Make the title describe the variant you sell.

For Shopify apparel catalogs · by Marco Nicola Messina · sources checked October 17, 2026.

Start with facts, then choose the wording

Build a small inventory of the fields you actually trust. A useful working row contains item ID, product name, garment type, confirmed material, color, size, current feed title and destination URL. Mark missing facts as missing. Do not turn a guess about composition, fit or audience into a product claim.

Google’s title guidance requires a relevant product description, distinguishes variants and excludes promotional wording. Titles can contain up to 150 characters. The title and landing page should describe the same product; adding size and color to a title does not replace the corresponding attributes.

Three fictional rows, three different decisions

These are review examples, not ready-to-import feed values. The second row stays “linen-blend”; shortening it to “linen” would change a fact. The third row illustrates a useful outcome of an audit: discovering that the source needs attention before copy does.

Keep product families and individual variants distinct

Review the item identifier and grouping separately from the title. Google’s item group ID documentation describes grouping variants of the same product while submitting each variant separately. Check the current requirements for all relevant variant attributes and your destination market. A shared parent title alone does not establish a correct variant group.

Open the submitted URL for each sampled row. If the navy medium row opens sand small, clearer wording will not fix the selection problem. Use our variant and price comparison to record that discrepancy before changing titles.

Choose where the mapping should live

Document the path from the Shopify source fields to the final feed title. Decide whether the approved change belongs in source data or a feed-specific rule. A feed mapping may allow channel-specific wording, while a storefront title change can affect what shoppers see elsewhere. Agree which surfaces may change before implementation.

Test one product family and its exceptions first. Export the original mapping and affected rows. After editing, compare the received data with the proposed values, check truncation and duplicated attributes, and confirm that the next synchronization preserves the result. Keep a rollback record.

If AI helps write the titles

Google currently requires AI-generated titles to use structured_title with digital_source_type set to trained_algorithmic_media. Check that your feed integration supports the required fields before deploying AI-generated copy. The examples above were generated for illustration; this page does not submit them to Google. Consult the linked title specification for the format used by your source.

Measure relevance without inventing a sales result

Before a live test, save the affected item list, original titles, proposed changes and a comparable unchanged group where practical. Record concurrent price, budget, bidding and inventory changes. Evaluate relevant search terms and item-level performance only when there is enough data to make a useful comparison.

A processing success proves that data was accepted at that stage. It does not demonstrate higher conversion or incremental profit. If the catalog is already accurate and there is too little traffic to assess wording, more rewrites may be a poor use of your budget.