The Sole-Mention Rate: Why Most AI Recommendations Are Unique to One Engine

The finding

Across 56819 company-question pairs where at least one AI engine made a recommendation, 50209 pairs (88.4 percent) were named by only one engine. The other two engines did not mention the company at all for that question.

This is not a finding about edge cases. Sole mentions are the dominant pattern, not the exception.

What this means

When ChatGPT recommends a company to a buyer, that recommendation reflects ChatGPT's view, not a consensus across AI assistants. Nearly nine in ten recommendations do not transfer to other engines.

For a company tracking its AI visibility, this creates a measurement challenge. A high score on one engine may coincide with invisibility on others. The score describes performance on that platform, not a general property of the brand's AI presence.

For buyers using AI assistants, this means the recommendations they receive depend heavily on which assistant they ask. A buyer using Claude would see substantially different company recommendations than a buyer using Gemini for the same question.

Why engines disagree

The study cannot determine why engines disagree, but several mechanisms are plausible:

Training data differences. Each engine was trained on a different corpus. A company prominent in ChatGPT's training data may be absent from Claude's. Neither engine is wrong; they have different information.

Retrieval system differences. Modern AI assistants often augment their base knowledge with real-time retrieval. Different retrieval systems surface different sources, leading to different recommendations.

Sampling from a large space. For many buyer questions, dozens of companies could be reasonable recommendations. Each engine may be sampling from this space, producing valid but non-overlapping lists.

The contrast with search engines

Traditional search engines also produce different results for the same query, but the overlap tends to be higher for competitive terms. The top results for "best CRM software" on Google and Bing might differ in order but often include many of the same companies.

The 88.4 percent sole-mention rate for AI engines suggests a different dynamic. These engines are not ranking a shared list differently; they are often constructing entirely different lists.

Implications for visibility strategy

A company that optimizes for ChatGPT visibility may see no improvement on Claude or Gemini. The strategies that work on one platform may not transfer, because the engines are drawing on different information and applying different reasoning.

This argues for platform-agnostic presence building: ensuring your company appears in sources that multiple engines are likely to trust, rather than chasing signals that one engine rewards.

What the study measured

The study examined 2500 prompts across 30564 companies. For each prompt, all three engines (ChatGPT, Claude, and Gemini) were queried, and every company named by at least one engine was recorded. The 88.4 percent figure represents the share of those company-prompt pairs where only one engine made the recommendation.