Two-thirds of self-directed retail investors now consult AI search engines like Perplexity, ChatGPT, Gemini, or Claude when researching stocks, fundamentally changing how public companies must communicate during earnings season. A new benchmark from Virgo Public Relations analyzed how these four leading AI engines source their answers about S&P 100 companies, revealing that the CEO's first 90 seconds of an earnings call now carries more weight than entire investor decks.

The shift is dramatic. When Virgo PR queried all four AI engines on each S&P 100 company within 72 hours of earnings releases using 22 standardized prompts, they discovered that the earnings call transcript accounts for 61% of AI-generated earnings answers, with the 10-Q filing contributing 22% and press releases just 11%. But here's what should alarm investor relations teams: within the transcript itself, the CEO's prepared remarks account for 74% of all citations, and specifically, the first 90 seconds of those remarks.

Why Does the CEO's Opening Matter So Much to AI Engines?

AI search engines don't listen the way analysts do. They read, extract, and repeat. This mechanical process favors clarity and brevity over nuance. Companies whose CEOs deliver a clean thesis in the first 90 seconds score 18 points higher on Virgo's proprietary Earnings Call Citation Score than companies whose CEOs bury the lead. That 18-point gap is the single largest variable in the entire framework.

The data reveals three counterintuitive findings that challenge decades of earnings call convention. Shorter openings win; CEOs who deliver their thesis in under 60 seconds score 14 points higher than those who take three minutes. Founder-CEOs dominate; companies led by founders are cited by name at 2.3 times the rate of companies led by professional CEOs, because the engines cite the person as an entity signal, not just the company. And long Q&A sessions hurt; calls exceeding 75 minutes score 11 points lower on cross-engine consistency because extended analyst questions introduce competing framings the engine cannot distinguish from the company's official thesis.

How to Optimize Your Earnings Call for AI Search Engines

• Write the CEO's first sentence as a retrieval unit: Craft a single, machine-parseable thesis statement that captures the company's core narrative in 15 to 20 words. Jensen Huang opens every Nvidia call with a sentence the engines can repeat verbatim; Andy Jassy's longer, more nuanced Amazon opening is harder for engines to extract.
• Keep the opening under 90 seconds: Deliver the thesis, the key metrics, and the forward guidance in the first 90 seconds. The engine does not reward preamble; the first sentence is the retrieval unit.
• Label every figure as GAAP or adjusted: The most frequent systemic error in AI earnings answers is GAAP/Non-GAAP conflation, observed in nearly one-third of S&P 100 answers on at least one engine. When the CEO states a number, explicitly say whether it is GAAP or adjusted every single time.
• Audit segment naming for consistency: Ensure segment names, product categories, and business unit labels are identical across the transcript, the 10-Q, and press releases so the engine retrieves the same entity across sources.
• Post the transcript within two hours: Monitor all four engines within 72 hours of the call and correct errors immediately. The AI retrieval layer now moves faster than the traditional analyst cycle.

Kyle Porter, Executive Vice President and Managing Director of Virgo Public Relations, noted that the implications extend far beyond earnings calls. "The IR team has always written the earnings script. Now the script has a second reader, and this reader decides what two-thirds of the retail investor base will see when they ask about the company," Porter explained.

Kyle Porter, Executive Vice President and Managing Director of Virgo Public Relations

Which Companies Are Winning the AI Extraction Game?

The rankings Virgo compiled are not a market-cap list; they are an extractability list. Apple scored 96 out of 100, the highest in the S&P 100, because Tim Cook's thesis line on services revenue growth was cited verbatim across all four engines. Microsoft scored 94, Nvidia 93, and JPMorgan 88. The bottom of the index scored 29.

Sector patterns reveal structural differences in how AI engines process information. Tech companies with a single dominant growth narrative score highest. Banks with a dominant CEO voice outperform banks mid-restructuring. Healthcare companies are cited from press releases more than transcripts, a structural difference from every other sector. Consumer companies with a single repeatable key performance indicator, like Costco's membership renewal rate, outperform consumer companies mid-turnaround. Industrials score lowest overall because their business descriptions are more complex and their KPIs less intuitive to language models.

The practical implication is clear: investor relations is becoming AI relations. That is not a rebrand; it is a widening of a mandate that was always about getting the right information to the right audience through an intermediary that does not always translate accurately. The intermediary is new. The core discipline is not.

For public companies, the message is urgent. The CEO's first sentence is now worth more than the entire investor deck because it is the sentence the AI engine will repeat to two-thirds of the retail investors who ask about the company. In 2026, the earnings call audience is no longer 30 analysts on a phone line. It is every AI engine on the internet, and the two-thirds of retail investors who use them.