Amazon is rolling out a feature inside Rufus, its AI shopping assistant, that allows shoppers to describe themselves in open-ended terms - their style, hobbies, household composition, and even their pets - and stores that information as a persistent profile. Every subsequent search, whether typed into the search bar or spoken to Alexa, is then filtered through that saved identity. The feature, surfaced today by Vanessa Hung, CEO of Online Seller Solutions and a widely followed e-commerce strategist, has immediate implications for how product listings are ranked and discovered on one of the world's largest retail platforms.

The interface itself is straightforward. Inside the Rufus panel, a prompt labelled "Tell us about you" presents a free-text input field accompanied by the instruction: "We will use these details to improve your personalization and shopping recommendations. You can tell us anything about you - your style, hobbies, interests or who you shop for, like family or pets." A separate section, titled "About you," confirms what the system does with that input: "As you shop on Amazon or chat with Rufus or Alexa, we'll save helpful information to personalize your experience. Once we have saved some details, you can manage them here."

Both interface descriptions are confirmed by screenshots shared in Hung's LinkedIn post on April 22, 2026.

What the profile layer actually does

The mechanics here go beyond simple preference tagging. According to Hung's analysis, this is a profile-matching update rather than a query modification. The search query itself stays the same. What changes is the shopper identity layer that the query is evaluated against. Two people searching for "storage bins" will receive different product stacks if one of them has told Rufus "I have three kids and a small apartment" and the other has said "I run a minimalist home office." Same keyword, different results.

That distinction matters structurally. Traditional Amazon search optimisation operated on the assumption that keyword relevance, review velocity, and conversion rate determined ranking. Those signals still exist. But now there is an additional layer - a persistent identity profile - sitting between the query and the ASIN stack. A listing's bullet points, A+ content, and backend keyword metadata are now evaluated not just against the search term but against what Rufus knows about the person typing it.

Hung's framing captures the shift precisely. According to her post, "the system is no longer just reading your listing, instead, it is reading it against a shopper identity layer." Listings that use explicit use-case language - phrases like "designed for families with young kids" or "built for small home offices" - become anchors that Rufus can map to real shopper profiles on file. Generic feature lists, which describe a product without addressing any identifiable type of person, do not provide that mapping signal.

The technical structure of the feature

The "Tell us about you" feature operates as a persistent memory store at the account level. Unlike session-based personalisation, where context resets after a browsing session ends, this profile persists across devices and across Amazon's surface area. According to the interface text confirmed in Hung's post, the saved details apply not just to Rufus conversations but to broader Amazon shopping interactions and to Alexa.

That cross-surface persistence is significant. It means a shopper who tells Rufus "I have three dogs and a small apartment" on a mobile device will receive product results shaped by that identity when browsing on desktop, speaking to Alexa, or using Amazon's visual search tool, Lens. The profile becomes the connective layer across the entire Amazon ecosystem.

The input mechanism is deliberately open-ended. There are no dropdown categories, no pre-set interest tiles, no structured demographic fields. Shoppers write in natural language. This approach has a specific consequence for how Rufus processes the output: the system must interpret free-text identity descriptions and map them to product attributes and listing signals at scale. That requires Rufus's underlying large language model to function as an identity-to-ASIN matching engine, not simply a query-answering tool.

Rufus operates on a multi-model architecture running through Amazon Bedrock, utilising models including Anthropic's Claude Sonnet, Amazon Nova, and a custom model trained on Amazon's product catalogue. A real-time router directs queries to appropriate models based on capability requirements, latency constraints, and answer quality optimisation. The addition of a persistent identity layer means that routing decisions now incorporate not just the incoming query but a structured representation of who is asking it.

Context: Rufus at scale

The scale at which this identity layer operates matters. More than 300 million customers used Rufus throughout 2025, and the assistant generated an estimated $12 billion in incremental annualized sales during that year, surpassing the $10 billion pace Amazon had discussed after the third quarter. Monthly average users grew 149% year-over-year and interactions climbed 210% over the same period, according to Amazon's fourth quarter financial results published in February 2026.

