TL;DR: Performance Max bids on revenue unless you tell it otherwise. A 5x ROAS earns $40 of profit on a 60% margin product and $0 on a 20% margin product. Google says only 5% of advertisers send cost data. Add COGS to your feed, tag products by margin, and rebuild your targets around breakeven ROAS.
Why ROAS Targets Hide Profit in Performance Max
Performance Max optimizes toward the conversion value you give it. For most eCommerce accounts, that value is revenue. Revenue says nothing about what a sale actually earned.
So PMax treats a $100 sale of a low-margin product the same as a $100 sale of a high-margin one. The algorithm will happily push whichever one converts cheapest. Often that is the product that earns you the least.
Same ROAS, Very Different Profit
Take two products, each priced at $100. Product A costs $80 to make and ship, a 20% gross margin. Product B costs $40, a 60% gross margin.
Both run at a 5x ROAS. That means $20 of ad spend for every $100 sale.
- Product A: $100 revenue, minus $80 COGS, minus $20 ad spend, equals $0 profit.
- Product B: $100 revenue, minus $40 COGS, minus $20 ad spend, equals $40 profit.
Your Google Ads dashboard shows two identical 5x ROAS results. Your bank account shows one winner and one product that is working for free.
Breakeven ROAS Is 1 Divided by Margin
The math behind this is simple. Breakeven ROAS equals 1 divided by your gross margin. Below that number, every sale loses money once ad spend is counted.
- 20% margin: breakeven ROAS of 5.0x
- 30% margin: breakeven ROAS of 3.33x
- 40% margin: breakeven ROAS of 2.5x
- 50% margin: breakeven ROAS of 2.0x
- 60% margin: breakeven ROAS of 1.67x
One account-wide ROAS target can’t fit a catalog where margins run from 20% to 60%. It will overspend on thin-margin products and underspend on the ones that pay your bills. Plug your own margins into our ROAS Calc to find breakeven by product line.
How Performance Max Decides Where to Spend
PMax uses Smart Bidding across Search, Shopping, YouTube, Display, Gmail, and Discover. It chases whatever value signal you define. If the signal is revenue, it maximizes revenue.
Google’s own pitch reflects that focus. In February 2023, Google reported that advertisers using Performance Max see, on average, over 18% more conversions at a similar cost per action. That is a conversion metric. It is not a profit metric.
More conversions at a similar CPA can still mean less profit. That happens when the extra volume comes from your lowest-margin SKUs.
PMax also competes with your other campaigns. iClick’s 2026 benchmarks, built from accounts we manage, show PMax cannibalizing Standard Shopping in more than 80% of accounts. If you run both side by side, margin data decides which campaign should own which products. Our breakdown of Google Shopping vs Performance Max and which should run your product ads covers that split in detail.
Four Ways to Feed Margin Data Into PMax
There is no single profit switch. There are four levers, and most accounts need at least two of them working together.
1. COGS in Merchant Center Plus Conversions With Cart Data
This is Google’s native path. You add the cost_of_goods_sold attribute to your Merchant Center feed. Google Ads then works out margin per product using a simple formula: profit equals revenue minus COGS.
For this to work, your conversion tag must pass cart data. That means item IDs, prices, and quantities for every purchase. Google’s help documentation is strict here. Item IDs in your tag must exactly match the id attribute in your Merchant Center feed.
Once both pieces are live, Google Ads reports gross profit and COGS by campaign and product. Profit-based bidding builds on that same data.
Few advertisers have done this. In July 2024, Smarter Ecommerce (smec) wrote that only 5% of advertisers are ready for profit optimization because so few submit COGS data. That gap is an opening for anyone willing to do the setup work.
No precise COGS data? Google allows approximations. Its help center gives the example of estimating COGS as 80% of product price. Treat that as a stopgap. A flat estimate applied to every product makes every margin look the same, and that erases the very signal you are trying to send.
2. Margin-Tier Custom Labels
Merchant Center gives you five custom labels, custom_label_0 through custom_label_4. Each label can hold up to 1,000 unique values across your account.
