Let’s start with an uncomfortable question. If you stripped out revenue and looked only at what your Google Ads spend is actually generating in gross profit SKU by SKU, would the numbers still look as good?
For most e-commerce brands running Performance Max, the answer is no.
And not by a small margin.
PMAX is a black box with a mandate: find conversions at your target ROAS. It does exactly that. The problem is that it has no idea which of your products are profitable and which are quietly bleeding you dry. It optimises for the metric you gave it. It doesn’t care about the ones you didn’t.
The ROAS Trap
Here’s a scenario that plays out constantly across e-commerce accounts. A brand is running PMAX at a blended ROAS of 6x. Leadership is happy. The agency is happy. The monthly report looks good.
But dig into the data, and you find PMAX has fallen in love with a category of products, say, entry-level power tools, or accessories, or consumables, that carry a 12% margin. At 6x ROAS, that product isn't breaking even. It’s destroying value on every click.
Meanwhile, a range of premium SKUs with 48% margins is barely getting any impressions. They’re harder to convert, so PMAX deprioritises those products. The algorithm isn’t wrong.
It’s doing exactly what you told it to do. The brief was just broken from the start.
PMAX doesn’t know your margins. It knows what your ROAS target is. Those two things are not the same instruction.
Why PMAX Funds Worst-Margin SKUs
PMAX was built to maximise conversion value. Unless you’ve explicitly told it otherwise, it treats a £50 sale with a 10% margin identically to a £50 sale with a 45% margin. Same conversion value. Same signal. Same optimisation weight.
The algorithm will naturally gravitate toward products that convert easily and often, which tends to mean lower-priced, high-volume, frequently repurchased items. These are rarely your most profitable products. They’re your most predictable ones.
Add to this PMAX’s tendency to consolidate budget into whatever is working by its own definition, and you get a compounding effect: high-margin SKUs get starved of impression share, their conversion data thins out further, PMAX loses confidence in them, and the cycle continues.
The Break-Even ROAS Problem
A product with a 20% gross margin requires a 5x ROAS to break even on ad spend before paying for fulfilment, returns, or overhead. At a 15% margin, that break-even point is 6.7x. Running a blended account target of 4x means every low-margin SKU PMAX touches is a guaranteed loss.
How To Fix: The 5-Minute Diagnostic
You don’t need a full attribution overhaul to get a read on whether this is happening in your account. You need three data sources joined together.
Pull your PMAX product-level spend data
Go to your Shopping tab in Google Ads and filter by PMAX campaigns. Export product ID, spend, clicks, conversions, and conversion value for the last 60–90 days. You need enough volume to see a signal.
Join it to your COGS or margin data
Match product IDs to your margin feed, whether that lives in a spreadsheet, a Shopify export, or your ERP. You need gross margin percentage at SKU level. If you have contribution margin (net of fulfilment and returns), even better.
Calculate break-even ROAS per SKU
The formula is straightforward.
Break-even ROAS = 1 ÷ Gross Margin %
A SKU with a 25% gross margin requires a 4x ROAS just to break even before fulfilment, returns, and overhead are considered. Any PMAX spend on that product below 4x is a direct loss. Any blended account target below 4x and that SKU is a guaranteed drain.
Compare actual PMAX ROAS to break-even ROAS by SKU
For each product in your export, calculate the actual ROAS PMAX has been achieving. Compare it to the break-even threshold. The gap (positive or negative) tells you your real situation.
Plot spend against margin tier
Sort your SKUs into margin bands (under 20%, 20–35%, 35%+). Look at where PMAX has concentrated spend. You’re looking for the pattern: is budget disproportionately sitting in your lowest-margin band?
What you’ll typically find
The output is rarely pretty the first time. Here’s a representative pattern from an e-commerce account. The specific numbers vary, but the shape is consistent.
