Black Friday 2025 revealed something most shoppers didn’t notice while hunting for deals — AI traffic to U.S. retail sites had increased by 805% compared to the previous year. That’s not a typo. The way people discover, evaluate, and purchase products through shopping apps has fundamentally shifted, and the apps sitting on your phone right now are probably smarter than you realize.
Amazon, Walmart, Sephora, Pinterest, and a growing list of retailers have spent the last two years quietly embedding AI features that change how their apps work. Visual search lets you snap a photo of someone’s outfit and find similar items instantly. Chatbots now handle complex product questions and can complete purchases without you ever leaving the conversation. Virtual try-on technology has gotten good enough that Nike reduced returns by 64% just by letting people see how shoes would fit before buying.
If you’ve noticed your favorite shopping apps getting eerily good at knowing what you want, this is why.
Visual Search: Your Camera Becomes a Shopping Tool
The idea is simple enough — see something you like, take a photo, find where to buy it. The execution has gotten remarkably sophisticated.
Pinterest Lens
Pinterest Lens has quietly become one of the most powerful visual search tools available to consumers, processing around 80 billion queries monthly as of late 2025. The platform’s more than 600 million monthly active users have driven 44% year-over-year growth in visual search usage, and Pinterest has leaned into this by making nearly every image on the platform shoppable.
You can photograph a lamp in a friend’s apartment, a jacket on someone walking down the street, or a piece of furniture in a magazine, and Pinterest will surface similar items available for purchase.
Google Lens
Google Lens operates on an even larger scale, handling nearly 20 billion visual searches every month. Visual search queries through Google grew 65% year-over-year as of July 2025, suggesting that consumers are increasingly comfortable using their cameras as search tools rather than typing descriptions into search bars. The integration with Google Shopping means results often include pricing, availability, and direct purchase links.
Amazon StyleSnap
Amazon approached visual search differently with StyleSnap, focusing specifically on fashion. Users upload images of outfits they’ve seen on social media or anywhere else, and Amazon’s AI analyzes:
- Color patterns and palette matching
- Fit and silhouette details
- Style category and aesthetic
- Specific clothing details and accessories
It’s particularly useful for those moments when you see an outfit on Instagram but have no idea where any of the pieces came from.
ASOS Style Match
ASOS built Style Match directly into their app, letting users photograph clothing items and find similar products from ASOS’s catalog within seconds. The speed is the selling point here — snap a photo, get results, start shopping.
eBay Image Search
eBay’s Image Search takes a broader approach, allowing visual search across their entire marketplace of new and used items. This is useful when you’re trying to identify something specific but don’t know what it’s called — vintage items, collectibles, or products where you’d struggle to describe the exact search terms.
The image recognition market behind all of this is projected to grow from $46.7 billion in 2024 to $98.6 billion by 2029, which tells you something about how much companies are investing in making cameras smarter shopping tools.
AI Shopping Assistants: Chatbots That Actually Do Something Useful
For years, retail chatbots were glorified FAQ pages that frustrated more customers than they helped. The current generation is different enough that the numbers are hard to ignore.
Amazon Rufus
Amazon’s Rufus has reached more than 250 million customers in 2025, with monthly active users up 149% and total interactions up 210% compared to the previous year. The more interesting statistic is behavioral — customers who use Rufus are 60% more likely to make a purchase than those who don’t. Amazon estimates Rufus will generate an additional $10 billion in annualized sales, which explains why they’ve made it increasingly prominent in the shopping experience.
Rufus handles the kind of questions that used to require either extensive research or hoping customer reviews addressed your specific concern. You can ask whether a tent is suitable for winter camping, whether a laptop can handle video editing, or how a particular brand’s sizing runs compared to another brand. The AI pulls from product descriptions, reviews, and Q&A sections to give answers that are actually useful rather than just linking you to generic help pages.
Walmart Sparky
Walmart launched Sparky powered by OpenAI’s technology, and they’ve pushed it further than most competitors by enabling ChatGPT Instant Checkout — one-click purchases that happen entirely within the chat interface.
