Why 2025's AI Shopping Boom Still Can't Fix Broken Business Models

Despite OpenAI, Google, and a wave of new startups promising to revolutionize fashion discovery, the fundamental problems remain unsolved

“As if!” - Cher Horowitz, Clueless (1995)

Author's note: I was the CEO of POPSUGAR, which owned ShopStyle as a wholly-owned subsidiary until we sold it to Rakuten in 2017. I have seen firsthand the challenges of fashion discovery platforms and their business model issues.

The AI revolution is here. ChatGPT can now shop for you, Google's virtual try-on is uncannily good, and a wave of startups is raising eight-figure rounds to build fashion's next great interface.

But despite all this progress, the business model is still broken. I saw it up close as CEO of POPSUGAR, which owned ShopStyle. Technology wasn't the problem then, and it isn't now.

The recent announcements tell the story. Daydream emerged from stealth with $50 million. OneOff promised celebrity-style emulation. Alta raised $11 million for AI styling straight out of "Clueless."

The technology demonstrations are impressive. Google's virtual try-on works across billions of items. ChatGPT processes rambling style descriptions and surfaces relevant products. Daydream wraps fashion search in a chat interface. Instead of matching keywords like "red dress," these systems understand concepts like "romantic evening wear." The AI can finally decode what humans actually want.

This represents genuine progress. The user experience improvements are real and meaningful. But technology was never the bottleneck. Better algorithms, same broken math.

Yet scratch beneath the surface, and you'll find the same fundamental problems that have plagued fashion discovery platforms for over a decade. The technology has advanced dramatically. The business model flaws remain just as broken.

Same Problems, Shinier Tools

Technological sophistication has never been the primary obstacle. Polyvore had elegant visual discovery tools a decade ago. ShopStyle offered comprehensive product search. Lyst, Wanelo, Fancy, Pose, Spring, and about fifty others you've probably forgotten all created powerful search capabilities. None became the billion-dollar businesses their backers envisioned.

The problem wasn't inadequate technology. It was flawed economics.

Better algorithms don't fix broken customer acquisition costs. They don't solve incomplete product catalogs. They don't eliminate attribution nightmares. They simply make the same old problems more expensive to solve.

The Attention Economy's Chokehold

When OpenAI adds shopping to ChatGPT, it's not creating a new shopping destination. It's embedding commerce into existing behavior patterns. Google's AI shopping features live within the world's dominant search engine. These aren't standalone platforms competing for attention.

Independent fashion discovery startups face a fundamental asymmetry. They must convince users to develop new habits while tech giants embed shopping into existing user flows. The attention economy has been carved up by platforms that own the top of the funnel.

The irony is stark. The same week Daydream launched its AI chatbot, Google announced features that let users shop conversationally within the platform they already use for search. The competitive dynamics haven't changed. They've intensified.

The Catalog Problem Persists

Despite technological advances, the product coverage problem remains acute. Daydream launched with 8,000 brands, positioning this as comprehensive coverage. But "comprehensive" in fashion means having virtually everything a consumer might want.

Luxury brands still refuse third-party aggregation. Amazon still operates as a walled garden. Shopify still keeps checkout behind its own walls. When someone searches for a specific brand and finds nothing, they don't blame the absent retailer. They blame the platform and rarely return.

In fashion, if something is missing, everything feels broken.

AI doesn't change these dynamics. It just makes the gaps more obvious.

The Attribution Nightmare Gets Worse

The affiliate revenue model has actually deteriorated. Browser extensions have become more sophisticated at hijacking attributions. Cross-device shopping has increased. And now AI itself creates new attribution challenges.

Meanwhile, the retailers that discovery platforms depend on for inventory are also their direct competitors for keywords and advertising. Nordstrom and Zara bid against Daydream for the same Google search terms. This makes it impossible for discovery platforms to compete. The retailers' customer acquisition costs will always be lower because they own the inventory and capture the full transaction value. Discovery platforms would need to make their service so sticky that they change user behavior entirely. No inventory-less aggregator has ever pulled this off.

When Google's AI completes purchases on behalf of users, it's disintermediating the discovery platforms that initially surfaced the products. The AI becomes the last touch before purchase, capturing the attribution that would have gone to the discovery platform that did the actual work.

The platforms invest in AI to improve discovery, then watch as the big tech platforms use their own AI to capture the transaction value. They do the work. The giants get paid.

A shopper discovers a dress on Daydream, clicks through to buy it, but Honey's browser extension hijacks the final commission. Daydream gets nothing for the discovery work. This scenario plays out millions of times daily.

The Funding Cycle Repeats

Daydream's $50 million seed round and OneOff's early-stage funding echo the investment patterns from previous waves of fashion tech. Venture capitalists are drawn to the same narrative. Massive market opportunity. Technological disruption. The promise of building the next Amazon in fashion.

The fundamentals haven't changed. Customer acquisition costs in fashion remain prohibitively high. Lifetime value calculations are challenged by low repeat usage. The AI wrapper doesn't change these realities.

That's the part most investors miss.

The Customer Experience Nightmare

We haven't even mentioned the horrible customer experience around multiple retailer checkout and shipping costs. A shopper finds three pieces for an outfit across different brands, then faces three separate checkouts, three shipping fees, and three different return policies. When something goes wrong, customer service becomes a nightmare of finger-pointing between the discovery platform and the retailers.

The only way to make this work is either a subscription model that justifies the friction, or become a massive destination and pivot to advertising like Pinterest did. Everything else is just expensive customer acquisition with no retention.

The Brutal Truth

This wave of AI shopping announcements represents impressive technological achievement. But these platforms are still building sophisticated bridges to other companies' checkout pages. They're still fighting for attention in an economy where distribution is controlled by dominant platforms. They're still trying to monetize through affiliate fees in an ecosystem designed to minimize those fees.

The paradox intensifies. The more sophisticated the technology becomes, the more it illuminates the fundamental business model challenges that no algorithm can solve. The venture capital community, seduced by the promise of AI, is funding the same flawed strategies with fancier technology.

The brutal truth is that the tech industry has spent over a decade and billions of dollars trying to disrupt fashion discovery. The incumbents have only grown stronger. Amazon dominates online retail. Google controls search. Social platforms own discovery. The AI revolution has given these giants even more tools to defend their positions.

Meanwhile, the new AI platforms face impossible economics. They compete against their own partners for customers. They build better discovery tools that bigger companies immediately copy. They solve attribution only to watch browser extensions steal their revenue. They improve user experience only to fragment it across multiple checkouts.

The real innovation won't come from better AI. It will come from entrepreneurs brave enough to acknowledge that the fashion discovery problem isn't technological. It's structural. Until someone builds a business model that actually aligns incentives rather than fighting them, we'll continue to see impressive technical demonstrations masquerading as breakthrough companies.

But here's the opportunity. The AI advances are real. The technology finally works. The entrepreneurs who recognize this moment and build around structural realities rather than against them will create something genuinely transformative. That might mean vertical integration, white-label solutions for existing retailers, or business models we haven't seen yet.

The fashion industry is ready for disruption. The technology is ready. The question is whether the next wave of entrepreneurs will learn from the graveyard of beautiful, broken platforms that came before them.

AI will not fix the business model. But the right business model, enabled by AI, might finally fix fashion discovery.