Retail returns have always been the cost of doing business online. The accepted tradeoff was: wider reach, more orders, more returns. Absorb the cost. Optimize the checkout. Move on.
That math is changing. The scale of returns has reached a point where absorbing the cost is no longer viable — and AI is creating new ways to solve the problem that simultaneously open marketing opportunities most e-commerce teams haven't considered.
The scale of the problem
The U.S. National Retail Federation reported that 15.8% of annual retail sales were returned in 2025, totaling $849.9 billion. Online returns were worse at 19.3%. Fashion is the most affected category, where sizing uncertainty and the gap between how products look on a screen versus on a body drive a significant portion of purchase-and-return behavior.
For years, the industry treated this as a logistics problem: faster processing, easier labels, restocking efficiency. But the root cause isn't logistics. It's information. Customers don't have enough data to make confident purchase decisions — so they buy multiple sizes, try them at home, and return what doesn't fit.
AI is now addressing the information gap directly.
What's actually changing
Three developments are converging to reshape how consumers evaluate products before purchase.
Google is embedding virtual try-on into search. Starting April 30, Google's virtual try-on technology will be accessible directly within product search results across Google platforms. This isn't a separate app or a third-party integration. It's inside the search experience itself — right where purchase intent is highest. Google's standalone Doppl app is shutting down to consolidate the technology into the core shopping experience. When virtual try-on is native to search, it becomes part of the discovery flow rather than an extra step.
AI shopping assistants are driving measurable revenue lift. Macy's rolled out "Ask Macy's," an AI-powered shopping assistant built on Google Gemini. The results are striking: customers who use the tool spend 4.75 times more per visit than those who don't. The assistant functions as a digital stylist, recommending complete outfits rather than individual items — which naturally increases basket size and average order value. This isn't incremental. It's a fundamentally different shopping behavior where AI replaces the browse-filter-compare pattern with conversational discovery.
Adoption is accelerating across the market. 40% of top U.S. retailers are now deploying AI shopping assistants. A wave of startups — backed by partnerships with Amazon, Adobe, and Google — is making virtual try-on and AI-powered fit recommendations accessible to mid-market brands that couldn't afford the technology even a year ago. What was an enterprise-only capability is rapidly becoming table stakes for any brand selling apparel, accessories, or home goods online.
The marketing implications
For e-commerce marketers, these shifts create three strategic changes that go well beyond "add virtual try-on to your product pages."
Product data is becoming creative.
When AI generates a virtual try-on experience, it works from your product feed — images, descriptions, sizing data, color attributes, fit specifications, and fabric details. The richer and cleaner your product data, the better the AI-generated experience renders. Brands with high-quality product imagery, complete attribute data, and structured sizing information will be featured in Google's try-on results. Brands with thin product data will be invisible.
This means your product feed is no longer just a commerce infrastructure concern. It's a creative asset. The same data that powers your shopping ads now powers the AI-generated shopping experience. Product data quality directly determines whether a consumer can see your product on themselves — or whether your competitor's product renders instead.
Conversion optimization is moving upstream.
Traditional e-commerce optimization focuses on the funnel: landing page to product page to cart to checkout. AI virtual try-on moves the highest-leverage optimization point upstream — to the discovery and consideration phase, before a shopper ever adds anything to a cart.
When a customer can see how a jacket looks on their body type inside a Google search result, the conversion decision happens earlier. The product page becomes a confirmation step rather than a persuasion step. This shifts where marketing teams should invest their optimization resources: less on checkout flow A/B tests, more on product imagery, size inclusivity in try-on models, and structured data that enables AI-powered discovery.
Return rate becomes a marketing KPI.
When virtual try-on reduces return rates by 30-40% — the range that early adopters are reporting — the financial impact is direct and compounding. On a product with a 25% return rate and a $15 average return processing cost, a 35% reduction in returns on 100,000 units sold saves $131,250 in processing costs alone. Add back the recovered inventory value, reduced shipping costs, and eliminated customer service interactions, and the total savings are substantially higher.
Those savings either flow to margin or get reinvested into customer acquisition. Brands that solve the returns problem aren't just cutting operational costs. They're funding growth with money that was previously lost to reverse logistics.
The return rate, traditionally a supply chain metric, now belongs on the marketing dashboard. It's a direct indicator of how well your product information — imagery, sizing guidance, AI try-on quality — is converting browsers into confident buyers.
What e-commerce marketers should do now
Audit your product data for AI readiness. Go beyond basic feed requirements. Ensure you have multiple high-quality images per product (including on-model shots), complete size and fit attributes, structured fabric and care data, and accurate color representation. This data is what AI systems use to generate virtual try-on experiences.
Prepare for Google's April 30 try-on integration. Review your Google Merchant Center feeds for completeness. Products with thin data will not render in try-on results. Products with rich data will gain a significant visibility advantage in shopping searches.
Track return rate by acquisition channel and product category. Start treating return rate as a leading indicator of product information quality. High return rates on specific products signal that the gap between online presentation and physical reality is too wide — exactly the problem virtual try-on solves.
Evaluate AI shopping assistant integration. The 4.75x revenue lift Macy's reports is an outlier at this stage, but it signals the direction. Even a more modest lift from AI-assisted product discovery would justify integration for most e-commerce operations above a certain scale.
The bigger picture
E-commerce is entering a phase where the shopping experience itself is AI-generated. Virtual try-on, conversational shopping assistants, and AI-curated product recommendations are converging into an experience that looks nothing like the static product grid that has defined online shopping for two decades.
The brands that prepare their product data infrastructure for this shift will have an advantage that compounds with every search query. AI systems that generate better try-on experiences will drive lower return rates, higher customer satisfaction, and stronger repeat purchase behavior — which feeds back into the AI models as positive signal.
This isn't about adopting a single new tool. It's about recognizing that your product data — its completeness, quality, and structure — is becoming the primary competitive variable in AI-mediated commerce.
*itscool.ai helps e-commerce brands build the product data infrastructure and AI-readiness strategies that convert in an era of virtual try-on and AI-powered shopping. If your product feed isn't ready for what's coming April 30, the time to fix that is now. Let's talk.*