Amazon sellers have until July 27 to shorten their titles, or its AI does it for them

Amazon sellers have until July 27 to cut their product titles to 75 characters, spaces included, across every category except media. After that, the platform enforces the cap itself: titles that run long get rewritten by Amazon’s own AI, using its relevance model, your backend keywords, and your bullet points. Brand-registered sellers get a 14-day window to review the AI’s version before it goes live; sellers without Brand Registry get no review at all. Alongside the shorter title, Amazon added a new 125-character “Item Highlights” field for materials and use cases, which surfaces in mobile snippets and appears to carry its own search weight.

So what: the tactical work is a two-week job (audit titles, rewrite the long ones, write Item Highlights), but the signal underneath is that Amazon is building a product-data layer for its AI to read. The room is in the tooling for it, and in learning to write for the model, not only the shopper.

---

The longer read

The surface read is that shorter titles convert better on mobile, and Amazon has the data to back it. But if brevity were the whole point, Amazon would enforce the limit and stop there. Instead it added a field. The 125-character Item Highlights box is Amazon restructuring its catalog for a reader that isn’t entirely human anymore: Rufus, its AI shopping assistant, already parses listings to make recommendations, and Item Highlights gives it a clean, structured input the brand controls instead of a keyword-stuffed title.

For sellers, the tactical distinction that matters is rewrite, not trim. Auto-truncation doesn’t optimize; it cuts at character 76, mid-keyword or mid-feature, and leaves a broken title with your brand on it. The rewrite strategy is to lead with your highest-converting keyword, not your brand name, which Amazon already shows separately in the mobile layout. Spending 15 of your 75 characters on the brand is paying for space you already own.

Item Highlights is where the displaced keywords find a new job, but it is not a second title or a dumping ground. Amazon built it for models to read, so write it as a plain statement of what the product is made of and what it is for. “600 GSM Turkish cotton, quick-dry, fits standard bath bars” gives an AI everything it needs to recommend your towel; “premium luxury soft towel set best seller” gives it nothing.

The review gap is the part to act on first. Brand-registered sellers get a first draft from Amazon’s AI to approve, edit, or replace; everyone else gets whatever the model writes. If your catalog is large, the path is either internal tooling that batch-audits titles through the Seller Central API, or one of the listing platforms shipping bulk-audit and rewrite features against the new limit. Either way the clock is short.

The broader signal is worth watching past this deadline. Amazon is standardizing how AI reads product data, starting with Item Highlights for Rufus. It is a short step from there to structured fields for the agentic checkout standards taking shape elsewhere, like the Shopify-Google Universal Commerce Protocol. Sellers who learn to write for both humans and models now are building a skill that compounds, not one that expires on July 28.

Related

• Walmart, Google Gemini and UCP: what agentic checkout means for ecom operators — platforms formalizing structured, AI-readable product/checkout data