Performance Max for Ecommerce: Structure It Around Margins

Key Takeaways

• Performance Max scales whatever your conversion values reward, so two products with the same revenue and the same ROAS can leave ten times more contribution than each other.
• In one account we took on, 65% of the Performance Max budget was going to searches for the client's own brand. After brand exclusions, first time customers rose 52% the following month.
• Margin tiers start in Shopify's Cost per item field, which Shopify uses to calculate margin as price minus cost, divided by price.
• Because Google usually records the VAT inclusive order value, a UK product at 70% margin breaks even near 171% ROAS while one at 30% needs about 400%, so a single account target is arbitrary.
• Google suggests giving a retail Performance Max campaign at least six weeks of data before judging it, so changing targets weekly stops it ever settling.
• Custom labels only create value when they change analysis, budgets, bidding or structure. Labels nobody uses are just tidier data.

Quick Answer

Structure Performance Max around margins by getting cost data into Shopify, turning it into margin tiers, passing those tiers into Merchant Center as custom labels, and then using them to separate campaigns, listing groups and targets. Set each target from the break even ROAS of that margin tier rather than one account wide number, keep brand demand visible or excluded, and give every campaign enough conversions to learn.

Table of Contents

• What Performance Max Does With an Ecommerce Budget
• Why Bidding on Revenue Scales the Wrong Products
• Getting Margin Data Out of Shopify
• A Margin Tier Structure That Works
• Setting Target ROAS From Break Even
• Asset Groups, Audience Signals and Search Themes
• Brand Exclusions and Standard Shopping Alongside
• Value Rules and New Customer Goals
• Reading What Performance Max Reports
• Budgets and Learning Periods
• Common Mistakes We Fix
• Why Matters Perspective
• Frequently Asked Questions
• Final Thoughts
• Related Articles

Performance Max is often described as a black box, which is only half true. It decides bids, audiences, creative combinations and placements, but it does all of that in pursuit of the conversion values you send it. Feed it revenue and it optimises revenue, including revenue that leaves you almost nothing. This guide covers how to give it commercial information instead, using data your Shopify store already holds.

What Performance Max Does With an Ecommerce Budget

Performance Max is a goal based campaign type that serves across Google Search, YouTube, the Display Network, Discover, Gmail and Maps from one campaign, using your budget, conversion goals and values, bidding targets, Merchant Center feed, creative assets, audience signals, search themes and landing pages as inputs. Google's automation then decides the bids, the audiences, the creative combinations and the inventory it thinks will hit your objective.

For a retail account, the product feed is the most important of those inputs, because listing groups organise which Merchant Center products a campaign can advertise using product attributes. That is the hook this whole article hangs on: anything you can express as a product attribute, including margin, becomes something you can structure and bid around.

Why Bidding on Revenue Scales the Wrong Products

Take two products that each generate £500 of revenue from £100 of ad spend. Both report a 500% ROAS, and any bidding system told to maximise conversion value will treat them as equals. Now add margin: product A runs at 70% gross margin and product B at 25%. Product A produces £350 of gross profit and keeps £250 after the ad spend, while product B produces £125 and keeps £25. The same headline number hides a tenfold difference in what the business actually keeps.

Performance Max is not making a mistake here. It is doing exactly what it was told, because both products passed the same £500 conversion value into Google. The fix is not to distrust the automation but to change what it is optimising towards, and there are three separate ways to do that: structure products by their economics, send better conversion values, or report on profit so that human decisions about budget and targets are better informed.

Getting Margin Data Out of Shopify

Shopify already has the number you need. The Cost per item field holds what the product costs you, which Shopify defines as the amount paid to the manufacturer for a reseller, excluding taxes, shipping and other costs, or labour and materials for a maker. Shopify calculates margin from it as price minus cost, divided by price, so its own example of a product selling at 50 with a cost of 30 gives a 40% margin.

Two practical points make this work at scale. Cost per item lives at variant level, and Shopify says variants should carry their own cost where costs differ, which matters for size runs and bundles. And the product CSV includes the field, so you can export price and cost, calculate margin in a sheet, assign each product a tier and send that tier back. On £100 of revenue excluding VAT, a £30 cost is a 70% margin and a £70 cost is 30%, which is all the precision most accounts need.

Merchant Center also has a cost of goods sold attribute, which supports gross profit reporting, calculated as revenue minus cost of goods, where conversions carry basket data. It can be supplied through the feed, Content API or a supplemental feed. Treat the two as different jobs: cost of goods sold gives you profit visibility, while custom labels give you grouping you can build campaigns and bids around. Sophisticated accounts use both.

A Margin Tier Structure That Works

Google gives you five custom label fields and suggests uses such as seasonal products, clearance and selling rate, so margin is a natural extension. Google does not define what counts as high or low margin, so the thresholds below are ours and should be set from your own catalogue.

Keep product role separate from margin. A hero product might matter because it acquires customers, starts subscriptions, drives repeat orders, anchors the brand, sits on deep stock or simply converts unusually well, and none of that is a margin percentage. We usually run one label for margin tier and a second for role, with values such as hero, standard and long tail, so the two can be combined without either being distorted.

