Amazon has finally put numbers on its shopping assistant, and the numbers are large enough to change how you write a listing. More than 350 million shoppers used Alexa for Shopping over the past twelve months. Active users nearly doubled in the second quarter year over year. Interactions in the same quarter were up five times year over year. Those are Amazon’s own figures, disclosed on its second quarter 2026 earnings call.
Read that alongside the other Amazon change landing on sellers this month, the 75 character cap on product titles, and the two stop being separate stories. Amazon is shrinking the space a human reads and expanding the space a machine reads. The consequence is concrete and uncomfortable: if a growing share of discovery runs through an assistant that parses structured attributes and review text rather than a shopper scanning a results page, the assets that win are the ones a machine can parse cleanly. Your listing now has two readers, and only one of them scrolls.
What Amazon actually disclosed
The scale numbers matter because they are the first clean read on the product. Alexa for Shopping replaced Rufus in May 2026, so the twelve month figure spans two names on one surface, but the second quarter growth numbers sit entirely inside the new product’s life.
In Amazon’s own words on the quarter: “Over 350 million customers have used it in the last 12 months, and engagement accelerated in Q2, with active users nearly doubling, and interactions up over 5x year-over-year.” The company describes the surface as offering personalised recommendations, product comparisons, price history, and automation through features including Price Alerts and Auto-Buy.
Two of those features deserve a second look, because they are the part of the disclosure that reaches into your pricing. Price Alerts and Auto-Buy move a shopper from browsing to a standing instruction. A shopper who has told an assistant to watch a price is no longer in your funnel on your timetable; they are in a queue that triggers on a number.
The figures landed inside a strong quarter: net sales up 20 percent year over year to 200.6 billion dollars, operating income up 43 percent to 27.5 billion dollars. Amazon also expanded Amazon Lens, its visual search tool, to ten additional countries, taking it to 21. That one belongs in this story rather than the financial one, because Lens is another entry point where the first thing to read your product is software, not a person.
The 40 percent claim, read carefully
The figure that will travel furthest is this one, and it needs the most care. According to Amazon chief executive Andy Jassy on the earnings call, US customers who use Alexa for Shopping spend 40 percent more per order on average than shoppers who do not use the assistant.
Three qualifications, all of them load bearing.
- It is US only. The scope stated was US customers. There is no equivalent figure in the public disclosure for the UK, Germany, Spain or any other European marketplace, and it should not be repeated as a European number.
- It is Amazon’s own measurement. This is a first party figure from an interested party, delivered on an earnings call to investors. No independent audit or third party dataset has been published against it.
- It is a comparison between two groups of shoppers, not a proven effect. Amazon stated that assistant users spend more per order. That is not the same as showing that the assistant caused the larger basket. The people who reach for an AI assistant may already be the heavier, more engaged, more Prime-committed buyers.
That third point is where the seller-side reasoning has to stay honest. Amazon offered a related figure on the broader assistant: shoppers who try Alexa+ join Amazon Prime at nearly 25 percent higher rates. Read the two together and a simpler explanation is available, that assistant users skew towards Amazon’s most committed customers. The direction of causation has not been established publicly either way.
So the operator conclusion is not “the assistant makes people spend 40 percent more”. It is narrower and more useful: the cohort transacting through the assistant is high value, it is growing fast on Amazon’s own count, and it does not see your listing the way your analytics dashboard assumes it does.
If an assistant is comparing your product, it is reading your attributes, not your adjectives. Helium 10’s listing and keyword tools let you audit which terms and attributes are actually indexed on your ASINs before you rewrite anything. Code ECOMMGM10.
Affiliate link. E-CommSphere may earn a commission at no extra cost to you.
The companion fact: 75 characters
Now the other half of the pairing. Amazon’s product title standard caps the item name at 75 characters including spaces. The policy was announced in June 2026, took effect from late July, and Amazon extended the start date to 3 August. Non-compliant listings stay active, editable and searchable, and Amazon generates an AI recommendation for them, with brand owners given 14 days to review and approve the change before it is applied. The scope is not limited to third party sellers: the policy applies to Amazon’s own listings too.
The character budget has not disappeared, it has moved. Sellers now have 200 characters of important product information split across two fields: 75 characters of item name and 125 characters of item highlights. Amazon’s position is that both fields feed search and neither is prioritised over the other. On volume, Amazon says sellers have updated over 984 million product titles since the policy was announced in June.
Put the two facts side by side. The human-facing headline has been cut to a third of the space many sellers were using. The machine-facing fields, item highlights, structured attributes, review content, are where the detail now lives. That is not a coincidence of timing so much as a coherent direction: a shorter title is easier for a person to scan and easier for a model to normalise, and the specifics get pushed into fields that are built to be parsed.
What an assistant reads that a shopper does not
A shopper on a results page reads roughly six words of your title, one image and a price. An assistant answering “which of these is quieter and fits a 60cm cabinet” is doing something different. It needs the dimension, the decibel figure, the material and the compatibility, in fields where those values exist as values rather than as prose buried in a bullet.
Three things follow for a seller.
- Attribute completeness stops being hygiene and becomes discovery. An empty attribute is not a cosmetic gap when the comparison layer is software. A missing value is a filter you cannot appear in.
- Item highlights is where your 125 characters have to earn their keep. If Amazon’s stated position is that item name and item highlights are both search inputs, the highlights field is the natural home for the specifics the title can no longer carry.
- Review text is now input, not just social proof. Amazon describes the assistant as answering comparisons and questions. The language buyers use in reviews is part of what a model has to draw on when it answers them.
There is a defensive angle too. If you leave a non-compliant title to the automatic rewrite, an AI recommendation decides which of your 75 characters survive. For brand-registered sellers that comes with a 14 day review window. That window is the cheapest control you have over how your product is described to both readers.
Cutting a title to 75 characters means choosing which keywords you keep. Helium 10 shows search volume and rank history per term, so the cut is a data decision instead of a guess you find out about at the next sales dip. Code ECOMMGM10.
Affiliate link. E-CommSphere may earn a commission at no extra cost to you.
What is still unknown
Amazon disclosed reach and engagement. It did not disclose the things a seller would most want to know, and it is worth being blunt about the gaps rather than filling them.
There is no published breakdown of what share of Amazon orders originate in an assistant conversation, no country-level split beyond the US framing on the spend figure, and no statement on how the assistant selects between competing products or whether sponsored placements sit inside its answers. The 350 million figure counts shoppers who used the surface over twelve months, which is a usage count and not a measure of how many purchases it decided.
What the quarter does establish is direction and pace. On Amazon’s own numbers, active users nearly doubled and interactions rose five times in a single year, on a surface that is barely three months old under its current name. Sellers who wait for a clean attribution model before making their listings machine-readable will be waiting past the point where it mattered.
The practical read for this week is unglamorous. Get the 75 character titles decided rather than assigned. Fill the attribute fields you have been ignoring. Treat item highlights as a search field. None of that is a bet on how big agentic shopping becomes, which is exactly why it is worth doing now.
Sources
- Retail Dive, Amazon customers are embracing Alexa for Shopping
- Amazon, Q2 earnings: CEO Andy Jassy on Amazon Stores growth, delivery speed, and AI shopping
- Amazon, Q2 2026 earnings report
- CX Dive, Amazon customers embracing Alexa for Shopping
- EcommerceBytes, Amazon’s New Product Title Policy Also Applies to Its Own Listings