Use one label for margin tier. Three tiers work for most catalogs:
- High margin: 50% or higher
- Mid margin: 30% to 49%
- Low margin: under 30%
Then split PMax by tier using listing groups. Each campaign gets a ROAS target set against its own breakeven. A high-margin campaign might run at 2.5x. A low-margin campaign might need 5.5x just to stay positive.
This approach works today, with no beta access and no COGS in the feed. The tradeoff is fragmentation. Every split divides your conversion data, and PMax learns slower on thin campaigns.
3. Product Value Optimization (Beta)
On September 14, 2026, Search Engine Roundtable reported a new Google Ads beta called Product Value Optimization. It works with Performance Max and Shopping campaigns.
Google describes it as a way to “create adjustments to optimize towards what matters, like profit, seasonal sell-through, or best-sellers.” You can apply value adjustments by product, brand, or category without changing your campaign structure.
That solves the biggest weakness of custom-label splits. You keep one consolidated campaign and its conversion data, while bidding still leans toward high-margin products. It also helps brands that don’t want to share exact cost data with Google.
It is a beta, so availability varies by account. If you have access, test it on one campaign against a holdout before rolling it out more widely.
4. Send Gross Profit as the Conversion Value
The most direct option skips revenue entirely. Your checkout passes gross profit, not order total, as the conversion value. Most brands do this through server-side tagging or a margin lookup inside Google Tag Manager.
Now tROAS becomes a profit target. A 1.5x target means $1.50 of gross profit per $1 of ad spend.
This is the cleanest signal and the easiest to get wrong. Your reported ROAS drops overnight. So does every revenue number in Google Ads. Finance and leadership need to know that before you launch.
Which Lever Fits Which Account
| Method | Setup effort | Keeps campaigns consolidated | Best for |
|---|---|---|---|
| COGS + cart data | Medium (feed + tag) | Yes | Brands with accurate per-SKU costs |
| Margin-tier custom labels | Low | No | Catalogs with clear margin bands |
| Product Value Optimization | Low (beta access) | Yes | Accounts already stable on tROAS |
| Profit as conversion value | High (dev work) | Yes | High-volume brands with tag control |
What Profit-First ROAS Looks Like by Category
Breakeven tells you the floor. Benchmarks tell you what a well-run account can reach above it.
These figures come from iClick’s 2026 benchmarks page. They are drawn from iClick-managed accounts in 2024 and 2025, not industry surveys.
- Google Shopping (well-managed): 480% ROAS
- eCommerce, Health & Beauty: 390% ROAS
- eCommerce, Fashion: 320% ROAS
- Music and Entertainment: 280% ROAS
Now compare those numbers to the breakeven table. A fashion brand at a 30% margin has a breakeven of 3.33x. A 320% ROAS sits just below that line. Revenue looks healthy while every order loses a few cents.
A health and beauty brand at a 50% margin breaks even at 2.0x. At 390% ROAS, it clears breakeven by nearly double. Same channel, same campaign type, very different economics.
For the full category and channel breakdown, see good ROAS for eCommerce benchmarks by category, margin, and channel.
What Scale Looks Like When Margins Drive Bids
Our named case studies show what happens when bidding follows real unit economics. Zager Guitars grew from $300K to $1.5M a month in revenue, 5x growth, at an 8.4 blended ROAS. Wish Rock Relaxation reached 10x ROAS on more than $80K a month in spend. Splendid Iris hit 6.2x ROAS on a $20K monthly budget and reached profitable scale in three months.
None of those brands got there with one blended target across the whole catalog.
A 30-Day Rollout Plan
Don’t switch everything at once. Smart Bidding needs stable signals, and a sudden value change resets what the algorithm thinks is working.
Days 1 to 7: Build the Margin Map
Pull COGS per SKU from your ERP, Shopify, or accounting system. Include shipping, payment fees, and average return cost if you have them. Calculate gross margin and breakeven ROAS for every product.
Days 8 to 14: Fix the Feed and the Tag
Add cost_of_goods_sold to Merchant Center through a supplemental feed. Assign margin tiers to a custom label. Confirm your purchase tag passes cart data with item IDs that match the feed exactly.
Check disapprovals while you’re in there. iClick’s benchmarks show a 12% feed disapproval rate in unmanaged accounts. A disapproved product sends no signal at all.
