Performance Max is spending evenly. Your margins aren’t even.
Open most Performance Max campaigns and you’ll find one asset group, every product in it, and a single target ROAS sitting over the lot. On paper it’s tidy. In practice it means Google is being asked to hit the same return on a product that keeps 65% of its price as on one that keeps 20%.
Google will happily do that. It’ll find the products that convert most easily at that target and push them hardest, and those are very often your cheapest, lowest-margin lines. The ROAS holds. The revenue looks healthy. And a chunk of the spend is going on sales that leave almost nothing behind once the product, the shipping and the fees come out.
The fix isn’t another round of tweaking the target. It’s telling Google which products are worth more to you. Your margin data already lives in Shopify. Custom labels in Merchant Center are how you get it to Performance Max.
Why Shopify margin data and Merchant Center custom labels for this
Shopify has a Cost per item field on every product and variant, in the pricing section. Fill it in and Shopify’s Gross profit by product report shows you what each line actually makes. One thing to know: Shopify only reports profit for products that had a cost recorded at the time they were sold, so if costs have only just gone in, the history will be thin.
Google Merchant Center gives every product five custom labels, custom_label_0 to custom_label_4. They’re your own tags. Shoppers never see them, and Google says they’re for Performance Max, Shopping and Demand Gen campaigns. Each label can hold up to 1,000 unique values across the account, and each value can be up to 100 characters. You can only give a product one value per label.
In Performance Max, listing groups can be split by custom label, and Google’s own recommendation is to group items with custom labels and then target those labels in listing groups. So the chain is simple: margin in Shopify, a margin band in a custom label, and a campaign structure that bids differently on each band. I’ve written more about the feed side on my Google Merchant Center tool page.
Where I’d start
This is the order I’d work through it with any brand running Performance Max on a Shopify catalogue…
1. Get a cost on every product. Go through the catalogue and fill in Cost per item for every variant that’s live in your feed. No cost, no margin, no label. If you’ve got hundreds of products, Shopify’s product CSV has a Cost per item column you can fill in and import, which beats clicking through one product at a time.
2. Work out a margin you’d actually bid on. Price minus cost per item is your gross margin. I’d go a step further and take off an average for shipping, payment fees and packaging too, because that’s the money the ad spend has to come out of. Then sense-check the result against the Gross profit by product report so you know the numbers line up with what the store is actually selling.
3. Turn margins into bands, not numbers. Don’t put “43.7%” in a label. Put products into three or four bands, something like margin-high, margin-mid and margin-low, and give the products that are barely worth advertising a band of their own. Bands keep you well under the 1,000-value limit, and they give each campaign enough products and enough sales to learn from. I’d use custom_label_0 for margin and keep the other four free for things like bestsellers, seasonality or stock.
4. Get the labels into Merchant Center with a supplemental source. Be careful here. Shopify’s product CSV has Google Shopping custom label columns, but Shopify says its own Google & YouTube channel doesn’t use them. So don’t rely on them. Instead, in Merchant Center, turn on the Advanced data source management add-on (Settings, then Add-ons), go to Data sources, open the Supplemental sources tab and choose Add supplemental product data. The Google Sheets template is the easiest route. Each row needs the product ID exactly as it appears in your main feed, plus the same feed label and language, and then your custom_label_0 value. A supplemental source can only add to products that already exist in the feed. It can’t add new ones.
5. Add cost of goods sold while you’re in there. Put a cost_of_goods_sold column in the same sheet, with a number and a currency (for example 12.50 GBP). Google Ads uses it, along with conversions with cart data, to report cost of goods sold, gross profit and gross profit margin. That needs your purchase conversions to send the items in the basket, through the Google tag or Google Tag Manager, with item IDs that match Merchant Center exactly. It’s a reporting feature, not a bidding one, but it’s how you’ll see whether the change worked in profit, not just revenue.
6. Split Performance Max by margin band. In the campaign, subdivide the listing groups by custom label 0. A target ROAS is set at campaign level, so if you want each band bid differently, each band needs its own campaign: one for high margin, one for mid, and so on. Exclude the barely-worth-it band, or give it a small budget of its own with a high target so it only spends when it’s clearly profitable.
7. Set each target from its break-even, not from last month’s ROAS. Break-even ROAS is 1 divided by the margin left after costs. A band keeping 50% needs a ROAS of 2 just to stand still. A band keeping 25% needs 4. Set each campaign’s target above its own break-even, by however much profit you want each sale to leave behind. That’s the whole point: the high-margin products get room to scale, and the thin ones have to earn their place.
8. Give it time, then review the bands monthly. Expect a few weeks of the campaigns finding their feet before you judge them. After that, check once a month: costs change, prices change, and a product in the wrong band quietly undoes the work.
What to measure
- Gross profit after ad spend, by campaign. Gross profit from conversions with cart data minus the campaign’s spend. This is the number the whole exercise is meant to move.
- Share of spend by margin band. Before and after. If most of the budget still lands on the low band, your targets aren’t doing their job.
- Blended cost per order against Shopify. Total Google spend against actual orders in the store. Platform-reported conversions are a guide, not the truth.
- Products with no label. Anything in the feed without a margin band is being bid on blind. Check the count after every catalogue change.
The mistakes I see most
Labelling with margins that aren’t real. If cost per item is missing, out of date or a guess from three years ago, the labels are wrong and Google will bid on fiction. Get the costs right first.
Splitting too far. Ten bands across four campaigns sounds precise. It usually means each campaign has too few sales to learn from. Start with three or four bands and only split further when the volume is there.
Mismatched IDs. The supplemental sheet has to match the product ID, feed label and language in your main feed exactly. Get one wrong and the label simply doesn’t land, with no obvious error in the campaign.
Switching off low-margin products without asking why they sell. Some cheap products are how new customers find you, and they come back for the expensive ones. Check what first-order buyers of those products go on to buy before you cut them. My guide on optimising Meta ad spend on contribution margin covers the same thinking for Meta.
Setting the labels and forgetting them. New products launch without a band, prices change, suppliers put costs up. Build the monthly review into the calendar.
My verdict
The brands I’m working with right now are fighting the same battles: ad spend that’s climbed without the profit to match, no clear view of where they can safely scale, and not enough modern ecommerce know-how in the team to steer Google’s automation instead of being steered by it. Performance Max sits right in the middle of that. It’s very good at finding sales. It has no idea which of your sales are worth having unless you tell it, and custom labels built from your Shopify margins are the clearest way to tell it.
It isn’t the right choice for everyone. If you sell a handful of products with similar margins, there’s nothing much to split, and you’re better off putting the effort into the feed and the creative. If your Google spend is small, splitting it across several campaigns can leave each one too little to learn from, so start with one campaign that simply excludes the lowest band. And if you haven’t got reliable cost data, that’s the job to do first. No campaign structure can fix numbers you don’t have.
For a Shopify brand with a real catalogue, a real spread of margins and Performance Max doing most of the Google work, though, it’s where I’d start. Fill in the costs, band the products, get the labels into Merchant Center, and give each band the target it can afford. Let me know how you get on.
Related reading: How to Optimise Your Product Feed Using Custom Labels and How Most Ecommerce Brands Are Getting Their PMax Campaigns Horribly Wrong. More guides on this problem: Wasted ad spend.
More on this lever: The Traffic Lever: How To Optimise Traffic Without Paying More For Every Visitor.
Ian