Customers who engage with Rufus during shopping sessions convert at rates 60% higher than those who do not. That figure, from Amazon's own data, reflects the assistant's existing effectiveness as a product matching tool. The "Tell us about you" feature adds a new upstream signal to that matching process - one that does not require Rufus to infer shopper context from browsing behaviour alone but instead draws on explicit self-reported identity.

Amazon launched Rufus in beta in February 2024, expanding availability to all US customers in July 2024 ahead of Prime Day. Since then, the system has received more than 50 technical upgrades, including account memory, automatic purchasing, price alerts, and visual search features announced in November 2025. The November update introduced account-level memory that understands individual shopping patterns - a customer who mentioned having sports-enthusiast children and a golden retriever would receive recommendations prioritising athletic gear and pet-hair-capable vacuum cleaners. The "Tell us about you" feature extends that memory system with a proactive, user-initiated identity declaration rather than a passively inferred one.

What changes for product listings

The identity layer introduces a structural split in how product content functions. According to Hung's analysis, bullet points, A+ content, and backend keywords now serve two distinct functions simultaneously. They continue to feed traditional keyword indexing - the established mechanism by which Amazon's search algorithm matches queries to ASINs. But they also operate as identity-matching signals that Rufus reads when evaluating a listing against a saved shopper profile.

A listing that says "designed for families with young kids" is not just describing a product feature. It is anchoring the ASIN to a use-case category that Rufus can match against profiles where shoppers have declared they have children. The same logic applies to phrases like "ideal for pet owners," "built for small home offices," or "perfect for minimalist spaces." These are identity anchors, not merely marketing copy.

Generic feature lists - specifications, dimensions, material descriptions without person-referencing language - do not provide that anchor. They describe the product but do not map it to any identifiable type of person. As Hung put it in her post, "generic feature lists do not anchor to anyone." In a system where the product stack is filtered through a shopper identity layer, a listing with no identity language is invisible to the profile-matching mechanism even if it ranks well on keyword relevance alone.

This is a meaningful structural shift from the keyword-centric paradigm that has governed Amazon listing optimisation since the platform's early years. Amazon began applying AI personalisation to product descriptions and recommendation types in September 2024, using large language models to reposition key product attributes for individual users - highlighting "gluten-free" at the top of a description for shoppers who frequently search for it, for example. The "Tell us about you" feature extends that approach from passive behavioural inference to active identity declaration.

Implications for the advertising layer

For the marketing community, the identity profile layer lands at a moment when Amazon's advertising infrastructure is itself becoming more AI-mediated. Amazon's AI shopping prompts moved to paid general availability on March 25, 2026, ending a free beta period and introducing cost-per-click charges for AI-generated shopping conversations. Advertisers are now paying for engagement with a surface that is simultaneously being personalised through shopper identity profiles.

The combination creates a new variable in sponsored product visibility. A listing may win an auction and appear in a shopper's results but still lose relevance if the listing's content does not contain the identity language that Rufus uses to confirm a match against that shopper's profile. In practical terms, paid placement and organic relevance are now both necessary but neither is sufficient.

Amazon's Persona Builder tool, launched in early 2024, already allowed advertisers to construct audience segments based on Amazon's first-party data, combining behavioural signals to define personas such as premium fitness headphone buyers. The shopper-declared identity profiles that "Tell us about you" generates represent a first-party signal of a different quality - explicit rather than inferred, structured around the shopper's own self-description rather than Amazon's behavioural modelling of them.

How Amazon surfaces these declared profiles to advertisers, if at all, has not been disclosed. The announcement from Hung does not indicate whether the saved profile data will be accessible through Amazon Marketing Cloud or incorporated into Amazon DSP audience segments. What is clear is that the profiles influence product ranking inside Rufus, which is the surface where, according to Amazon's own data, shoppers convert at 60% higher rates.

