A Shopping campaign can report an acceptable ROAS while quietly spending on products with thin margin, weak stock cover or poor conversion rates. This Google Merchant Centre feed optimisation guide 2026 explains how to make your feed useful for matching, eligibility and profitable customer acquisition – not merely free of errors.
Direct answer: optimise a Merchant Centre feed by first fixing required data and landing-page mismatches, then improving titles, identifiers, product types, images and attributes using real search, margin and conversion evidence. Finally, segment products by commercial value so campaign spend reflects contribution, stock and customer value rather than catalogue size.
What feed optimisation actually changes
Your product feed is not a back-office export. It is the structured product information Google uses to understand what you sell, assess whether an item can serve, and match it with relevant searches across Shopping and Performance Max inventory.
A clean feed prevents avoidable disapprovals. A commercially useful feed does more: it gives Google clearer signals about the product, its variants and its category. That can improve query relevance and make product-level reporting easier to act on. It cannot repair a weak price position, an unconvincing product page or inaccurate purchase measurement.
Google’s current Product data specification defines mandatory and conditional attributes by product category and market. Treat this as the rulebook for eligibility, rather than relying on a historic feed template. (Source: Google Merchant Centre Help, Product data specification.)
The priority depends on the diagnosis. If half the catalogue is disapproved, resolve policy and data issues first. If products are approved but spend concentrates on low-margin items, the problem is usually product segmentation and campaign control, not a missing adjective in a title.
Start with commercial data, not title rewrites
Before changing the feed, decide which products deserve acquisition spend. Export product-level revenue, units, returns where available, gross margin, shipping or fulfilment cost, stock status and repeat-purchase signals. You do not need perfect contribution reporting to make better decisions, but you do need more than platform ROAS.
A practical starting calculation is:
Allowable acquisition cost = net revenue – cost of goods – fulfilment and payment costs – target contribution
For example, a £100 sale with £45 product cost, £10 fulfilment and payment costs, and a £20 contribution target has an allowable acquisition cost of £25. A 4x ROAS may look attractive, but it is only viable if it produces that £25 cost per purchase or better. VAT treatment, returns and new versus returning customers can change the calculation, so use the definition your finance team actually trusts.
Create a simple product decision framework:
- Scale: healthy margin, reliable stock, competitive conversion and sufficient demand.
- Protect: profitable but limited stock, seasonal demand or narrow capacity.
- Fix: viable economics but poor feed data, weak imagery, low conversion or unclear positioning.
- Exclude or limit: low margin, frequent returns, poor availability or a product that only works for existing customers.
This classification should inform custom labels. Custom labels are not visible to shoppers; they are feed fields that let you group products for reporting and campaign structure. Google allows up to five custom-label attributes per product. (Source: Google Merchant Centre Help, custom labels.)
Useful label logic might include margin tier, bestseller status, seasonality, stock band and new-customer suitability. Avoid labels that no one will maintain. A single accurate margin tier is more valuable than five inconsistent classifications.
Google Merchant Centre feed optimisation guide 2026: fix the foundations
Match feed data to the landing page
Price, availability and condition in the feed must match the product page. These are common sources of disapproval and can also create a poor shopper experience when they slip through. If prices change frequently, ensure your feed refresh cadence and structured data reflect that reality.
Availability deserves particular attention. Do not leave products marked as in stock when variants are unavailable, fulfilment is delayed beyond your promise, or a product is effectively made to order. If your platform cannot express the distinction cleanly, resolve the operational mapping before scaling spend.
Use the correct product identifiers wherever they exist. For new products sold globally, GTIN is generally required; brand and MPN requirements depend on the product and identifier availability. Do not invent GTINs or use a generic value to pass validation. (Source: Google Merchant Centre Help, unique product identifiers.)
Build titles for query matching and product clarity
Titles should describe the product in the order a buyer would search for it. For many catalogues, a sensible pattern is brand, product type, defining attribute, size or compatibility, and variant detail. The exact order depends on the category.
A fashion retailer may need brand, gender, product type, material and colour. A parts retailer may need brand, part type, compatible model, dimensions and specification. A homeware title may need product type, material, size and colour. Do not force one formula across the catalogue.
