AI-Mediated Commerce Isn't For Everyone

Written By Michael Quinn

Imagine. Someone asks a mutual friend of yours where they can buy something you make. They might recommend you, but you’ll never know if they did or how they described you. What you do know: your mutual friend will recommend 5-10 of your competitors, but you won’t know which ones or what was said about them.

This is not an imaginary problem. About half of US consumers are already using AI to research categories, brands, and products (McKinsey poll, 2025). Most brands are either unaware or unprepared to manage if or how they are included in AI recommendations. This is AI-mediated commerce, not for a time in the future when we trust machines to buy things for us, but a time now when we are trusting machines to inform, compare, and rationalize what to buy.

Brand AI Report is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber.

The good news is that brands can have an impact on how AI “sees” them. The bad news is that impact is indirect, imprecise – inclusion based on facts and consensus in AI, or exclusion based on their absence. And it can take months of human effort for brands to be competitive in AI, to establish their position, or to ward off competitors who see the opportunity and are quietly acting on it.

Most marketing categories in AI have an “anchor brand”. AI will almost always include them, depending on the prompt. They are usually the oldest, the most prolific, or the industry reference other brands are compared to. Then, as with a target, there is a circle of peers, brands that may not be the oldest or quite as revered as the anchor but AI trusts enough to recommend fully, a solid #2, depending on the prompt; they too usually make the list, along with a few other brands. For everyone else in the category, it’s a real land-grab for those few remaining spots in an AI’s response. Most brands are unaware they’re in this fight, let alone what they can (and cannot) do to survive it.

Most consumer prompts about brands fall into one of 3 buckets of Intention:

1. Research/Discovery: “Teach me about this category.”

2. Brand Comparison/Inquiry: “What brands compare well with….?”

3. How to Buy: “Where can I buy….?”

Based on the intention, the AI system cobbles together a response from a hierarchy of sources unique to that system and that intention. Some sources are already crawled, indexed, or cached, and easy for the system to reference; others require live web retrieval to verify, especially for ‘How to Buy’ prompts. Not all intentions use the same sources to form a response. And sources that are used are weighted differently, depending on the intention. For example, a prompt about the History of Hermès, a prompt about what brands compare to Hermès, and a prompt about where you can buy a Hermès jacket will all call upon different sources.

The good news is that there are only roughly 8 “Tiers” of sources most AI will often look to for data. Each tier is defined by the amount of trust AI assigns them. For example: Wikipedia and Wikidata score very high for trust because they are managed by 3rd party volunteer peer-editors. When asked about the history of Hermès, most AI systems will look there for an authoritative response, as well as a few other sources that carry less weight. It’s much less likely to look to Reddit or other forums, but those sources carry more weight when frequency, recency, and relevance are essential for a trustworthy response. For example, “How do I buy a Hermès jacket?” will often look to Hermès.com and other e-commerce feeds.

The tiers have other distinguishing characteristics. They offer different levels of direct editorial involvement and different timeframes for changes appearing in AI responses. Also, different prompt intentions will reorder which sources AI trusts most. The following table reflects the hierarchy for a category-level research prompt, “Tell me about….”. For a more commercial prompt, like “Where can I buy…?”, an AI system integration with, say Google Merchant Center or a Shopify storefront, carries maximum authority. Sources and timeframes vary widely by model, market, crawl frequency, and whether the system uses retrieval. So, this list is illustrative, not exhaustive:

Conflicting descriptions or facts about a brand across sources also impact whether or how AI describes it. Consistency and consensus are of paramount importance, not only for being included and described correctly in responses, but also for preventing competitors from structuring their qualifications more credibly and consistently (and so, framed more confidently by AI).

For example, when an AI system can only find self-published marketing language, it can hedge, using language like “The brand is described by some sources as...” or “According to its own website, the company offers...” These are the specific “red flags” that signal to a buyer that the AI doesn’t fully trust the brand yet.

This is the gap between being mentioned (hopefully) and being recommended with conviction to consumers using AI systems. Even so, the value to brands is actionable and measurable. Anchors can be challenged, peer groups can expand, and inclusion in AI recommendations is within reach. For qualified competitors, it is mainly a matter of being intentional, systematic, and patient with the process. The real risk to a brand is assuming that AI will both: include them when asked for recommendations and describe them accurately compared to their competition.

Onto the AI Headlines for Brand Leaders and Boards:

AI in Marketing, Sales, and Advertising

  • IBM-NRF Study: Brands and Retailers Navigate a New Reality as AI Shapes Consumer Decisions Before Shopping Begins – IBM Newsroom

  • Reddit and AI search are reshaping how brands get found – Marketing Tech News

  • ChatGPT Gets Ads: Omnicom, WPP, and Dentsu Line Up Brands for OpenAI Pilot – www.adweek.com

  • OpenAI exec wants advertisers to ditch agencies and just prompt ChatGPT for ads – PPC Land

  • Selling in the Age of Algorithms: Why Being Human Is Your Greatest Competitive Advantage – The AI Journal

  • Why Authenticity Still Wins in an AI-Driven Marketplace Full of Numbers – AutoGPT

Workforce Impact and Organizational Readiness

The News Media and AI Search

  • Google hit by European publishers’ complaint to EU over AI Overviews – Reuters

  • Few say Americans have a responsibility to pay for news – Pew Research Center

  • How The New York Times uses a custom AI tool to track the “manosphere” – Nieman Lab

  • What does it mean to ‘do your own research,’ and how often do Americans do it? – Pew Research Center

AI Ethics, Policy, and Global Trends

  • A Long Road Ahead: Effective Remedies for Artificial Intelligence Harms in the European Union – Center for Democracy and Technology

  • When Tech Hubs Fail Ethics: How Shillong Built Accountable AI – SmarterArticles

  • Exponential View #560: $1 trillion panic; decimal upgrades, exponential outcomes; attention vs intention; wars, CAR-T therapy & time++ – www.exponentialview.co

Agentic Commerce and Practical AI Implementation

Michael Quinn
Next

Rebranding? Don’t Forget AI, Because AI Won’t Forget You.