January 12, 2026
AI agents do not care about your logo. They care about your data.
There is a prevailing fear in the boardroom today: The AI game is rigged in favor of the giants. The logic seems sound. Large language models (LLMs) require massive amounts of data, and legacy brands have decades of it.
Conventional wisdom suggests that global leaders will dominate AI search results, effectively pushing agile competitors off the digital shelf.
But Joe Davis, CEO at Bazaarvoice, argues that the fundamental nature of how AI “thinks” suggests the opposite is true.
We are shifting from the era of search (keywords and endless scrolling) to the era of agentic shopping (bots that research and recommend). While the technology is new, the consumer has already moved.
Our latest research on generative engine optimization (GEO) shopping found that 83% of shoppers have used an AI-powered tool in the last six months, and 50% feel it improves the research phase of their journey.
Now, brand legacy matters less than data transparency.
The death of brand recall and the rise of attribute relevance
Before we break down the buckets, we need to understand the economic engine driving holiday In traditional search engines, dominant brands won through mental availability. We type “running shoes” and click the most recognizable logo. It is a game of memory.
But AI agents are immune to nostalgia. They don’t prioritize a brand because of a Super Bowl ad or a decades-old tagline. Instead, they prioritize the specific parameters of a user’s prompt.
Big brands have historically relied on broad, mass-appeal marketing claims like “Just Do It” or “Open Happiness.” AI agents cannot process that level of abstraction. They require granular, attribute-level data.
Davis notes that this shift in mechanics changes the definition of “market power.” “The nature of an AI prompt, as opposed to a Google search, often leads to a more detailed and nuanced request. To respond accurately to a detailed prompt, an LLM model needs deep product details in order to make a good recommendation,” he explains.
Consider the shift in discovery:
- Old search: “Best face cream” (Advantage: Big brands).
- New prompt: “Find me a fragrance-free face cream under $40 that works well under makeup for dry skin and doesn’t pill.” (Advantage: Challenger brands).
The AI is not looking for market share. It is looking for a semantic match for fragrance-free, under $40, and no pilling.If a challenger brand has authentic reviews that specifically mention these niche details, and a legacy giant relies on generic marketing copy, the AI will recommend the challenger. Every time.
Authenticity is the API
If the AI is the engine, data is the fuel. We are entering a data war where the differentiator is no longer volume, but quality.
This creates a unique challenge for AI models: they are fundamentally risk-averse. To avoid “hallucinations,” (the industry term for AI-generated errors) these models are hungry for human validation. They want to cite sources to prove they’ve “done the homework.”
Shoppers trust the word of other shoppers more than most advertising, and all of the major LLM models appear to have that same trust
Joe Davis, CEO, Bazaarvoice
Our research validates this “people-first” approach. Despite the AI hype, a trust deficit remains. Only 12% of shoppers trust AI-generated summaries, while 45% still trust human reviews. To bridge this gap, the AI must prove its recommendation is rooted in real-world experience.
For a marketing manager, the strategy is clear. You are no longer just optimizing for a human skimming a page for five seconds. You are optimizing for a machine that reads every word of every review to validate a specific use case.
The mandate for 2026: Optimize for specificity
The transition to agentic shopping isn’t coming; it’s here. Our data shows that 73% of consumers feel comfortable researching fashion and beauty products through an AI chatbot.
| About the data: Cited from the GEO Shopping Bazaarvoice Consumer Survey, November 2025. N=3,093 adults (A18+) in the US, APAC, and EMEA. This research explores how consumers utilize AI-powered tools, including generative AI (ChatGPT/Gemini), recommendation engines, and voice assistants, throughout the shopping journey. |
So, how does a challenger brand capitalize on this shift? By focusing on three tactical inputs.
Audit for attributes, not just sentiment
Stop looking at your star rating as the primary metric. Start looking at the attributes within your user-generated content (UGC).
“If you want to raise the odds that an AI agent will pick your product, you want both more reviews and more detailed reviews,” Davis advises.
Do your reviews mention fit, texture, longevity, and use cases? If the data isn’t in the reviews, the AI won’t find it.
Feed the machine the truth
In the AI era, perfection is a red flag. AI agents are programmed to be skeptical of anomalies. This human instinct is now coded into machine logic.
Davis notes that human shoppers have always sought out friction as a trust signal. “Personally, the one thing I never skip are the worst reviews … The level of discontent I see in the bad reviews gives me the best insights,” he explains.
AI models use “discontent signals” as a form of bot detection. An AI agent scanning a product with 5,000 five-star reviews and zero nuance flags it as “synthetic.” A product with debate, nuance, and occasional dissatisfaction of real customers looks “human.” If your review profile lacks these, the LLM’s safety filters may flag your entire catalog as bot-generated spam.
Ensuring your reviews are free of bot spam isn’t just an ethical choice anymore; it is a visibility strategy. To be chosen by AI, you must first prove you are real.
Break out of the walled gardens
Retailers like Amazon or Walmart often build “walled gardens” to protect their proprietary data, ensuring the AI only trains on their specific inventory.
However, brands need their data to be “free” and portable. If your review data is locked within a single retailer’s ecosystem, you remain invisible to agnostic LLMs like ChatGPT or Gemini.
“Retailers are reluctant to give over their proprietary data to a model that might cause a sale to happen at a competitor,” Davis explains.
This is a business strategy issue, not just a technical one. As a brand manager, you must fight for syndication. You must ensure your authentic content is accessible everywhere the AI is looking.
The North Star
The era of buying your way to the top of search results is fading. In an agentic world, you earn the recommendation through data density.
For challenger brands, this is the open door. You do not need a global advertising budget to generate thousands of detailed, authentic reviews that describe exactly how your product solves a niche problem. You just need a community.
“Brands still need to understand what it is that consumers are looking for and try to provide that,” Davis concludes.
The mission has not changed, but the tools have. And for the first time in a decade, the tools favor the agile.
Ready to get your products discovered, trusted, and chosen by AI and humans? Check out the blog below.