Then choose the right tool for the job, because campaigns, listing groups and asset groups solve different problems.

Setting Target ROAS From Break Even

This is where margin tiers earn their keep. The standard UK VAT rate is 20%, so a product selling at £120 including VAT is £100 of revenue to the business, but Google usually records the £120 checkout value as the conversion value. That gap alone shifts every target, and it is why a target copied from a case study is rarely right for your account.

Work it through on that £120 order. At 70% margin, the cost is £30 and gross profit is £70, so break even against the reported £120 is about 171%. At 50% margin, £50 of gross profit means 240%, and at 30% margin, £30 of gross profit means 400%. Break even is not a target though, because it leaves nothing, so we usually allow advertising to consume no more than 80% of gross profit before ad spend, which is our rule of thumb rather than a Google recommendation.

Two caveats matter. These figures use gross margin only, so delivery subsidy, fulfilment, packaging, payment fees and returns all need adding before you set a real target. And the arithmetic assumes Google is recording the VAT inclusive value, which is worth confirming rather than assuming, because sending ex VAT values changes every number in the table.

Asset Groups, Audience Signals and Search Themes

An asset group is a set of creative built around a theme or audience, holding headlines, descriptions, images, logos, videos and final URLs that Google combines into ads for different inventory. Google recommends organising them around coherent products, categories, themes or audiences, and the important structural point is that asset groups have no budget of their own, so they shape messaging rather than spend. Commercial segmentation and creative segmentation do not have to match.

Audience signals are suggestions, not targeting. They tell Google which audiences you believe are valuable, drawing on customer lists, website visitors, app users, video viewers, custom segments and demographics, and Google will still serve beyond them where it predicts conversions. Search themes work the same way for queries: optional, additive signals rather than keywords, currently capped at 50 unique themes per asset group. Use them where you know something the feed and website do not say clearly, such as "waterproof hiking boots", "vegan protein powder" or "monthly coffee subscription".

Brand Exclusions and Standard Shopping Alongside

Performance Max will advertise against searches for your own brand unless you stop it, and the effect can be larger than most merchants expect. In one account we took on, 65% of the Performance Max budget was going to brand searches before we were involved. After brand exclusions were applied, first time customers rose 52% the following month, because the same money went to demand the brand was not already winning. That is one account rather than a benchmark, but it shows what a headline ROAS can hide.

Google's detailed brand exclusion controls cover how Shopping traffic is treated on excluded brand searches, and for Performance Max the exclusions apply to the relevant Search and Shopping inventory. We do not exclude brand in every account, because sometimes defending brand terms is the right call, but every account should at least be able to see branded and non branded performance separately.

Standard Shopping can run alongside Performance Max, and Google lets Ad Rank decide which campaign serves where both target the same products. Creating a Standard Shopping campaign therefore gives it no automatic priority, so if you want hero products genuinely ringfenced, remove them from the Performance Max listing groups. If you want to know which approach performs better rather than argue about it, Google provides experiments for comparing the two.

Value Rules and New Customer Goals

It is tempting to assume conversion value rules can multiply value by margin label. They cannot. Google documents their conditions as audience, location and device, so there is no native rule saying "if custom label 0 is high margin, multiply by 1.5". What they can do is adjust values that Smart Bidding then uses, including with target ROAS, maximise conversion value and Performance Max. Margin belongs in labels, structure, cost of goods reporting or the conversion values themselves.

Customer type is a different question, and here Google does have a native answer. Customer lifecycle goals let you bid higher for new customers, higher again for high value new customers, or only for new customers, with retention modes for lapsed ones. A £30 repeat order from an existing customer and a £30 first order from a new one are worth very different amounts over a year, and this is the setting that lets bidding act on the difference.

Reading What Performance Max Reports

Performance Max reports more than its reputation suggests. Channel performance reporting shows how a campaign delivers across Search, Display, YouTube, Discover, Gmail, Maps and Search partners, which is the fastest way to understand where budget is going. Resist the instinct to read a channel with high spend as one that is stealing budget, because the campaign is optimising as a whole.

Search terms insights group demand into categories and subcategories across Search, Shopping and Performance Max, which is where branded demand, new product themes, competitor intent, irrelevant themes and seasonality show up. Product reporting was expanded in June 2026 so that product metrics are available across all Performance Max networks, not just the narrower view available before, which makes product level profit analysis far more practical. Placement reporting shows where ads appeared, but Google is explicit that it is not a complete performance report and is primarily a brand safety tool, so use it for that and nothing else.

Budgets and Learning Periods

Google publishes no universal minimum budget, and it could not sensibly do so when one merchant sells £20 consumables in a cheap auction and another sells £4,000 furniture. The useful test is whether each campaign gets enough conversions to learn. Split £3,000 a month across ten campaigns and each gets £300, which is roughly £10 a day, and precision in the structure becomes pointless when no campaign has enough data. Where volume is thin, put high and medium margin products together and keep the labels for reporting.