Use words supported by the product page and real product attributes. Avoid promotional claims, excessive capitals, delivery promises and keyword repetition. The aim is accurate matching, not squeezing every possible query into one field.
Review search terms, product performance and on-site search together. A title change is a hypothesis: it should clarify the item for searches that are commercially relevant, not simply increase impressions.
Complete category-specific attributes
Product type is your own taxonomy and is useful for reporting and campaign organisation. Google product category uses Google’s taxonomy. Populate both where practical; they solve different problems.
Then check the attributes that change how a product is understood: colour, size, gender, age group, material, pattern, size system, item group ID and multipack or bundle details. Apparel and variant-heavy catalogues often lose relevance because this information is absent, inconsistent or placed only in a description.
For variants, item group ID matters. It tells Google that several SKUs are versions of one parent product rather than unrelated items. Make variant attributes consistent and ensure each landing page selects the advertised variant correctly.
Images, descriptions and annotations need a business case
Images influence clicks and product confidence, but creative changes should be judged against conversion and margin, not click-through rate alone. Use clear, high-resolution primary images that show the product accurately. If the category and policy allow it, additional images can address objections such as scale, material, fit, usage or included components.
Descriptions should answer useful questions that titles cannot carry: key specifications, compatibility, materials, care, dimensions and use case. Keep claims verifiable and consistent with the landing page.
Promotions, sale prices and shipping settings can improve competitiveness where they genuinely apply. They are not a substitute for a credible offer. A discount that destroys contribution margin should not be used to manufacture a feed advantage.
Turn feed quality into campaign control
Do not let a single Performance Max campaign decide that every product in the catalogue deserves equal investment. Use custom labels and product groups to create deliberate budgets or separate campaigns where product economics, inventory and strategic role differ.
For instance, high-margin bestsellers may justify a dedicated campaign and tighter monitoring. Clearance lines might have a fixed budget and a different target. Low-margin accessories can be excluded from prospecting while remaining available for branded demand. The right structure depends on spend, catalogue size, conversion volume and how quickly the business can act on product-level insight.
Measurement is the limiting factor. Merchant Centre data cannot tell you whether a purchase was profitable if purchase values are wrong, refunds are ignored or new-customer reporting is unreliable. Validate that order IDs deduplicate purchases, currency and tax handling are deliberate, and product-level values reconcile sensibly with the store and finance data.
Common mistakes that waste time
The most common mistake is treating diagnostics as a strategy. Fixing every warning can consume weeks without changing commercial outcomes. Prioritise errors that block serving, attributes that affect relevant matching, and data needed to control spend by economics.
Another mistake is bulk rewriting titles without a control group or before fixing the landing pages. If the product page is vague, out of stock or uncompetitive, a better title may only buy more unprofitable clicks.
Finally, do not confuse an approved feed with a healthy acquisition system. Feed quality, campaign structure, tracking, pricing, delivery proposition and product-page conversion work together. A weakness in any one can limit the others.
A sensible implementation sequence
In week one, audit disapprovals, account-level shipping and tax settings, price and availability mismatches, identifier coverage and top-spend products. In week two, add commercially useful custom labels and improve titles and attributes for the products that already drive meaningful spend or margin.
Then restructure product groups or campaigns around your scale, protect, fix and limit decisions. Give changes enough data to assess, while watching for stock shifts, pricing changes and seasonal effects. Review monthly at minimum, and more often for fast-moving catalogues.
Software can automate feeds, rules and supplemental data when the catalogue is large or changes frequently. It is unnecessary when a small, stable catalogue can be maintained accurately in the ecommerce platform. No feed tool can decide your margin thresholds, define a viable customer-acquisition cost or resolve disagreement between marketing and finance.
If your feed is approved but Shopping profitability is still unclear, the issue may sit across product economics, measurement, campaign structure and store conversion. The €650 eCommerce Google Ads Profitability Diagnostic is designed to identify which constraint should be fixed first – before more budget is committed.
A better feed should make the next commercial decision easier: which products to scale, which to protect, and which should stop absorbing spend.