Why AI Assistants Recommend Brooks Over Nike

When researchers queried ChatGPT, Claude, and Gemini for running shoe recommendations, a pattern emerged: Brooks, a relatively small technical brand, appeared consistently across all three platforms. Nike, the world's largest athletic brand, surfaced far less reliably.

That discrepancy reveals a fundamental shift in how brands compete when AI systems mediate product discovery. The advantage no longer belongs to brands with the biggest marketing budgets or strongest emotional narratives. It belongs to brands that AI can interpret—those whose value translates into attributes and evidence that algorithms can process and recommend.

Researchers at Georgetown University's McDonough School of Business and the University of Virginia's Darden School of Business analyzed more than 1,000 brand mentions across 716 unique brands in 15 retail categories. Their findings, first reported in Harvard Business Review, show that traditional brand-building advantages are losing relevance in AI-mediated environments.

Why it matters

As AI assistants become the primary interface for product discovery, brands face a new competitive bottleneck. The question is no longer whether consumers remember your brand, but whether AI systems can retrieve it as a credible solution to a specific problem. This shift requires fundamental changes to how companies structure product information, cultivate third-party validation, and coordinate across marketing, product, and engineering teams.

What makes brands interpretable to AI

The research identified four patterns that explain why many recognizable brands fail to appear in AI recommendations:

Fragmentation across platforms. Only 8.4% of brands appeared consistently across ChatGPT, Claude, and Gemini. Most surfaced on just one platform, suggesting that visibility alone doesn't determine inclusion.

Inconsistent positioning. Among brands appearing on multiple platforms, 55% were framed differently across systems. AI doesn't reproduce brand messaging—it infers positioning from available third-party information.

Query-dependent competition. Exploratory queries generated 95% more brand mentions than goal-oriented queries, and only 11% of brands appeared in both types. The way consumers articulate their problem determines which brands the AI considers.

Inclusion as the real filter. Once a brand appears in an AI response, 78.7% of mentions carry positive sentiment. The competitive challenge isn't managing tone—it's getting included in the first place.

Brooks succeeded because it built what the researchers call an "interpretable brand." Under CEO Jim Weber, the company narrowed its focus to technical performance, invested in biomechanical research, and developed technologies like GuideRails and DNA LOFT cushioning to address clearly defined problems. Critically, Brooks cultivated an ecosystem of coaches, clinicians, and specialty retailers who could explain those solutions in precise terms.

AI systems favor this approach because they construct recommendations by working forward from a user's condition to product requirements to brands that satisfy them. Brands appear when models can build a clear chain connecting user needs to measurable attributes to supporting evidence.

Three practices for AI recall

The researchers propose a new metric: AI recall share, measuring how reliably a brand is retrieved when its attributes match a query. Building AI recall share requires three shifts:

Replace subjective claims with verifiable specifications. AI systems struggle with vague positioning. "High quality" means nothing; "1,000-cycle durability, ISO-certified" gives the model something actionable.

Cultivate independent validation. Brooks spent 20 years building relationships with specialty running stores, coaches, and podiatrists who needed to explain product choices. That third-party credibility now powers AI recommendations.

Shift from symbolic to evidentiary structure. Many trusted brands underperform in AI recommendations because their strengths are expressed through emotional positioning and lifestyle associations rather than attributes and evidence that models can process.

The brands positioned to win in AI-mediated discovery aren't necessarily investing most in AI technology today. They're the ones that have spent years building a body of evidence that makes them easy to retrieve—often because they were built for human experts who needed to explain choices, which turns out to be the same thing AI systems require.

The research was conducted by John Gale, Luca Cian, and Luc Wathieu and first reported in Harvard Business Review.