view. Most agencies do not pull it because the answer makes their PMax recommendation look bad.

The honest audit position is the opposite of the agency line. PMax should be a small additive layer on top of a healthy manual Search and Shopping account. Not the engine.

Section 1: The placement breakdown (the hidden report)

What to check: how much of your PMax budget is actually going to Search vs Shopping vs Display vs YouTube vs Discovery vs Gmail and Maps.

Where to find it inside Google Ads: Google hides this on purpose. The path is Campaigns → [Your PMax campaign] → Insights → Performance → Where audiences came from. Toggle the view to "By placement." If the view is empty, the campaign is too new (Google typically gates the data for the first 14 days).

A composite from a recent audit. A DTC home goods brand running $35K per month on Google, with 65 percent of that budget allocated to a single Performance Max campaign. Founder named Vikram. When the placement breakdown finally loaded, the distribution showed: 8 percent Search, 12 percent Shopping, 41 percent Display, 26 percent YouTube, 13 percent Discovery and Gmail. The account was acquiring almost nothing from Search (its highest-intent placement) while bleeding into retargeting inventory that was rehammering existing site visitors. Restructuring the Google budget to cut PMax to 20 percent of spend and rebuild manual Search and Shopping reclaimed roughly $8K to $12K of monthly recoverable spend in the first 90 days.

What good looks like: at least 60 percent of spend in Search and Shopping placements. Less than 30 percent combined in Display, YouTube, and Discovery.

What bad looks like: 50 percent or more of spend in Display and YouTube. The acquisition campaign is functioning as a retargeting campaign without you knowing.

What to do: if your distribution is bad, you have two options. Either cut PMax back to a supporting role (under 20 percent of total Google budget) and rebuild manual Search and Shopping, or restructure your asset groups to bias the algorithm toward Search and Shopping placements. The second option is harder and only partially effective because of how PMax is built. For context on whether your spend is even high enough for PMax to learn properly, see the cost breakdown by spend tier. PMax also does not expose Quality Score at the keyword level, which is one of the structural reasons Quality Score becomes less useful at scale.

Section 2: Asset group structure (PMax's campaign structure)

What to check: whether your asset groups mirror your product categories on the site, or whether they are a "one bucket for everything" setup.

Where to find it inside Google Ads: PMax campaign → Asset Groups tab. You should see one asset group per distinct product category or distinct audience signal. Google's own asset groups documentation walks through the basics.

What good looks like: 3 to 7 asset groups for a typical DTC catalog. Each one targets one product category or one audience signal. Clear thematic separation between groups.

What bad looks like: 1 asset group covering everything. Or 15+ asset groups with no clear logic, created as the brand tested its way through PMax over 12 months.

What to do: consolidate to one asset group per product category. Mirror the same structure as your manual Search and Shopping campaigns. Asset group sprawl is the PMax equivalent of campaign structure sprawl, which is one of the leaks I cover at stage 2 of the diagnostic order I run on every Google account. The principle carries over: structure controls the spend. Without structure, you have no levers to pull when something breaks.

The simplest version of the test. If you cannot answer the question "which asset group made us the most money this month, and at what return" in under 30 seconds inside Google Ads, the asset group structure is the next finding.

Section 3: Audience signals (the part PMax hides hardest)

What to check: which audience signals you have fed PMax, and whether those signals are non-overlapping.

Where to find it inside Google Ads: Asset Group → Audience signals. Look at custom segments, customer lists, and interest categories.

What good looks like: 2 to 4 distinct audience signals per asset group, non-overlapping, with clear thematic logic. Customer match lists weighted higher than broad interest categories.

What bad looks like: 10+ audience signals stacked on one asset group, with heavy overlap. Or zero audience signals (the account is hoping the algorithm figures it out alone, which means it will default to broad placements).

What to do: simplify signals to 2 to 4 per asset group. Prioritize first-party data (customer lists, website visitors) over broad interest categories.

This is the part of PMax that pushes back hardest against "feed it more signals" agency advice. More signals make the algorithm more confused, not better. The signal-stacking habit is the same mistake brands make on Meta retargeting, where overlapping audiences end up bidding against the same user across multiple ad sets.

Audience signals in PMax are a hint to the algorithm, not a hard targeting filter. The cleaner the hint, the closer the algorithm gets to the real buyer. The more cluttered the hint, the more PMax falls back on the default it likes most: cheap Display and YouTube impressions.

Section 4: Creative asset quality and variety

What to check: whether your PMax campaign has enough creative variety to support the spend, and whether the creatives are fresh.

