Performance Max can scale an e-commerce account quickly, but it can also make an account look more profitable than it really is if you don't understand where the conversions and revenue are coming from.
The biggest mistake is treating Performance Max as a campaign you simply launch, give a ROAS target, and leave alone. PMax uses Google's automation across Search, Shopping, YouTube, Display, Discover, Gmail, and other eligible inventory, so the quality of the information you give it matters enormously.
For e-commerce advertisers, that information starts before the campaign is even created: your Merchant Center product feed.
Product titles, identifiers, product attributes, images, prices, availability, and custom labels give Google the information it needs to understand and organize your products. Google explicitly states that product data shapes how ads and free listings behave and perform.
This guide breaks down how to structure PMax for scalable e-commerce growth, separate products intelligently, feed useful audience signals, control branded traffic, and decide when creative assets should be added instead of relying primarily on product-feed-based advertising.
1. Your Product Feed Is the Foundation of PMax
Before optimizing a Performance Max campaign, optimize what the campaign is actually advertising.
For an e-commerce business, that means looking closely at Merchant Center.
A useful way to think about the relationship is:
If the first part is weak, the rest of the system has less useful information to work with.
Google recommends using important product details in titles, including information such as brand, product name, color, size, gender, and other relevant attributes. It also recommends putting important information toward the beginning because product titles can be truncated in ad formats.
Product titles
Compare:
- Weak: Premium Running Shoes
- More useful: Nike Air Zoom Pegasus 41 Men's Running Shoes Black
The second title gives Google's product matching systems substantially more information.
Your title should describe what the product actually is, using the attributes shoppers are likely to search for. Don't turn it into keyword stuffing.
GTINs and product identifiers
GTINs help identify products and variants. Google recommends submitting accurate product identifiers where they exist, alongside attributes such as brand and MPN where applicable. Products without legitimate identifiers should not be given invented GTINs.
This is particularly important for retailers selling established branded products.
If your feed contains incorrect identifiers, you're not improving the campaign—you are feeding incorrect product information into the system.
Custom labels
Custom labels are different. They aren't primarily about helping Google understand what the product is. They are extremely useful for your campaign organization and bidding strategy.
For example:
| Custom Label | Example |
|---|---|
| profit_tier | High / Medium / Low |
| margin | 60+ / 40-59 / Under 40 |
| season | Summer / Winter / Evergreen |
| stock | Overstock / Normal / Low |
| priority | Hero / Core / Test |
| sale_status | Sale / Full Price |
This lets you create meaningful product groups instead of treating every SKU equally.
2. Structure Asset Groups Around Meaningful Product Differences
One giant asset group containing every product can become difficult to control. But creating dozens of asset groups simply because your store has dozens of categories is also unnecessary.
Google recommends organizing asset groups around meaningful themes, products, services, or audiences. It also recommends keeping campaign structures consolidated where possible unless different budgets, ROAS targets, countries, languages, or business objectives justify separate campaigns.
A practical structure could look like:
Campaign: PMax – UK Ecommerce
- Asset Group: Running Shoes
- Asset Group: Hiking Shoes
- Asset Group: High-Margin Products
- Asset Group: Seasonal Collection
But don't create separate campaigns for every category unless the business needs independent budget or bidding control.
Profit-margin segmentation
This is where custom labels become powerful.
Imagine your store sells:
- Product A: £100 revenue / £20 profit
- Product B: £100 revenue / £50 profit
- Product C: £100 revenue / £70 profit
If you optimize purely around revenue, all three can look identical. But commercially, they aren't.
You could use:
Then use listing groups and campaign structures to analyze how those products behave.
This doesn't mean automatically excluding every low-margin product. Some low-margin products may have excellent conversion rates or generate repeat customers. The point is to make business economics visible inside the advertising structure.
3. Don't Give PMax Random Audience Signals
Performance Max doesn't require you to manually select a traditional audience in the same way as older campaign types.
Audience signals are suggestions that help Google's AI understand potential customers. Google specifically says these signals can include first-party data such as customer lists, website visitors, app users, and video viewers, as well as custom segments based on keywords, URLs, or apps.