Shopper control and data persistence

The interface confirms that profile data is manageable. According to the "About you" section visible in Hung's screenshots, once details are saved, shoppers can "manage them here" - implying the ability to edit or remove saved information. The system stores profile data at the account level, meaning it persists across sessions and devices rather than expiring with a browsing window.

Amazon has not published detailed documentation on data retention periods, the specific ways in which declared identity attributes are weighted against behavioural signals, or whether profile data is used for advertising targeting beyond the Rufus and Alexa surfaces. Those questions are material for privacy-conscious shoppers and for brands seeking to understand how their listing content is evaluated.

The broader context of Amazon's data practices is relevant here. Amazon updated its Business Solutions Agreement in February 2026 to include formal restrictions on the use of Amazon's materials for AI development purposes, and has moved to block external AI bots from accessing platform data while expanding its own internal AI capabilities. The "Tell us about you" feature generates shopper identity data that remains within Amazon's ecosystem - a pattern consistent with the company's broader strategy of consolidating first-party signals inside its own infrastructure.

Reaction from the e-commerce community

The feature generated immediate discussion among practitioners. Shane Barker, founder of TraceFuse.ai, noted in a comment on Hung's post that "generic feature lists not anchoring to anyone is the listing problem most sellers don't even know they have." A commenter using the handle Shangruff, describing themselves as an e-commerce growth expert, characterised it as "a big shift - it's no longer just about ranking for keywords, it's about matching real people. Listings that speak to a specific lifestyle or use case are going to win." A third commenter, Madha Noor Akmal, an Amazon catalogue and account health specialist, offered a compressed operational response: "Make an ideal customer persona first then write the copy."

Hung credited Ritu Java with sharing the update first. Java, a specialist in Amazon advertising and AI-readiness for listings, has been documenting the implications of Rufus personalisation for seller strategy and was scheduled to host a session covering how to audit listings for Rufus personalisation, what shopper-profile signals mean for Rufus visibility strategy, and listing edits designed to improve AI-readiness.

The Rufus personalization thread connects to a wider industry shift. As answer engine optimization becomes a discipline in its own right across general-purpose AI platforms, Amazon is constructing its own version of that optimisation surface - one where the answer engine is embedded directly inside the product catalogue and the identity of the person asking is a declared, persistent variable rather than something inferred from clicks.

Timeline

Summary

Who: Amazon, through its Rufus AI shopping assistant; documented by Vanessa Hung, CEO of Online Seller Solutions, with attribution to Ritu Java for first surfacing the update. Commentary from practitioners including Shane Barker (founder, TraceFuse.ai), Shangruff (e-commerce growth expert), and Madha Noor Akmal (Amazon catalogue specialist).

What: A new feature called "Tell us about you" inside Rufus allows shoppers to enter free-text descriptions of themselves - style, hobbies, household composition, pets - which are saved as a persistent identity profile at the account level. The profile is applied to all future product searches and recommendations across Rufus, Alexa, and broader Amazon shopping surfaces. Listings with explicit identity-anchoring language - phrases that address specific types of people - are positioned to match more effectively against saved profiles than listings that use generic feature descriptions.

When: The feature is actively rolling out as of April 22, 2026, the date on which Hung published her analysis. The broader Rufus personalisation infrastructure has been under development since Rufus launched in beta in February 2024.

Where: Amazon's platform, accessible through the Amazon Shopping app and the Rufus conversational interface. The saved identity profile applies across devices and surfaces, including Alexa, according to the interface text confirmed in Hung's post.

Why: The feature represents Amazon's move from passively inferred shopper context - derived from browsing history and purchase behaviour - to explicitly declared identity. For shoppers, the stated purpose is more relevant product recommendations. For the broader marketplace, it introduces a new matching layer between queries and ASINs that rewards listings containing identity-specific language and disadvantages listings that describe products without addressing any particular type of person. The development arrives as Amazon's advertising infrastructure grows more AI-mediated, with AI shopping prompts moving to paid placements and Rufus driving conversion rates 60% higher than non-assisted shopping sessions.

Share this article
The link has been copied!