Google Shopping Feed Optimisation: 12 Fixes by Profit (2026)

Google Shopping feed optimisation improves the product feed you send to Google Merchant Center, so Google can approve your products, match them to relevant searches and present them well. This guide orders 12 fixes by the profit at stake and covers the margin information needed to decide which traffic is worth buying. If you searched for Google Shopping feed optimization, you are in the right place.

Start with Merchant Center’s Needs attention issues. A better title cannot help a disapproved product. Once products are eligible, use these 12 fixes to prioritise work by the profit at stake.

The order below is an editorial priority for a store with unresolved feed issues, not a measured ranking of revenue uplift. Your biggest opportunity depends on affected products, demand, contribution margin and advertising spend. This guide covers field-level feed fixes. For campaign structure and the wider review routine, use our Google Shopping ads optimisation guide; for a new account, begin with what Google Merchant Center is.

How does product data affect Shopping performance?

Your feed supplies the product information behind Shopping ads. Google uses product data rather than a positive keyword list to understand what you sell. Campaign settings, landing pages and auctions also affect delivery; the feed is a starting point, not the only control.

Read how Merchant Center fits into Shopping if you are setting up your first data source. For account setup and your first campaign, use the Google Shopping beginner’s guide. For an ongoing account review, start with Google Shopping management.

Google Shopping feed optimisation best practices: the 12 fixes

1. Restore price and availability accuracy

Fix mismatches first when they prevent otherwise profitable products from serving. Check the exact variant, currency and checkout price, not only the product family’s default page. Google requires the submitted price to agree with the landing page and checkout; availability must also match. Follow its price requirements for your target country, including the relevant VAT treatment.

Use a daily refresh as a working baseline; send updates sooner for fast-changing prices or stock. Automatic item updates can correct some discrepancies, but your data source still needs to stay current.

Use a platform connection, scheduled fetch or the Merchant API for updates. Google says the Content API for Shopping reached sunset on 18 August 2026, with progressive request errors from 1 September 2026 and full shutdown expected in early 2027.

2. Exclude products that cannot support paid traffic

Review products with persistently poor returns against their own contribution margin. Low margin, expensive fulfilment or frequent returns can make extra traffic uneconomic. Low stock can also justify lower priority when replenishment is uncertain.

Exclude at campaign level when the decision concerns one campaign. Standard Shopping uses product groups; Performance Max uses listing groups. Google’s product groups guide covers subdividing and excluding products; allow 24 to 48 hours for such changes to sync.

Use excluded_destination = Shopping_ads when the offer should stop appearing in Shopping ads across campaigns. That destination does not itself exclude free listings. Do not falsify availability to stop advertising a product that remains in stock.

3. Send margin and performance bands through custom labels

Custom labels let you organise products using your own business rules. Google’s custom label specification provides five fields, custom_label_0 to custom_label_4, with one value per field per product. Each value is 1–100 characters; each field supports up to 1,000 unique values across the account.

Use a small vocabulary. Put margin bands in one field and observed performance in another. Do not create a unique label for every product or overwrite fields another campaign already depends on. Label values are not shown to shoppers.

These are illustrative label names, not Product Metrics segment definitions. A label does not set a target ROAS or move budget by itself. Configure the campaigns that consume it. Separate campaign targets require appropriate campaign structure; listing groups inside one Performance Max campaign do not each get an independent target ROAS (Google’s campaign-structure guidance).

The diagram below follows one illustrative source product through its supplemental row and the processed field you must verify. The example thresholds are margin_high at 40% or more, margin_mid from 20% to below 40%, margin_low below 20%, and margin_unknown when costs are missing. These are example business rules, not Google’s or Product Metrics’ thresholds.

Illustrative attribute flow: SKU-A → 60% margin → custom_label_0 = margin_high → processed SKU-A field → campaign group. The label carries a rule’s output; the configured group supplies the advertising action.

Calculate the floor with the break-even ROAS calculator or the margin calculator: break-even ROAS = 1 ÷ contribution margin before advertising. Do not subtract expected ad cost inside the margin and then use that margin to calculate the advertising allowance: that counts advertising twice. On a consistent revenue and cost basis, POAS = ROAS × contribution margin and break-even POAS is 1. See POAS vs ROAS.