Think of them as: “Here is useful information about the people who tend to buy from us.”
Not: “Only show ads to these people.”
Performance Max can still find people outside your supplied signals if Google's systems believe they are likely to convert.
First-party data
For an established store, useful signals might include:
- Previous purchasers
- High-value customers
- Website visitors
- Cart abandoners
- Product viewers
- Email subscribers
- App users
- Previous converters
A particularly useful distinction is between all customers and valuable customers.
If you have enough data, a list of customers who purchased £500+ over their lifetime can potentially provide a more useful signal than simply uploading everyone who ever visited the website.
High-intent custom segments
You can also provide custom audience signals based on relevant search terms, URLs, or apps.
For a premium running shoe store, for example:
Keywords:
URLs:
- Competitor product pages
- Running publications
- Relevant product comparison websites
Don't turn this into a huge list of unrelated interests. A smaller set of genuinely high-intent signals is generally more useful than throwing every imaginable keyword into the audience signal.
4. Control Brand Traffic Before Celebrating ROAS
This is one of the most important PMax checks for e-commerce.
Suppose your campaign reports:
- £20,000 revenue
- £4,000 spend
That's a 500% ROAS. It looks excellent.
But now imagine £12,000 of that revenue came from people searching for your own brand name.
Your PMax campaign may be receiving credit for customers who already knew the company and were searching specifically for it. That doesn't make the sales worthless. It changes how you should interpret the campaign's incremental acquisition performance.
Use brand exclusions when appropriate
Google now provides campaign-level brand exclusions for Performance Max. These can prevent PMax from serving for selected brand searches, including variants and misspellings, and Google recommends brand exclusions rather than relying on negative keywords specifically for excluding brand traffic.
For example, suppose your brand is ExampleStore. You might create a brand list containing:
- ExampleStore
- Example Store
- Common brand variants
- Relevant subsidiary brands
Then apply that brand exclusion to the PMax campaign.
Google notes that for retail advertisers with product feeds, there is also an option to continue allowing Shopping ads on searches mentioning excluded brands while excluding other relevant inventory, depending on the campaign setup.
Don't blindly exclude every brand
This is important. Brand exclusion is a strategic decision, not a mandatory optimization step.
If branded Shopping traffic is highly valuable to your business, removing it may reduce useful conversion opportunities. Google itself recommends considering the impact of exclusions on reach and performance rather than applying them universally.
The right question is:
“Do I want PMax to capture this branded demand, or do I want to measure and manage branded demand separately?”
That is a measurement and business-strategy decision.
5. Feed-Only vs Full Creative Asset Groups
This is where many e-commerce advertisers get confused.
Performance Max can work with a Merchant Center feed, and Google states that when a Merchant Center feed is attached, assets can be automatically generated from the feed. Additional headlines, descriptions, images, and other assets are still recommended to help Google serve across applicable surfaces.
So there isn't a universal rule that says “Never upload creative,” nor is there a rule that says “Always upload every possible creative asset.” The decision should depend on the campaign objective and the quality of your creative.
When a feed-led setup can make sense
A product-feed-led approach can be useful when:
- Shopping is the primary sales channel
- Product images are strong
- Product titles are well optimized
- The landing pages are highly relevant
- You want to test product demand before investing heavily in creative
- You don't yet have strong video assets
But don't interpret “feed-only” as “Display is completely off.” Performance Max is designed to operate across multiple Google surfaces. If you want strong control over where ads can appear, you need to understand the available campaign and account-level inventory and suitability controls.
When full creative assets make sense
Add strong creative when your product benefits from demonstration, lifestyle imagery, brand storytelling, or visual differentiation.
Examples:
- Fashion: lifestyle photography
- Cosmetics: product demonstration
- Fitness equipment: product-in-use video
- Furniture: room-context photography
- Consumer electronics: feature demonstration
A good video isn't automatically a junk placement generator. A poor video or irrelevant creative can simply give automation more opportunities to distribute weak messaging.
