The way Amazon ranks and surfaces products is changing, and sellers are going to feel it first. The shift from keyword-matching to intent-modeling may seem subtle, but it’s about to reshape how Amazon listing optimization actually works.
We’ve spent years building Amazon content for enterprise brands like adidas, Ray-Ban, and Hydro Flask. The game we played was based on keyword coverage. Amazon’s search engine matched the words a shopper typed against the words in your listing, so the winning move was to pack every relevant term into the title, the bullets, and the backend. That era is coming to an end.
That intent layer is called COSMO. It’s a system Amazon built to read why people buy, not just what they type. Instead of matching your listing’s words against a shopper’s search, it maps products to real human context: what a product is for, who it’s for, when it’s used, what else it can do.
The search logic changed. It’s no longer asking whether your listing contains the words the customer typed. It’s asking whether your product fits what they’re trying to do. Amazon has tested it in search navigation on about 10% of US traffic and the roll-out looks set to continue.
COSMO doesn’t replace Amazon’s older search algorithms, the ones sellers call A9 and A10. Those still run. It adds an intent layer on top, and that layer carries the most weight on broad, open-ended queries.
A bare “running shoes” search can’t tell Amazon much, so it returns the broadest, highest-volume listings and lets them fight it out. But “shoes for marathon training with flat feet” tells Amazon what the shopper actually needs, and that’s where intent now influences what results show up.
What we’re seeing on Amazon reflects a much bigger shift in the way we interact with our devices. People have learned to ask LLMs for things in full, specific sentences, and they’re bringing that habit to Amazon’s search bar. Amazon is rebuilding its search function to reward listings that best answer those questions.
As of July 27, 2026, Amazon caps product titles at 75 characters in every category except media (down from 200). A new “Item Highlights” field gives you an additional 125 searchable characters for the details that no longer fit. If your title is still over the limit after that date, Amazon’s AI will rewrite it for you, and brand owners get 14 days to review the result.
Your first instinct might be to see this as an editing challenge. It’s not quite as simple as that. The title stopped being a place to carefully stack keywords. Now it has a more direct job: say what the product is, who makes it, and the one thing that sets it apart.
The new title cap only hurts listings that leaned on the title to do work the rest of the page should have shared. If your bullets, product description, and A+ content already carried those keywords, you lose almost nothing.
If they didn’t, cutting keywords out of the title exposes a bigger problem. It’s a real fix to make either way, but a listing built right comes through this almost unscathed.
Amazon is betting shoppers will increasingly ask questions instead of typing keywords. Rufus (rebranded Alexa for Shopping in May 2026) turns search into more of a conversation. To answer those questions, Alexa reads the actual product details in your listing, your bullets, product description, and A+ content. It can’t do much with a title that’s just a string of search terms.
The title used to hold a lot of key information. Now it can’t, so the rest has to be written into the listing itself. The four things COSMO reads (what the product does, who it’s for, when it’s used, and what else it can do) need to appear in your bullets, product description, and A+ content, not just the title.
Most people in the industry, us included, think COSMO is the secret sauce that powers Alexa’s recommendations. Amazon hasn’t confirmed that publicly. But, regardless, it doesn’t change what you should do. Whether or not the two are explicitly connected, the answer is the same: put your product’s key information everywhere a shopper or Alexa might look, not just in the title.
Luckily, you don’t have to guess what shoppers are looking for. The data is already in Seller Central. Pull a year of Search Query Performance, check your Brand Analytics by ASIN, and read your reviews, where customers tell you in their own words what they use the product for.
Nevertheless, intent-based optimization is more work than keyword mapping. Someone has to pull the data, read it, and write listing content from what it shows. While we are blessed with software that can help with the pulling, it can’t do the reading or the writing for you.
Pro tip: most “AI listing optimization” on the market is just keyword research with an AI writing assistant. The test is simple: ask the vendor to show you the intent data behind a rewrite. If they can’t, they’re doing the same keyword work as before, just faster.
Plenty of listings will feel this cut, because they kept too many of their keywords in the title and nowhere else. When the title shrinks, those terms have no other place on the page to fall back on.
We’ve spent the last two years building this into a system. We merged four AI technologies and developed our own keyword and intent mapping rubrics, the ones that balance search terms against the language shoppers actually use. Getting this right under the new algorithm is closer to a science than an art. You have to know the rules, and you have to know how to adapt your content, as this went live on July 27.
This is the work we do for enterprise brands on Amazon every day. The intent analysis, the rebuild, the A+ content. If your keywords are in the wrong place, we’ll fix it before Amazon does.
