Alexa shops by text now. Your listing copy decides

Amazon has opened Alexa for Shopping to text messages, so an invited US shopper can now search, fill a cart and buy inside a message thread. Amazon describes the assistant messaging back with suggestions, so plan for your title and bullet text to do work your photography does elsewhere on Amazon. Invited customers first, more in the coming weeks.

What Amazon switched on

Amazon said on Friday, November 6, 2026, that it is launching a way for customers to chat with its Alexa agent over text messages to shop. Its own announcement describes selected customers texting requests while the assistant finds the right products, builds the shopping cart, makes purchases and sends delivery updates. It started rolling out that day to a small group of US customers by invitation only, expanding to more customers over the coming weeks. Getting started means accepting the terms and linking a phone number, after which the contact appears in the phone as Alexa Shopping.

The thread is not a search box in different clothing. Amazon's examples include reordering the laundry detergent bought last month, setting up a weekly grocery order, comparing top-rated robot vacuums, pulling up to a full year of price history, sending a photo to find similar products, and surfacing options from stores across the web through Shop Direct. Amazon also says a request started over text can carry over to the Shopping app, Amazon.com and Alexa-enabled devices, and the other way round.

The scale behind it is the part worth sitting with. Amazon says more than 400 million customers have used Alexa for Shopping in the last 12 months, and that the number of customers using its agentic features has grown more than eight times since the start of the year. My read is that a company does not wire a purchasing agent into SMS for an audience it still considers a side experiment.

What a text reply does to your listing

Start with the shopper's screen. It is a message thread, and Amazon describes the assistant searching for products, comparing options and messaging back with suggestions. The feature is days old and limited to invited users, so treat what follows as a guess rather than a verdict. In a thread, the comparison is carried by the assistant's sentences. Your main image, your lifestyle photography and your A+ modules are, at best, one tap away. What travels into the conversation is likely to be the text Amazon holds about your product, which means the title, the bullets, the structured attributes and the language in your reviews.

I have argued for a while that listing content is advertising, because the same copy decides whether a paid click converts and whether an assistant recommends the product. A text surface removes the ambiguity from that argument. Amazon's own examples have a customer telling the assistant to only show organic options, or saying they only want to see vegan products, and the assistant remembering it. If a claim like that lives only in a comparison chart inside your A+ content or in text baked into a packaging photo, it is not in the material an assistant can filter on.

Which sellers feel it first

This lands hardest on three groups of US brand owners.

Visually sold categories come first. Beauty, apparel and home brands win the click with the hero image and the swatch, and a text summary strips that advantage down to whatever the copy says. Consumables and replenishment brands come second, for the opposite reason. Amazon's own example is a shopper texting to reorder the detergent they bought last month, which names an incumbent and skips comparison altogether. My read is that on this surface, winning the first purchase is worth more than it used to be, and losing it costs more. Third are brands whose differentiation sits mostly in A+ content rather than in the listing text.

The read on why Amazon did this now

Two pieces of public timing point the same way. TechCrunch reports that the move comes after Amazon blocked Meta's Muse from querying its site in September, and that the company is rolling the feature out just before the start of the holiday shopping season. Read together, the motive I would name is keeping discovery inside Amazon. Shut the door on an outside agent, then open your own one in the place shoppers already type, in the weeks when they buy most.

A second motive I would name is collecting data Amazon did not have. Setup links a phone number to the account, and Amazon says the more customers use it, the more it learns their personal preferences, shopping habits and tastes. Stated preferences are a different kind of signal from inferred ones, and they are hard to get from a search box. Both of these are reads on the public evidence, not statements of Amazon's intent. The pattern is not unique to Amazon either. TechCrunch notes that DoorDash said third-party agents had so far placed only hundreds of orders in total on its platform, and that it plans to keep developing its own text-based agent.

Auditing a catalogue that has to read well as plain text

The work scales with the catalogue, and the gap is wide. Across the nine published Sellrbox case studies at /customers, which are a selected published set rather than a sample of every account we run, catalogue sizes range from 22 SKUs to more than 22,000 ASINs, and the set spans both Managed and Self-Service engagements as well as single-marketplace and multi-marketplace accounts. A 22 SKU catalogue can be read end to end. A catalogue of 22,000 ASINs cannot, so the honest answer is to start with the SKUs carrying the ad spend and the SKUs that get reordered.

For each of those, check that the text alone answers the questions a shopper types into a thread:

• The use case and the person it is for, inside the title, not inferred from the image
• Any attribute a shopper might filter on in conversation, such as organic, vegan, fragrance-free or a size, written as words in the bullets and filled into the structured attributes
• The reason to pick you over the next product, in the first bullet rather than the fourth
• Pack size, count and refill compatibility, because reorder requests depend on matching a previous purchase

If you want the broader mechanics of how products get surfaced on Amazon's AI surfaces, our Alexa for Shopping ranking playbook covers the search side in more depth than a news piece can, and the listing conversion guide is the right companion for the copy itself.

One small test while invites are still limited

Take ten ASINs that matter, strip out the images, and put the title and bullets of each next to the same text from two competitors in a plain document. Hand it to someone who has never seen the products and ask which they would buy and why. That is close to the material Amazon describes the assistant working from when it searches, compares and messages back, though the feature is new enough that the comparison is a guess. Then change one thing on one product and watch conversion rate and the queries that convert over the next few weeks.

The decision in front of you this month is small and cheap, which is the best kind. Rewrite the text for your top-selling and most-reordered SKUs so it stands up without a photograph, and keep watching as Amazon sends more invitations over the coming weeks. If you would rather have someone else run the catalogue sweep alongside your ad work, that is what managed service is for.