Google Shopping & Performance Max: The Feed Decides

With Google Shopping and Performance Max, the biggest lever is rarely the campaign structure – it’s the product data feed. Get sloppy here and you give away visibility, no matter how carefully the campaign setup is otherwise put together.

Key takeaways

• The feed is to Shopping and PMax what keywords are to search campaigns – the actual targeting decision.
• Title structure following the pattern brand – product type – attribute clearly beats generic item names.
• GTIN, the correct Google product category and a reliable price and availability sync prevent disapprovals and account suspensions.
• PMax stays controllable through target ROAS, asset groups and exclusions – but only with margin logic rather than revenue logic.
• A clean feed pays off twice: on Google Shopping today, and with AI shopping agents tomorrow.

Why the feed is the real targeting

With classic search campaigns you choose keywords; with Shopping and PMax, that role falls to the product data feed: Google matches search queries directly against your products’ titles, descriptions and attributes. A title that only contains the internal item name simply matches fewer search queries than one that speaks the customer’s actual language.

That has a consequence many retailers underestimate: feed maintenance isn’t an IT ticket you set up once and forget, but an ongoing marketing task. Every new product line, every attribute change, every price adjustment directly affects visibility – often more than a change to your bidding strategy. If you read campaign reports but never touch the feed, you’re optimising the wrong lever.

This applies especially to seasonal ranges and international feeds: a feed optimised for barbecue accessories in summer loses visibility in winter if nobody updates the seasonal attributes and titles. And a feed built purely for the German market rarely works unchanged in Austria or Switzerland – even small linguistic differences in search terms affect matching.

Title structure that actually works

The best title structure follows a fixed order: brand, product type, the most important distinguishing attribute, optionally target group or use case. This order matches more real search queries than a marketing title, and can be set up for an entire range in five steps:

1. Study search behaviour: What terms do customers actually use (search suggestions, internal store search, Search Console data)?
2. Set attribute priority: For shoes, colour often matters more than material; for electronics, capacity often matters more than colour – priority varies by category.
3. Build a title template per category: for example {Brand} {Product type} {Main attribute} {Size/Variant}.
4. Test against bestsellers: manually check the titles of your ten highest-revenue products before rolling the rule out across the whole range.
5. Feed rules, not manual work: store the structure as a rule in Merchant Center or a feed tool so new products get titled correctly automatically.

One example makes the difference concrete: “Item no. WS-4471-42” matches practically nothing. “Salomon XA Pro 3D men’s hiking boot, grey, size 42” hits exactly the search terms someone actually types before buying.

The most common feed mistakes

Five mistakes crop up again and again in practice – each one costs visibility or budget directly:

1. Technical instead of natural-language titles: internal item numbers or SKU codes in the title instead of customer language. Fix: build the title template around search behaviour, not ERP logic.
2. Missing or incorrect GTIN: without a valid GTIN, visibility drops noticeably, because Google struggles to categorise products without a unique identifier. Fix: make GTIN mandatory in the product-creation process.
3. Wrong Google product category: a category that’s too broad or simply wrong dilutes matching. Fix: set up category mapping properly once and maintain it for new ranges.
4. Price or stock drift: landing page and feed fall out of sync, often through delayed sync intervals. Fix: increase sync frequency and set up automated alerts for discrepancies.
5. Images with overlay text or watermarks: breaches policy and lowers the click-through rate. Fix: check image rules before every product upload, ideally automated.

Feed management: tool or in-house?

Whether a dedicated feed management tool pays off depends on your range and how often it changes, not on company size alone. If you run a few hundred stable products, well-maintained feed rules in Merchant Center are often enough. If you’re managing thousands of products, frequent price changes or several country feeds at once, a tool noticeably saves time and reduces errors.

As a rough rule of thumb: in-house rules are usually enough if your range stays under a few hundred SKUs, price changes are rare and you’re only serving one market. A feed tool pays off once you’re maintaining several thousand SKUs, several country feeds, frequent price or stock changes, or several marketplaces in parallel – at that point, manual maintenance no longer scales linearly with the range, but disproportionately with complexity.

From experience: the tipping point is rarely a fixed SKU count, but the question of how many hours a week get spent correcting the feed manually. Once that’s regularly more than two to three hours, a tool usually pays for itself within a few months – through saved working time alone, before you even factor in the visibility gain from better feed quality.

Performance Max, seen soberly

Performance Max is deliberately a black box: how the system distributes budget across channels stays largely opaque, and that’s exactly what makes many advertisers nervous. Control is still possible, just through different levers than before – target ROAS, well-cut asset groups (split by margin, category or seasonality rather than gut feeling), and consistent exclusions: brands, irrelevant search terms, unprofitable categories.

More important than pure revenue control is margin-based control. A high ROAS on low-margin products isn’t a win if it comes at the expense of high-margin products competing for budget in the same campaign pot.

Margin control instead of flying blind on revenue

A simple worked example shows why pure revenue control is misleading. Product A costs €50, has a 40 percent margin (€20 contribution margin) and a ROAS of 4. Product B also costs €50, but has only 15 percent margin (€7.50 contribution margin) and a ROAS of 6.

On pure revenue, Product B looks stronger – a higher ROAS. Once you factor in contribution margin, the picture flips: with the same €100 ad budget, Product A brings in €400 revenue and €160 contribution margin (400 × 40%), while Product B brings in €600 revenue but only €90 contribution margin (600 × 15%). Anyone optimising PMax on ROAS alone would shift budget towards Product B – and actually lose profit, even as revenue rises.

From experience: the most robust approach is to roughly tie the target ROAS per asset group to the average margin of the products it contains, rather than setting a single value for the whole range. High-margin categories can tolerate a lower target ROAS, low-margin ones need a higher one – otherwise the high-margin range ends up quietly subsidising the low-margin one.

The bridge to agent visibility

A clean, structured product feed doesn’t only pay off on Google Shopping. The same discipline – clear attributes, correct categorisation, consistent price and availability data – is also what allows shopping agents and AI systems to capture and recommend products correctly (see our article on agent-ready commerce). The same feed hygiene that secures Google Shopping visibility today increasingly also decides whether a language model can even cite your product correctly – a connection we explore in more depth in our article on LLM optimisation for stores.

Keep your feed clean for humans and machines alike today, and you won’t have to redo that work tomorrow. Feed quality is one of the few e-commerce investments that keeps its value across channel shifts.

The bottom line

Campaign structure and bidding strategy are interchangeable; a clean feed is the investment that keeps paying off through every Google change. Fix titles, attributes and sync quality first, and only then turn the bidding dial – that consistently gets you more visibility per advertising euro. The sensible next step: go through your ten highest-revenue products in the feed before your next PMax campaign goes live.