Illustrative data, calculated from the stated assumptions. All three products produce a ROAS of 4 at €25 spend, with different profit outcomes. These are calculations, not a before/after customer result. Break-even covers the variable costs included here; it does not fund fixed overhead or a desired profit reserve.

Product Segmentation writes performance segments to Merchant Center labels for configured campaigns to use. Its six segments use advertising clicks and return, measured as ROAS or POAS. A ROAS segment alone does not establish profit. Keep product costs and target ROAS assumptions explicit.

4. Fix product titles for the searches you want

Lead with the details that identify the product: product type, meaningful brand or model, then distinguishing attributes. Google permits 1–150 characters and says shoppers usually notice the first 70 or fewer, depending on screen size. Google advises putting the most important details first and using the available characters, but every word should be a true, useful attribute rather than padding. See the title specification.

For an illustrative fictional product, change Aster 2 to Aster 2 Men's Running Shoes, Black, UK Size 10 only when those details are true. This is an example edit, not evidence of improved performance. A recognised brand can go first; for an unfamiliar brand, product type may explain the offer faster.

To test it, save the product ID, original title and revised title. Check the revised value on the processed product, then compare mature periods with an unchanged product group where possible. Record clicks, spend, purchases and contribution after ads; changes in price, stock or promotions make a simple before/after comparison inconclusive.

Avoid promotional phrases, repeated keywords and attention-seeking capitals. Generate titles from verified attributes with rules rather than editing thousands manually. Google requires AI-generated product titles to use structured_title, with digital_source_type set to trained_algorithmic_media; check your feed tool’s support before generating product text.

5. Supply accurate GTIN, brand and MPN values

Use identifiers assigned to the product you actually sell. A GTIN helps Google recognise products across sellers. Submit brand and manufacturer part number (MPN) when available, following Google’s unique product identifier guidance.

Check the GTIN’s check digit using a validator. A valid check digit does not prove that the GTIN belongs to your product.

Do not use identifier_exists = no just because your records are incomplete. Apply Google’s category-specific identifier rules. Products without a GTIN can still have brand and MPN values; handmade does not automatically mean identifier-free.

6. Check Google’s category and correct it where appropriate

Google automatically categorises products. The google_product_category attribute is optional, and Google accepts overrides in specified cases, including category-based campaign targeting, incorrect category-specific requirements and alcohol compliance. It is not a mandatory field for every product.

Where an override is appropriate, choose the most specific accurate entry in the current taxonomy. Submit either the numeric ID or the full category path. Google’s category specification explains accepted overrides; use the taxonomy linked from that specification to find the correct ID or full path.

Avoid copying a plausible-looking category from an old guide. If Google has no suitable category, describe the product accurately and use your own product_type for your business structure.

7. Improve images without breaking the rules

Use a sharp image of the exact product and variant. Remove promotional overlays, watermarks and placeholders. Add alternative angles through additional_image_link; check category exceptions before applying one presentation rule to every product.

Google’s current image specification announces a 500 × 500 pixel minimum for all products from 31 January 2027, and recommends 1,500 × 1,500 or larger. Its specification table already shows the new minimum. Prepare images to meet it now rather than relying on the older 100/250-pixel guidance in many articles.

A model photo is not automatically disallowed for apparel: Google’s apparel guidance recommends showing clothing on models. A plain product background is useful in many categories, not a universal reason to reject every lifestyle image. Judge image tests on qualified sales and contribution as well as click-through rate.

8. Complete category and variant attributes

Fill required fields for the destination, country and product category. Apparel commonly needs color, size, gender and age_group; material, pattern and size system can make the offer clearer. Use Google’s product data specification to check the exact requirements rather than treating one apparel list as universal.

For example, Google’s age group specification sets requirements for apparel Shopping ads in named countries, including the UK, and for apparel free listings, with exceptions. Populate verified values across variants instead of filling missing cells with guesses.

9. Build a consistent product type taxonomy

Use product_type for a stable, store-defined path, such as Shoes > Running > Men's. It gives campaign reporting a usable structure without forcing your store’s taxonomy into Google’s category field. Google supports product type as a product group subdivision.

Keep paths specific enough to act on, but avoid changes that constantly move products between reporting groups. Document your mapping and review new products so they do not accumulate in an unhelpful catch-all category.

10. Make sale prices, shipping and tax treatment agree

Submit a genuine sale_price and use sale_price_effective_date for its active period. The effective date specification explains the date range. A sale badge has additional conditions. The discount must be greater than 5% and less than 90%. Show both original and active sale prices on the landing page, with the sale price most prominent.