Google recommends covering asset types and providing quality, varied assets in asset groups. The practical rule is: Don't avoid creative because you fear Display. Improve your creative and monitor where your budget and conversions are actually coming from.
Best when you lack custom video/lifestyle assets and want to validate direct SKU demand first:
- Shopping is the primary revenue driver
- Clean product images & titles carry the click
- Testing baseline product conversion rates
Best when your products thrive on visual demonstration, lifestyle context, and multi-channel reach:
- High-converting product-in-use videos
- Lifestyle photography & tailored headlines
- Scaling across YouTube, Discover & Search
6. Don't Optimize PMax From ROAS Alone
Performance Max can report impressive top-line numbers while hiding important differences between products, channels, customers, and search behavior.
Start looking at:
- Conversion value
- ROAS
- Cost
- Conversion volume
- New vs returning customers
- Product-level performance
- Search terms
- Landing pages
- Asset performance
- Audience insights
- Profit margin
Google's current PMax search-term reporting provides search terms and corresponding landing pages, but Google recommends prioritizing conversion data rather than judging PMax primarily through clicks or impressions.
This matters because PMax isn't simply a Search campaign with a different interface. It can generate conversions across different inventory types, so CTR alone doesn't tell you whether the campaign is commercially healthy.
A simple product profitability example
Suppose PMax generates:
| Product Group | Revenue | Ad Spend | ROAS | Gross Margin | Gross Profit | Est. Contribution Profit |
|---|---|---|---|---|---|---|
| High Margin | £30,000 | £6,000 | 5.0x | 65% | £19,500 | +£13,500 Net |
| Medium Margin | £40,000 | £10,000 | 4.0x | 40% | £16,000 | +£6,000 Net |
| Low Margin | £50,000 | £16,000 | 3.1x | 15% | £7,500 | -£8,500 Loss |
If you look only at revenue, the low-margin group appears strongest. If you look at the economics, the decision becomes more complicated. This is why custom labels based on margin, product type, stock status, or strategic importance can be valuable.
7. A Practical PMax Scaling Framework
Don't scale simply because ROAS looks good for seven days. Use a structured process.
Step 1: Fix the feed
Audit:
- Product titles
- Descriptions
- GTINs
- Brand
- MPN
- Images
- Price
- Availability
- Product variants
- Shipping information
- Custom labels
Google recommends keeping product information accurate and up to date, including price and availability.
Step 2: Segment meaningful products
Create labels for:
- Margin
- Season
- Bestseller status
- Stock level
- Product category
- Promotional status
Then use listing groups and asset groups where meaningful differences actually exist.
Step 3: Add useful audience signals
Start with:
- Customer Match / customer lists
- Website visitors
- Purchasers
- High-value customers
- Relevant custom segments
Don't confuse signals with strict targeting.
Step 4: Audit branded traffic
Check how much revenue comes from branded searches.
If you have a separate brand Search strategy and want PMax to focus more heavily on non-brand demand, test the appropriate brand exclusion setup.
Step 5: Evaluate product economics
Don't ask only: “Which product has the highest ROAS?”
Ask: “Which products generate the most valuable contribution after advertising costs?”
Step 6: Scale gradually
Once the campaign has reliable conversion data and acceptable economics, increase investment carefully and monitor whether efficiency changes.
A larger budget can expose the campaign to additional opportunities, but it can also force the system into less efficient auctions.
Scaling means finding additional profitable demand, not simply spending more money.
Conclusion
Performance Max is powerful because Google handles a significant amount of the targeting, bidding, and placement decisions.
That does not mean advertisers should surrender control. Your control moves upstream.
You control the quality of the product feed, the way products are classified, the business economics you expose through custom labels, the audience signals you provide, the brand traffic you allow, the creative assets you supply, and the conversion data you optimize toward.
The most important mindset shift is this:
Don't try to manually control every auction. Build a better information system for the automation to work with.
A well-structured PMax campaign isn't simply a collection of products with a ROAS target. It is a system where product data, customer data, creative, business margins, and conversion signals work together.
When you look at your current e-commerce campaign, is your biggest opportunity in the product feed, campaign structure, audience signals, or controlling branded traffic?