Google suggests allowing a retail Performance Max campaign at least six weeks of data before judging its performance. It separately advises against frequent or large changes to budget, bid strategy, targets or targeting while automated bidding is still optimising, and suggests allowing one to two conversion cycles after any significant change. It also recommends giving asset groups around two to three weeks before judging asset performance. Moving a target from 400% to 350%, then to 500%, then back to 400% is not management, it is three restarts, so change targets when the economics or the evidence justify it rather than because a week looked weak.

Common Mistakes We Fix

Seven problems account for most of the underperformance we see in ecommerce Performance Max accounts:

1. One asset group for everything. Skincare, haircare, supplements and candles sharing a single creative theme gives Google nothing to work with.
2. The same value logic for every product. A 75% margin line and a 20% margin line passing identical revenue values will be scaled identically.
3. No margin data at all. Without Cost per item in Shopify, product level profit analysis cannot start.
4. Brand demand hidden in the headline. An 800% ROAS that is mostly brand searches is not an 800% return on new demand.
5. Segmentation past the point of data. Eight campaigns at £10 a day each is worse than three campaigns that can learn.
6. Targets that change constantly. Every significant change restarts learning, as Google's own guidance says.
7. Labels nobody uses. Custom labels that never change analysis, budgets, bidding or structure are decoration.

Why Matters Perspective

We describe our PPC work as managed for profit, and this article is what that phrase actually means. It is not a claim that we can see your profit and loss account inside Google Ads, because nobody can. It means giving the advertising system better commercial inputs, then judging the outputs against the economics of the business rather than against a ROAS number with no context.

The brand exclusion case above is the clearest example. A headline return looked strong, 65% of the budget was buying customers the brand already had, and fixing that produced 52% more first time customers the following month without a penny more spend. Nothing in Google Ads was broken. The account was scaling exactly what it had been asked to scale.

An ecommerce business keeps money only after VAT, product cost, fulfilment, payment fees, returns and advertising, so a £120 order is never £120 of value. Shopify holds the cost data, Merchant Center holds the labels and cost of goods attribute, and Google Ads holds listing groups, targets, lifecycle goals and product reporting. None of those features is a profit strategy on its own. If you would like us to look at how your account is structured, our free Google Ads account review covers the feed, the structure and the targets, and our PPC work is scoped to the account, billed a month in arrears and not tied to long contracts.

Frequently Asked Questions

Should ecommerce stores use Performance Max?

Usually yes, provided conversion tracking is reliable, Merchant Center data is healthy and you know which products you want scaled. Performance Max reaches Search, YouTube, Display, Discover, Gmail and Maps from one campaign, but it scales whatever your conversion values tell it to scale, so the inputs decide whether that is profitable.

How many asset groups should a Performance Max campaign have?

There is no correct number. Google advises organising asset groups around coherent products, categories, themes or audiences, so the right count is however many genuinely need different messaging, imagery or landing pages. One asset group covering unrelated ranges is the version that consistently underperforms.

Should I split Performance Max campaigns by margin?

Split when margin tiers genuinely need different budgets or different targets and each campaign will still get enough conversions to learn. Where volume is limited, keep the products together and use custom labels and listing groups instead, so you keep the reporting without starving each campaign of data.

Does Performance Max bid on my own brand name?

It can, unless you apply brand exclusions. In one account we took on, 65% of the Performance Max budget was going to searches for the client's own brand. After brand exclusions were applied, first time customers rose 52% the following month, because the same budget went to demand the brand was not already winning.

Can I run Standard Shopping and Performance Max together?

Yes. Google allows both to target the same products and lets Ad Rank decide which serves, so creating a Standard Shopping campaign does not automatically give it priority. If you want products genuinely ringfenced in Standard Shopping, exclude them from the Performance Max listing groups.

What budget does Performance Max need to work?

Google publishes no universal minimum, because a store selling £20 consumables and one selling £4,000 furniture face different auctions. The practical test is whether each campaign gets enough conversions to learn: £3,000 a month split across ten campaigns is £300 each, which is rarely enough.

How do I get product margins from Shopify into Google Ads?

Fill in Cost per item on every product and variant, export price and cost, calculate the margin, assign each product a tier, then pass that tier into Merchant Center as a custom label using a feed app or supplemental data source. Google Ads has no margin data of its own until you send it.

Final Thoughts

Performance Max rewards the quality of what you feed it. Put cost data in Shopify, turn it into tiers, send the tiers to Merchant Center, and set targets that reflect what each tier can actually afford. Then leave the campaigns alone long enough to learn.

Start with fewer, better funded groups rather than a precise structure nothing can learn from, and judge a ROAS by whether it is profitable for the products that produced it. If you would like a second opinion before you restructure, ask for a free Google Ads account review.

Related Articles

• Google Shopping for Shopify: Setup and Feed Optimisation
• Google Ads Audit Checklist for Ecommerce Stores
• How Much Does a PPC Agency Cost in the UK?
• Why Your Shopify Conversion Rate Matters and How to Measure It
• How to Prepare Your Shopify Store for Black Friday
• The 20 Best Shopify Apps for Every Job Your Store Needs
• Your First 90 Days: A Marketing Plan for New Shopify Stores
• Google Merchant Centre for Shopify: Setup, Disapprovals and Fixes