How to write Shopify product titles that AI shopping agents actually read

How agents read titles (different from how Google reads them)

Traditional SEO advice says: stuff your title with keywords your buyer searches for. AI shopping agents flip that. They want titles that are specific, factual, and parseable, close to a structured field. The agent's internal task isn't "rank pages by keyword density" but "tell the user which product matches what they asked for."

Three downstream behaviors fall out of that:

• Product-type match.Titles with the product-type noun in them get matched cleanly into category queries.
• Attribute match.Titles with a distinctive attribute (material, ingredient, use-case) get matched into attribute queries, which are the majority of queries in our buyer-intent benchmark suite (see the methodology section of the Q2 report).
• Fluff downweighting.Titles padded with marketing fluff ("premium", "amazing", "best") are downweighted as low-confidence. Agents are explicitly trained to mistrust brand self-claims.

The 6 criteria from the rubric

Source: commerce-agentic/agentic-catalog-scanner (CC0). Every criterion below is implemented in the open rubric, so you can verify exactly how we score.

The two highest-weight criteria (#1 length and #3 product-type noun) are also the two most commonly failed in our public audits. Fix those first.

The default title formula that works

A reliable template that hits criteria #1, #2, #3, and #4 in a single line:

{Brand or gender qualifier} · {Product type} · {Distinctive attribute or spec}

Most catalogs that score well across all 6 criteria are some variant of this. The pattern works because it pushes the product-type noun into the visible portion of the title and forces at least one factual marker.

Vertical patterns: before / after

The pattern is consistent across all 10 verticals: name the product type, then specify the disambiguating attribute. The "before" examples all fail criterion #3 or #4 (or both), which is the most common reason a catalog underperforms on AI recommendations.

The 5 antipatterns to drop

How to bulk-fix titles in Shopify

For catalogs under 50 SKUs you can rewrite manually in the product edit page. Past that, the productive path is:

1. Export your products to CSV (Products → bulk actions → Export).
2. Open the CSV in a spreadsheet. Audit the existing titles against the 6 criteria. Quick wins: length out of range, missing product-type noun, fluff openers.
3. Rewrite a sample of 5-10 titles using the formula above. Run our public audit on your store after re-importing to verify the title-score increase.
4. For full-catalog rewrites, the Shopify app bulk-runs the same rubric with Claude-driven rewrites under a per-criterion pre-flight check (it won't ship a rewrite that doesn't improve the title score).

What this article doesn't claim

• We don't have controlled before/after data showing "fixing title criterion X increased AI mentions by Y%." The required experiment is hard to run cleanly.
• We do have the open rubric, calibrated against the brands AI agents cite most. The catalogs that follow the 6 criteria predominate in those captures.
• We do observe that catalogs where titles violate the antipatterns above are systematically absent from top recommendations in our captures.

Once your titles are fixed, descriptions are next. They're signal #2 (20 of 100 pts, the highest single-signal weight). See the descriptions deep-dive. After that, set the structured fields the agents prefer: the metafields guide.