How to rewrite product titles for AI search.
Your titles were written for Google snippets — keyword-front-loaded, adjective-padded, optimized for a ten-blue-links world. Read one aloud inside an AI answer and it sounds wrong, because the engine has nothing in it to quote. Nobody on the team knows what the new format is supposed to be.
A title an AI engine can use states function, audience, and one or two quantified capabilities. eCommerce Insights drafts per-SKU rewrites — title, description, bullets — as reviewable diffs tuned per engine and channel; a human approves every change. See the listing-rewrite docs and PDP optimization.
The slow way: rewrite by intuition, argue by taste
The manual version starts with a doc titled "Title guidelines v3" and ends in a taste argument. Someone rewrites twenty titles; someone else objects that they're too long, too plain, or off-brand; nobody can point to evidence either way because nothing is being measured. The rewrites that ship are the ones that survived the meeting, not the ones that work.
If you do it manually, anchor on a pattern instead of taste. The strongest public evidence is Amazon's COSMO research (SIGMOD 2024): shopping engines traverse intent relations — function, audience, capability, use context — and a title that states none of them is unreachable. "Premium Stainless Steel Travel Mug" surfaces nothing actionable. "16oz Insulated Travel Mug for Coffee on Long Commutes — Leakproof, Fits Standard Car Cupholders, Keeps Hot 8 Hours" surfaces four relations an engine can match against a real buyer question. Rewrite your top twenty titles against that pattern, ship, and track citation status weekly for a month. That works; it just doesn't scale past the hero SKUs, and the measurement discipline rarely survives the quarter.
The eCommerce Insights way
- Find the titles that fail. Filter the catalog for SKUs with low answer-coverage scores or stuck at BRAND-only citation status — the engine knows the brand, never quotes the product. Those titles are the queue.
- Apply the pattern. Function + audience + quantified capability, primary keyword still early. The rewrite engine drafts against the buyer prompts the SKU is actually losing, so the title answers the questions buyers actually ask.
- Review the diffs. Each SKU's proposal arrives as a before/after diff — title, description, bullets — in the listing-rewrite workflow. Tuned per engine and channel: the Amazon variant respects Seller Central style rules; the Shopify variant feeds the six-engine scan.
- Approve as a human. Edit, approve, or reject. Nothing auto-writes; brand voice survives because a person owns the final text. The same review queue covers the schema diffs.
- Ship and measure. Push to Shopify (Growth, early access) or export CSV, then watch per-engine citation status over the next two to three weeks. Titles that moved the needle stay; titles that didn't go back in the queue with the evidence attached.
The body-copy side of the same fix — answer blocks, FAQ coverage, the first 300 characters — is covered in the optimize content for AI search guide.
What "good" looks like
Ask AI about this job
Have your favorite AI engine apply this walkthrough to your titles.
Frequently asked questions
What does an AI engine actually want from a product title?
Will rewriting titles for AI hurt my Google rankings?
Does eCommerce Insights write the titles automatically?
Are titles really weighted that heavily by AI engines?
Should titles differ per channel?
Titles an engine can quote. Voice you still own.
Per-SKU rewrite diffs, human-approved, measured per engine.