Marketers are confronting a new discovery problem: AI assistants don’t simply rank brand content lower or higher—they decide whether a brand gets mentioned at all. That change is pushing companies to hire dedicated heads of AI search, move budgets to influence how chatbots describe them, and demand better measurement standards.
The Interactive Advertising Bureau is assembling a framework to standardize AI visibility measurement, but the industry still lacks an agreed definition of trustworthy data or a reliable driver of recommendations. Reddit’s sudden drop in ChatGPT recommendations has added urgency, showing brands why they need to act on multiple fronts at once.
No single playbook across models
Citation behavior differs sharply by platform. Tinuiti data for July showed Amazon accounted for 4% of Microsoft Copilot citations, just 0.1% on ChatGPT and zero on Gemini. That Copilot lift materially affected Amazon’s average citation share across product categories.
“We’re now influencing discoverability across a multitude of surfaces,” said Angela Seits, vp strategy at Dept. “We’re focused on guiding clients to think about how they’re building content for all the ways that people interact with their brand.”
LLMs have clear source preferences
The same question about sneakers may cite two different brands depending on which model answers. A Semrush study of U.S. data across 22 categories found ChatGPT leans on community and reference sources, with Reddit as its top citation at 28.9%. Google AI surfaces skew toward video and social: YouTube leads at 21.1% and Facebook follows at 17%.
- Amazon: 4% of Copilot citations, 0.1% of ChatGPT citations, zero on Gemini.
- ChatGPT: Reddit is the top cited source at 28.9%.
- Google AI: YouTube and Facebook are the two most-cited sources.
- YouTube’s reach: Google AI Mode, AI Overviews and Perplexity cite it regularly; ChatGPT and Copilot do not.
“There is no silver bullet,” said Gabe Feldman, co-founder of The Now Agency. “It’s YouTube, schema on your website, PR, creators, publishers—all working together.”
Trust is rising faster than measurement
Adobe’s July 2026 consumer survey of more than 5,000 U.S. respondents found 95% of AI users rate its answers at least as trustworthy as a traditional search engine, with 29% calling it very trustworthy. Among regular users, that figure climbs to 97%.
As a result, brands are pre-checking whether creators already appear in AI-generated answers before campaign events, building internal tools to track which sources get cited and why, and running executive content through the same visibility tests—an emerging discipline sometimes called AI engine optimization (AEO).
“We’re going to see AEO have the same seat at the table as some of the largest media channels,” said Crystal Duncan, evp of brand engagement at Tinuiti.
Why measurement remains messy
The same prompt can return different answers nearly every time, and model updates can shift a brand’s visibility overnight. That instability has pushed the IAB to build standards for trustworthy data, even as vendors disagree on what a reliable score looks like. In an IAB survey of 200 brand and agency decision-makers, 45% said comparing AI-driven and traditional customer journeys is their biggest challenge.
“The market is messy,” said Caroline Giegerich, vp of AI at the IAB. “You could run ‘what’s the best sneakers for under $200?’ and get different results 100 times.”
What marketers should do now
Treat AI visibility as a distinct channel rather than an extension of SEO. Map which models matter for your category, audit citation sources by platform, and track answer stability over time. Brands with long consideration journeys, such as insurance and automotive, are already using shifting search behavior to make the case for additional media investment.
Until standards mature, the practical play is to diversify: build for community platforms where ChatGPT looks, invest in video and social sources that Google AI cites, and keep site schema and PR assets clean because no single surface will win across every model.
Source: Digiday