You can ask Sparky to find a birthday gift for a ten-year-old who likes dinosaurs, refine the suggestions through conversation, and complete the purchase without ever navigating to a product page.
eBay AI Agent
eBay introduced their AI shopping agent this year, delivering what they call “hyper-personalized” product recommendations and guidance based on your shopping history and stated preferences. This year alone, eBay has rolled out five new AI features including:
- AI-backed shipping estimates
- Shopping chatbot for product discovery
- Real-time personalized product picks
- Expert guidance based on preferences
Alibaba Wenwen
Alibaba’s Wenwen/Taobao Wenwen assistant is moving toward what the industry calls “agentic” behavior — AI that doesn’t just answer questions but takes actions on your behalf. Within Taobao and Tmall, Wenwen blends conversational product discovery with embedded calls-to-action like “Track Price” and “Buy Now” that appear naturally within the conversation. It’s the early version of AI that shops for you rather than just helping you shop.
L’Oréal Beauty Genius
L’Oréal took a specialized approach with Beauty Genius, an AI assistant focused entirely on beauty products. The feature set includes:
- Selfie scanning for personalized skin diagnostics
- Hair analysis and product recommendations
- Database of insights from 150,000 dermatologists
- AR try-on integration for recommended products
Shoppers arriving at retail sites from AI services were 38% more likely to make purchases compared to traditional traffic sources, which suggests these assistants aren’t just novelties — they’re actually helping people find and buy products they want.
Personalized Recommendations: The Algorithm Knows Your Style
Personalized product recommendations now account for up to 31% of eCommerce revenues, and stores using these systems see conversion rates up to 4.5 times higher than those showing generic product grids. This isn’t new technology, but the sophistication has increased dramatically.
How Modern Recommendation Engines Work
The basic mechanism analyzes multiple data points about your behavior:
- Browsing history and time spent on products
- Past purchases and purchase frequency
- Items viewed but not purchased
- Wishlist and saved items
- Data from across the broader platform ecosystem
Amazon’s recommendation engine is the most famous example, powering everything from the “Customers who bought this also bought” sections to the personalized homepage that looks different for every user.
eBay Shop the Look
eBay’s “Shop the Look” feature represents a more curated approach — an immersive carousel of complete outfits tailored to your shopping history, with interactive hotspots that reveal similar items and outfit inspirations. Their AI-generated subject lines for email campaigns drove a greater than 40% increase in quality visits, showing that personalization extends beyond the app itself into how retailers communicate with customers.
Pinterest AI Feed
Pinterest feeds are now AI-powered at every stage, from the initial pins you see to the search results to the shopping recommendations embedded throughout. For a platform built around visual discovery, AI personalization determines not just what products appear but which aesthetic directions the algorithm thinks you’ll respond to.
The Consumer Comfort Shift
A 2024 Boston Consulting Group survey of 23,000 consumers found that 80% of customers worldwide are comfortable with personalized experiences, which represents a significant shift from the privacy concerns that dominated discussions just a few years ago. Retailers using personalized recommendation algorithms reported a 22% increase in customer lifetime value — people who feel like an app understands their preferences tend to keep coming back.
The adoption curve is accelerating. According to a 2025 Adobe survey, 39% of consumers already use generative AI for online shopping, with 53% planning to do so this year.
Virtual Try-On: See It Before You Buy It
Return rates have always been one of eCommerce’s most expensive problems, and virtual try-on technology is finally mature enough to make a real dent.
Google Try It On
Google’s Try It On feature now works with just a selfie, powered by their Gemini 2.5 Flash Image model. Shoppers across the U.S. can upload a photo and see how clothing items might look on them, with the feature available across billions of items from retailers including:
- Macy’s
- Kohl’s
- Walmart
- Nordstrom
The technology has improved enough that results look plausible rather than like obviously photoshopped images.
Sephora Virtual Artist
Sephora’s Virtual Artist set the standard for beauty try-on and the numbers reflect genuine adoption — within two years of launch, users had tried on over 200 million shades and the feature logged more than 8.5 million visits. Sephora reported a 200% increase in user engagement after rolling out the AR-powered tool, and it’s become central to how many customers shop for makeup on the app.
The technology comes from ModiFace, which L’Oréal acquired and now powers try-on features across multiple beauty brands.
Warby Parker Virtual Try-On
Warby Parker built virtual try-on using Apple’s ARKit platform and Face ID technology, going beyond simple image overlay to analyze:
- Face shape for frame recommendations
- Skin tone for color matching
- Facial proportions for fit assessment
For a company that disrupted eyewear retail by making online glasses purchasing viable, virtual try-on addresses the obvious concern about buying frames you can’t physically try.
Nike Fit
Nike Fit delivered the most dramatic results — a 64% reduction in returns and 73% improvement in customer confidence about sizing. Across the broader industry, virtual try-on technology has reduced online returns by an estimated 28%. When returns cost retailers money on shipping, restocking, and often result in items that can’t be resold at full price, those percentages translate to significant savings.
