AI Merchandising: The Quiet Revolution

The numbers don’t lie. Last year, retail customers returned nearly 850 billion dollars worth of merchandise. That’s 16% of everything sold, or one out of every six items changing hands twice. Online? Even worse – 19% of web purchases get sent back, often at a cost to the retailer that exceeds the refund itself.

Gen Z is driving this trend. Shoppers aged 18 to 30 averaged nearly eight online returns per person last year, according to NRF data. That’s not just a problem – it’s a pattern that changes how merchants need to think about their entire online experience.

I see this as the silent killer of retail margins. For years, we’ve talked about the returns problem without real solutions. But now, generative AI is giving merchants actual tools to fight this battle – not just theoretical fixes, but measurable improvements.

The Problem Merchants Actually Face

Let’s be clear. The single biggest driver of returns isn’t quality or price – it’s uncertainty about fit and suitability. Customers hesitate to buy online because they can’t try things on. They worry a dress won’t drape right, that shoes will pinch, that a color won’t match their existing decor. This hesitation kills conversion rates and guarantees costly returns.

But here’s what’s different in 2026. Early virtual try-on technologies from the 2010s felt clunky and unconvincing. The technology wasn’t there yet. Now? AI-driven virtual try-on platforms like Catches are giving customers detailed, realistic feedback about how items will actually look and fit. “It pinches here; drags there; the draping is wrong” – this is the kind of specific guidance that actually helps people make better buying decisions.

AI That Actually Drives Results

What gets me excited isn’t the technology itself – it’s what the technology enables when merchants use it strategically. Take Walmart’s approach during the Super Bowl. The company used generative AI to suggest complete party assortments based on a customer’s past behavior, not just individual items. A shopper who previously bought chips and salsa got suggested snack packs. Someone who bought beer got coolers and glassware. The result wasn’t just better experiences – it was bigger basket sizes and higher conversion rates.

Macy’s found similar results with their “Ask Macy’s” shopping assistant. Early testing showed shoppers who use AI assistance spend 400% more than those who don’t. Four times the revenue. That’s not a marginal improvement – that’s a fundamental shift in how customers engage when they have the right guidance.

David’s Bridal built an AI wedding planning platform called Pearl that increased time on site and drove higher average order values. These aren’t theoretical improvements – they’re measurable business outcomes.

Where Merchants Should Focus

I think there are three areas where AI delivers the most value for merchandising teams:

First, reducing returns through better visualization. If a customer can see exactly how something will look and fit, they’re far more likely to be satisfied with their purchase. This isn’t just about trying on clothes – it’s about visualizing how a sofa will look in their living room, how a paint color will appear on their walls, or how a piece of jewelry will complement what they already own.

Second, increasing basket size through intelligent suggestions. The key here isn’t just recommending more items – it’s understanding customer intent and suggesting the complementary items that actually make sense. A shopper buying a laptop doesn’t need random accessories – they need the right carrying case, external monitor, and software that solves their specific problem.

Third, improving decision-making speed. Merchants can now test assortment hypotheses in real-time rather than waiting weeks for sales data to roll in. This means faster response to trends and less dead inventory. We worked with a fashion retailer last quarter that used AI to adjust their spring assortment based on early social media trends – they ended up selling 300% more of the trending styles compared to their traditional planning approach.

What I’m Watching For

The ChatGPT Instant Checkout experiment failure tells us something important: customers aren’t ready to complete purchases within AI engines yet. The technology still has limits around trust and security. But what’s working is using AI as a guide during the consideration phase – not as the checkout process itself.

I also notice the conversation shifting from “how can we use AI?” to “what should our actual AI strategy be?” We’re moving beyond experimentation toward structured implementation. That’s when the real business value emerges.

Another thing I’m watching: the difference between hype and results. Early in 2025, everyone was talking about AI replacing human merchants. By late 2025, the conversation had shifted to “How do AI tools augment human expertise?” That’s the right question. The winning formula isn’t AI replacing people – it’s AI helping people make better, faster decisions.

Nosto is another example I’ve been tracking. They’ve combined enterprise-grade AI search, personalization, and merchandising control into a single platform that works for Shopify Plus merchants. The results they’re showing aren’t theoretical – they’re measurable improvements in conversion rates and average order values.

What are you seeing in your own merchandising efforts? Are virtual try-on tools actually reducing returns? Has AI-driven personalization moved beyond hype into tangible results? I’d love to hear what’s working in practice versus what’s still theoretical.

One thing I’ve learned in my years working with retail clients: the companies that succeed with AI aren’t the ones with the fanciest technology. They’re the ones with clear problems to solve and realistic expectations about where AI can actually help.

The quiet revolution in merchandising AI is happening now – not with fanfare, but with measurable improvements to margins and customer satisfaction. And that’s the kind of revolution I can get behind.