For Shopping ads in the UK, the Netherlands and Germany, the base price must have been valid for 30 days within the previous 200; those days need not be consecutive. In the US, the base-price history is 5 days within the previous 30 or 15 within the previous 200. Check Google’s country and destination rules: free-listing conditions differ, and meeting the requirements does not guarantee an annotation.

Keep delivery charges outside the product price in Merchant Center’s shipping settings or the relevant shipping attribute, following the price specification. Cheaper shipping may make the offer more attractive, but absorbing it also reduces the contribution available for ads. Recheck the break-even floor after a promotion or delivery change.

11. Write factual descriptions with key facts first

Describe the product itself: use, material, dimensions, compatibility and distinguishing features. Google allows 1–5,000 characters. Its description specification excludes promotional content and links. Google advises listing the key details in the first 160 to 500 characters, because shoppers may need to click to see the rest.

Use structured_description with digital_source_type = trained_algorithmic_media for AI-generated descriptions. Review generated facts against the manufacturer’s specification; fluent wording does not validate compatibility or safety claims.

12. Scale corrections with supplemental sources and attribute rules

Merchant Center attribute rules, formerly feed rules, transform incoming values. Enable the Advanced data source management add-on to use them. Build a rule, save it as a draft, test it, inspect affected offers and apply it only when the output is correct. See Google’s attribute rule instructions.

A supplemental data source adds or updates data for products already in a primary source. It cannot create or remove products by itself. The supplemental source setup guide also requires the add-on.

No feed rule can recover a missing cost of goods figure from a title. Start with reliable inputs. Test a small group, preserve stable product IDs and retain the old rule so an incorrect transformation can be reversed.

Which feed fix should you test first?

Choose a group with a clear commercial problem and measurable potential. Review actual margin and return alongside visibility; a high-margin product with weak demand is not automatically a winner.

Change one type of input in a defined group. Record product IDs, dates, prices, spend and the change made. Compare equivalent periods and, where practical, an unchanged comparison group. Account for seasonality, promotions, stock, conversion delay and returns. An increase in clicks is not proof that the edit increased profit.

Common Google Shopping feed mistakes and fixes

Feed optimisation checklist

Download the feed optimisation checklist as a CSV and open it in your spreadsheet. It is free and needs no email address. Record an owner and the verification evidence for every fix.

• Start: review Merchant Center Needs attention and resolve blocking issues.
• Fix 1: match price, currency and availability to the exact landing-page variant and checkout.
• Fix 2: review products with persistently poor returns against their own contribution margin, then lower priority or exclude them from paid groups.
• Fix 3: calculate contribution margin before advertising and set an unknown-cost band.
• Fix 3: assign documented margin or performance labels without overwriting existing uses.
• Fix 3: configure campaign groups to consume labels and review contribution after ads.
• Fix 4: put identifying details first in titles and distinguish variants.
• Fix 5: use current identifiers; never replace GTINs with store SKUs.
• Fix 6: check automatic categories and apply valid overrides where appropriate.
• Fix 7: use accurate, high-resolution images, at least 500 x 500 pixels from 31 January 2027, without promotional overlays.
• Fix 8: complete required country, category and variant attributes.
• Fix 9: standardise product type paths.
• Fix 10: match sale dates, shipping charges and tax treatment to checkout.
• Fix 11: keep descriptions factual and use structured fields for AI-generated text.
• Fix 12: test supplemental sources and attribute rules before applying changes.

What to do next

Start with one affected product group. Check eligibility, then use the break-even ROAS calculator to find its floor and choose the feed fix with the clearest profit opportunity. When you want observed clicks and return written into Merchant Center labels for configured campaigns to use, see Product Segmentation.

Sources

All Google pages were checked on 15 October 2026.

• Google Merchant Center Help: title, description, image and apparel image specifications.
• Google Merchant Center Help: price, sale price effective date, sale price annotations and automatic item updates.
• Google Merchant Center Help: unique product identifiers, product data requirements, Google product category, age group and product data tips.
• Google Merchant Center Help: custom labels, excluded destination and product groups.
• Google Merchant Center Help: attribute rules and supplemental data sources.
• Google for Developers: Content API for Shopping sunset.
• Worked example: calculated from the stated assumptions; no customer results are used.