ASOS See My Fit
ASOS became the first European retailer to trial “See My Fit,” an AR tool offering customers simulated views of products in different sizes and on different body types. Rather than just showing how an item looks, it addresses the more practical question of how it would look on someone with your specific proportions.
Market Growth
The market reflects the growing importance of this technology. Projections suggest the global virtual try-on market will grow from somewhere between $5.8 and $12.5 billion in 2024 to between $27.7 and $48.8 billion by the end of the decade. For glasses specifically, 29% of shoppers have already used virtual try-on technology, more than doubling from 13% in 2022.
Why Brands Are Racing to Add AI
The pressure to adopt these technologies comes from multiple directions simultaneously.
- Consumer expectations have shifted. That Boston Consulting Group finding that 80% of customers are comfortable with personalized experiences represents demand as much as acceptance — people expect shopping apps to know their preferences and surface relevant products. Apps that don’t meet this expectation feel broken rather than just less sophisticated.
- The ROI numbers are difficult to ignore. When Amazon’s Rufus users are 60% more likely to purchase, when personalization drives 31% of eCommerce revenue, when virtual try-on cuts returns by double digits — the financial case for AI integration is straightforward. The 40% increase in quality visits from AI-generated personalization at eBay represents real money.
- Competitive pressure compounds the effect. When Amazon, Walmart, and Alibaba all offer AI shopping assistants, brands without similar capabilities are competing with a structural disadvantage. The gap between leaders and laggards widens as AI features improve and users grow accustomed to them.
This is pushing brands to partner with mobile app development studio that specialize in AI integration. Building these capabilities in-house requires expertise that many retailers don’t have, and the timeline pressure means waiting to develop internal capabilities isn’t viable. Visual search, recommendation engines, chatbots, and AR try-on all require different technical approaches, and the implementation complexity means most brands need external help to catch up.
What’s Coming Next
The current generation of AI shopping features is already being superseded by more ambitious approaches.
Agentic AI
Systems that take actions on your behalf rather than just providing information — this is the direction Alibaba’s Wenwen and others are heading. Instead of helping you shop, these systems will eventually shop for you within parameters you set:
- Reordering household staples automatically
- Finding the best price on items you’ve previously purchased
- Monitoring wishlist prices and purchasing when they drop below your threshold
- The shopping assistant becomes a shopping agent
Expanded AR Try-On
AR try-on is expanding beyond makeup, glasses, and clothing into furniture, home decor, and other categories where seeing items in context matters. IKEA’s Kreativ already lets users visualize furniture in their spaces, and the technology is becoming standard across home goods retail.
Multimodal Shopping
Voice and visual multimodal shopping combines camera-based search with conversational interfaces. Rather than choosing between typing a search, taking a photo, or talking to a chatbot, future interfaces will blend all three into more natural interactions. You might photograph an item while asking questions about it verbally and receive responses that combine visual and textual information.
The 805% increase in AI traffic to retail sites suggests we’re still in the early adoption phase. The shopping apps on your phone a year from now will be meaningfully smarter than what you’re using today.
References
- PYMNTS – Pinterest monthly active users and query statistics
- PPC Land – Pinterest query volume and Google Lens growth statistics
- Daffodils – Amazon StyleSnap feature details
- G2 – Image recognition market projections, eBay Image Search, IKEA Kreativ
- Focale – ASOS Style Match
- Fortune – Amazon Rufus sales projections
- Stansberry Research – Walmart Sparky and ChatGPT Instant Checkout
- eBay – eBay.ai Agent announcement
- CNN – eBay AI feature rollout
- Feedonomics – Alibaba Wenwen, Adobe survey statistics, virtual try-on market projections
- Modern Retail – L’Oréal Beauty Genius
- TechCrunch – Black Friday AI traffic statistics, Google Try It On Gemini integration
- Techbuzz – AI traffic conversion rates
- Rep AI – Personalized recommendations revenue and conversion statistics
- Shopify – Boston Consulting Group personalization survey, Warby Parker ARKit details
- Digital Commerce 360 – eBay Shop the Look, Google virtual try-on availability, virtual try-on market growth
- EComposer – Sephora engagement statistics
- Onix – Sephora Virtual Artist usage statistics, Warby Parker technology details
- BrandXR – Nike Fit return reduction statistics
- Medium – Virtual try-on industry return reduction
- RetailWire – ASOS See My Fit launch
- Quartz – ModiFace and L’Oréal relationship
