Optimise your product feed for CSS: how to improve your Google Shopping data

Do you need a special product feed for a CSS Partner?

The product data requirements of Google Merchant Center apply to your shop's Shopping ads even when you work with a CSS Partner. Switching to a different CSS on its own therefore does not require a fundamentally different feed. Whether you also need an export for the provider's comparison website depends on how that provider is connected.

This is separate from the product data a CSS submits to CSS Center for its own product comparison pages. There is a separate specification for that. This article is about your merchant feed in Merchant Center and the Shopping ads for your online shop.

Agree with your CSS Partner which data sources are used and who makes changes. That way you know whether a correction belongs in your shop system, in your feed tool or directly in Merchant Center.

1. First fix errors that keep products from serving

A disapproved product cannot reach customers through the affected surface. So first check in Merchant Center which items are disapproved or limited and which causes Google gives. Missing or wrong product data, image problems and mismatches with the website can all affect serving.

Work through the issues by their business impact: an error on a high-revenue, well-stocked product deserves more attention than a problem with an item you are dropping from your range anyway.

A sensible order is:

1. Handle account-level issues and notices with deadlines.
2. Fix disapproved bestsellers and important product groups.
3. Remove recurring errors in the data source.
4. Then improve additional information and display details.

For every issue, check several affected items. If, for example, sizes are missing for all shoes, there is probably an error in the field mapping. A corrected export rule then achieves more than editing individual products by hand.

2. Write product titles that describe the item unambiguously

A good product title helps match the offer to a relevant search. At the same time, buyers should quickly see whether the product is what they are looking for. An internal model name like “Urban Pro 24” is often not enough for that.

Add the attributes that matter for the choice. Which ones those are depends on the product category:

The examples show a principle: brand, model, product type and relevant variant attributes belong together in a clear way. Only use properties the item actually has.

Google allows up to 150 characters for the title attribute. Put the most important information first, because titles are shortened depending on the display. The available length is not a target you have to fill with extra terms. Avoid keyword repetition and sales phrases like “Buy cheap now”.

Start with one important category. Define a title pattern for it and check samples before you roll it out to your whole range.

3. Maintain GTIN, brand and manufacturer part number correctly

Unique product identifiers help Google match your offer to the right product. So check whether existing GTINs are transferred correctly and belong to the respective variant.

Variants are a typical source of errors: different sizes or pack sizes can have different GTINs. So do not simply copy the identifier of the main item to all versions. Use the manufacturer's data and never make up substitute numbers.

If a GTIN is missing from your export, that does not mean none exists. First check your inventory system, supplier data and product packaging. The identifier_exists attribute with the value false is only meant for products that genuinely do not have the relevant unique product identifiers. It is not a fix for incomplete master data.

4. Keep price and availability in sync with your shop

Correct prices and stock levels prevent disappointment after the click and problems with product approval. Compare the submitted data with the actual product page: do price, currency, selected variant and availability match?

Choose the update frequency to suit your business. With frequent price changes or fast-moving stock, a daily export may not be enough. Make sure changes reach Merchant Center promptly, and check that the upload was processed successfully.

Google can correct certain mismatches with automatic item updates based on your website. However, this feature does not replace a regularly updated data source. If the same mismatches keep occurring, fix the transfer and the product data in your shop.

Pay particular attention to the start and end of promotions. Also check whether a sale price is submitted at the right time and whether the advertised variant can be bought at that price.

5. Submit every variant with the matching data

Clean variant data helps buyers land directly on the item they searched for. Anyone who clicks on a black sweater in size M should arrive on a product page with exactly that selection.

Submit every advertised variant as its own product with its own ID. Variants that belong together get the same item_group_id. Details such as colour, size, image, price and availability must match the respective variant.

A simple example:

Open the links from your feed directly in the browser. If all variants only point to the default selection, you may need to add suitable variant parameters to the URL.

6. Use images that show the offered product clearly

Product images should show what buyers will get before they click. So check whether your main image shows the right item in the right colour and stays understandable in a small preview.

Use sharp images with sufficient resolution and a clear view of the product. Extra promotional elements such as discount banners, shipping promises or merchant logos added afterwards do not belong on the main image. Brand markings that are on the product itself are a different matter.

Prioritise products with high visibility and weak images. Then also check the image URLs: Google must be able to reach them, and they must lead directly to the right image.

7. Add purchase-relevant attributes to the description

A specific description complements the title and explains what the item does and what it is suitable for. Generic statements like “premium quality for the highest standards” hardly help with choosing a product.

Instead, describe material, dimensions, features, compatibility or what is included, as far as this information is relevant and verified. For a laptop bag, for example, the usable inner dimensions and the layout of the compartments are more helpful than a long paragraph about modern design.

Use these questions as an editorial checklist:

• What exactly is the product?
• Which properties influence the purchase decision?
• Which version is on offer?
• What is included in the delivery?
• Which limitations or compatibility details matter?

Only automate descriptions once your source data is reliable. A neatly worded description with invented properties creates new problems.

8. Keep the Google product category separate from your own range structure

A clear categorisation makes it easier to classify your products and to work with your range. Distinguish between google_product_category and product_type.

The Google product category comes from Google's predefined taxonomy. Google assigns products automatically; with a valid value of your own, you can make the assignment more precise if needed. You use the product_type attribute for your own structure instead, such as “Sport > Running > Running shoes > Men”.

First check for obvious misassignments. Then standardise your own product type structure. Do not alternate between “Running”, “Jogging” and “Run sports” for the same category if you want to analyse them together later.

9. Use custom labels for product groups that make business sense

Custom labels make your feed more useful for campaign management. They let you tag products by criteria that matter to your business, such as margin, season or sell-off status. Google provides five attributes for this, custom_label_0 to custom_label_4; the values are not shown to customers.

A possible setup:

Only adopt this scheme as far as it supports your actual work. Define each group clearly and update changing values regularly.

The labels alone do not change any bids or budgets. You then have to use the groups in Google Ads for analysis and a suitable campaign structure. A margin label is helpful, for example, when you derive different profitability targets from it.

10. Measure whether your changes bring better results

Feed optimisation is meant to improve the marketing of your products. So before every larger change, document which items you are editing, what you are changing and what result you expect.

After fixing price errors, first check whether the affected products are approved again. For revised titles, also watch impressions, click-through rate, search terms where available, and the quality of the resulting sales.

For a meaningful comparison, try not to change titles, images, budgets and bidding targets at the same time. Choose a product group with enough data and take price promotions, season and changes in the competition into account. A before-and-after comparison gives you indications, but on its own it does not prove a causal effect.

How to get started with feed optimisation

You do not have to revise every item straight away. Start with the problems and product groups where an improvement can achieve the most for your business.

First: Check disapprovals, price mismatches and wrong availability. Fix recurring causes in the data source.

Next: Revise titles, identifiers, variants and images of your most important products. Check the data in Merchant Center after the upload.

Then: Add a consistent range structure and custom labels. Use the new groups for concrete decisions in Google Ads.

Regularly: Check error messages and the results of your changes. Note which rules work and which need adjusting.

Tackle CSS and feed optimisation together

Choosing your CSS Partner and maintaining your product feed are two decisions that belong together in your Shopping setup. The partner provides the agreed CSS service; your product data describes which offers you market.

At csspartner.io you can use feed management alongside the CSS service. This is especially useful if product data regularly arrives incomplete, variants are mapped incorrectly or you want to steer your range in Google Ads in a more targeted way.

Want to improve your product feed? Talk to csspartner.io about your shop, your data source and the problems you want to solve first.