Entry-level accessories
- Gross margin: 14%
- Break-even ROAS: 7.1x
- Actual PMAX ROAS: 4.2x
- PMAX spend share: 38%
- Status: Loss-making
Mid-range tools
- Gross margin: 27%
- Break-even ROAS: 3.7x
- Actual PMAX ROAS: 4.8x
- PMAX spend share: 41%
- Status: Marginal
Premium / pro range
- Gross margin: 44%
- Break-even ROAS: 2.3x
- Actual PMAX ROAS: 3.1x
- PMAX spend share: 21%
- Status: Profitable
In this example, 38% of PMAX budget is actively destroying value. Another 41% is generating marginal returns. The only profitable segment with a break-even ROAS well below what PMAX is actually achieving is receiving barely a fifth of the spend.
The blended account ROAS? A respectable 4.4x. The board sees healthy ROAS. Profitability tells a different story.
What Advertisers Can Do About It
There are three levers, in order of difficulty and impact.
1. Fix your conversion value signals
If you can pass margin-adjusted conversion values back to Google Ads through a data feed or custom label mapping, PMAX starts optimising toward profit rather than revenue. This is the cleanest fix. It requires technical work but changes the fundamental brief you’re giving the algorithm.
At minimum, pass back values that reflect margin tiers rather than revenue: assign a conversion value multiplier by margin band. It’s a proxy, not a perfect signal, but it shifts the algorithm’s incentives.
2. Segment PMAX campaigns by margin tier
If value signal work isn’t viable right now, campaign segmentation is the blunter alternative. Separate your SKUs into at least two PMAX campaigns: high-margin products and everything else. Set different tROAS targets for each campaign that reflect the actual break-even points.
Your high-margin campaign can afford a lower tROAS target because the products are more profitable. Your low-margin campaign needs a higher tROAS target or shouldn’t exist as a PMAX campaign at all.
3. Use custom labels to exclude or suppress the worst offenders
In the immediate term, custom labels let you carve problem products out of PMAX and either exclude them entirely or move them into Standard Shopping where you have bid control. A product with a 12% margin and no hope of hitting break-even ROAS should not be in a PMAX campaign. That product likely shouldn’t sit inside a PMAX campaign at all.
Yes, this may cause your blended ROAS to dip. That’s the point. You’re removing loss-making spend. Revenue will fall; profit will improve. That’s a trade worth making every time.
When your blended ROAS improves after restructuring, something has gone wrong. That number should drop. Your profit shouldn’t.
The Conversation With Your Audience
This is where the analysis gets complicated. Showing that PMAX has been funding loss-making products for 18 months is not a comfortable presentation.
Frame it this way: the campaigns were performing exactly as instructed. The optimisation brief was wrong. ROAS is a ratio that says nothing about absolute profitability. The fix isn’t a campaign rebuild from scratch. It’s feeding the algorithm better information and giving it better guardrails.
The metric shift you’re proposing isn’t “we’re going to make ROAS worse.” It’s “we’re going to make your business more profitable.” Those are different things. Most boards, once they see the margin-adjusted numbers side by side with the revenue numbers, make the right call quickly.
Final Thoughts
PMAX isn’t broken. Your brief to it probably is. The campaigns are doing exactly what they’re designed to do: find conversions at scale. The problem is that not all conversions are created equal, and the algorithm has no way of knowing that unless you tell it.
This problem is getting worse with AI-driven bidding. As Google’s automation becomes more aggressive, the importance of feeding the algorithm the right commercial signals only increases. AI bidding systems are extraordinarily efficient at chasing the KPI they’re given, even when that KPI conflicts with business profitability.
The five-step diagnostic above takes an afternoon. What you find will either confirm that your campaigns are in good shape, or give you the data you need to make a compelling case for change. Either outcome is worth knowing.
Need a fresh perspective? Let’s talk.
At 360 OM, we specialise in helping businesses take their marketing efforts to the next level. Our team stays on top of industry trends, uses data-informed decisions to maximise your ROI, and provides full transparency through comprehensive reports.
.png)