Days 15 to 21: Report Before You Bid
Run gross profit reporting for one full week before changing any bids. Find which products and asset groups earn money and which only earn revenue. This week usually changes the plan.
Days 22 to 30: Change One Bidding Lever
Pick a single method from the table above. Apply it to one campaign. Reset tROAS to match the new value definition. Hold everything else constant for at least two full conversion cycles.
Mistakes That Break Profit Bidding
Keeping a Revenue Target on Profit Values
If you switch conversion values to gross profit, your old target no longer means anything. At a 40% margin, a 480% revenue ROAS equals a 192% profit ROAS (4.8 times 0.4). Leave the target at 480% and PMax will choke spend almost immediately.
Using One Flat COGS Estimate
Google’s 80% approximation example is fine for turning on reports. For bidding, it tells PMax that every product has the same 20% margin. That is exactly the revenue-only problem you started with.
Mismatched Item IDs
If the IDs in your tag don’t match your feed, Google can’t join cart data to COGS. Profit metrics come back empty, and nobody notices for weeks. Test with a live order before launch.
Ignoring New Versus Returning Customers
Margin isn’t the only value signal. A first order from a new customer can be worth more than its gross profit suggests once repeat purchases arrive. PMax’s customer acquisition goal lets you bid higher for new customers. Layer it on only after your margin signal is stable.
Where Profit Bidding Fits in Your eCommerce PPC Structure
Margin data is one layer in a larger account structure. Feed quality, campaign segmentation, and channel mix all decide how much profit PMax can find.
Our eCommerce PPC playbook lays out that full system. For a look at how those decisions play out across real budgets, read inside the eCommerce PPC playbook and how we structure $61M of ad spend.
If you run a Shopify store, a DTC brand, or a health and wellness brand, margin structure varies a lot by model. Our eCommerce PPC industry hub breaks it down by vertical.
Find Out What Your PMax Campaigns Actually Earn
Most accounts we review have never compared ROAS to breakeven by product. Our Free Written Audit does exactly that. It is a 47-point written PDF covering feed health, margin signals, PMax and Shopping overlap, and bidding targets. You get it in five business days, with no sales call.
Related on iClick
Sources
- Google Ads Help: About conversions with cart data
- Google Ads Help: Provide cost of goods sold (COGS) feed attribute to report on profit margins
- Google Merchant Center Help: Custom label [custom_label_0-4]
- Google Blog (Tim Frank, Feb 23, 2023): New ways to multiply your results with Performance Max
- Smarter Ecommerce (Manuel Baudisch, July 3, 2024): Profit Optimization for Smart Bidding, Are you ready?
- Search Engine Roundtable (Sept 14, 2026): Google Ads Product Value Optimization Beta
- Google Business: Product Value Optimization
- iClick Advertising: 2026 PPC Benchmarks (iClick-managed accounts, 2024 to 2025)
- iClick Advertising: Case Studies
Frequently asked questions
What is Performance Max profit optimization?
It means making Performance Max bid toward gross profit instead of revenue. You feed Google margin data through COGS in Merchant Center, margin-tier custom labels, Product Value Optimization adjustments, or gross profit sent as the conversion value. Smart Bidding then favors products that earn more per sale, not just products that sell cheaply.
How do I calculate breakeven ROAS for PMax?
Divide 1 by your gross margin. A 40% margin product breaks even at 2.5x ROAS. A 20% margin product needs 5.0x. Below breakeven, each sale loses money once ad spend is counted. Calculate it per product or margin tier, because one account-wide number hides the products that lose money.
Do I need to share exact COGS data with Google?
No. Google accepts approximated COGS values, and margin-tier custom labels need no cost data at all. The Product Value Optimization beta, reported in September 2026, also lets you adjust product values without submitting exact costs. Approximations work for reporting, but accurate per-SKU costs give Smart Bidding a much stronger signal.
Should I split PMax campaigns by margin tier?
Only if each tier has enough conversion volume to learn on its own. Splitting lets you set a ROAS target against each tier’s breakeven, but it also divides your data. If you have access, Product Value Optimization keeps one campaign while still weighting high-margin products. For low-volume catalogs, consolidation usually wins.