Where to find it inside Google Ads: Asset Group → Assets tab. You should see headlines, descriptions, images, videos, logos.

What good looks like: at least 5 headlines, 3 descriptions, 5+ images at multiple aspect ratios, 1+ video per asset group. Assets refreshed quarterly.

What bad looks like: the minimum Google requires (3 headlines, 2 descriptions, 1 image, 1 logo) and nothing more. Or the same assets running for 12+ months.

What to do: build the creative assets out to at least Google's "Strong" rating per asset group. Refresh creatives quarterly to avoid frequency saturation on the placements where PMax bleeds budget (Display and YouTube).

Two patterns I see most often at this step. The first is the asset starvation account, which gives PMax the bare minimum and then blames the algorithm when results stall. The second is the asset graveyard account, which uploaded 20 assets at launch and has not touched them since. Both kill performance. The fix is the same: enough variety to give the algorithm room, refreshed often enough that frequency does not saturate.

If your account is under $5K per month total Google spend, skip the creative variety push. You do not have enough conversion volume for PMax to learn from the extra assets. Fix conversion tracking first. That is stage 1 of the Google Ads audit method.

Section 5: Conversion goal weighting

What to check: which conversion actions PMax is optimizing for, and whether they match the actual business priority.

Where to find it inside Google Ads: Campaign → Settings → Conversion goals.

What good looks like: 1 primary conversion goal (purchase, qualified lead) with value-based bidding. Secondary goals (add-to-cart, form-view) tracked but not optimized for.

What bad looks like: all conversion actions weighted equally. Or PMax optimizing for "lead" when the business cares about "qualified lead" or "purchase."

What to do: set the primary goal explicitly. De-prioritize secondary actions. For value-based bidding to work, your conversion values need to pass through correctly (a $200 sale recording as $200, not as $1). If you want a fast gut-check on what your break-even ROAS (return on ad spend) actually is at your margin and average order value, the break-even ROAS calculator runs the math in 30 seconds.

If the account's conversion tracking is fundamentally broken, fix that before touching PMax goal weights. The pattern I see most often: a brand reports 4x ROAS on PMax. Their P&L disagrees. The conversion event is firing on cart visits, not purchases. PMax has been optimizing for cart-clickers for months.

That is not a PMax problem. That is a data problem. PMax cannot be audited on broken conversion data, and goal weighting cannot fix what conversion tracking is reporting wrong.

Section 6: Search term insights (the visibility Google sometimes gives back)

What to check: which search queries triggered PMax ads, and whether any are clearly off-intent.

Where to find it inside Google Ads: Campaign → Insights → Search terms insights. This is the closest PMax gets to a real Search Terms Report.

What good looks like: most queries are on-intent for your product category. Brand vs non-brand queries clearly separated. You can identify the queries driving conversions.

What bad looks like: the report shows aggregated topic clusters ("home and garden," "online shopping") without specific queries. Most off-intent queries hidden under "Other."

What to do: PMax search insights are intentionally lower-resolution than a real Search Terms Report. Use what you can see. Add account-level negative keywords based on the visible off-intent queries.

Where the data is too aggregated to act on, that is itself a finding. The lack of resolution is a structural reason to keep PMax small and manual Search large. Manual Search gives you a full Search Terms Report. PMax gives you a topic cloud. If you want control over which queries you pay for, the platform forces a choice, and PMax is not the side of the choice that gives you control.

When to cut Performance Max entirely

Four conditional branches. Use whichever fits the account.

PMax does not provide brand defense either. The campaign type is built to find new buyers across Google's surfaces, not to hold Position 1 on your branded SERP against competitor and Amazon bids. For that, see the Self-Defense Campaign methodology.

Want this run on your account? Get a Free Quick Scan in 48 hours.

The closing argument: PMax should be a layer, not the engine

The honest budget allocation across a healthy Google account looks the opposite of what most CSMs recommend.

The placement-bleed pattern also shows up on the other half of the picture, because the Display and YouTube placements where PMax bleeds budget overlap heavily with the retargeting inventory most brands also run on Meta. If you are running both, you are often paying twice to hit the same warm user. The equivalent diagnostic for Meta accounts walks through the audience-overlap side of the same leak.

PMax is a layer. Not the engine. Treat it as the additive third campaign type on top of a healthy manual Search and Shopping account, and it earns its place. Treat it as the place to put 25 to 30 percent of your budget because Google's dashboard said so, and the placement breakdown will tell you exactly where that money went, six months later, when you finally find the report.