Watch: AI Brand Authority Index: Ranking 2.9 Million Brands by Associative Embeddedness in Gemini’s Memory
This research presents a methodology for quantifying brand authority in large language model memory using Personalized PageRank and directed association graphs.
Transcript
If you ask an artificial intelligence model to name one hundred brands at random, it will not actually be random. Instead, the AI reveals which brands occupy the most real estate in its memory.
To study this, researchers prompted Google’s Gemini model two hundred thousand times to name random brands. Giants like Google, Microsoft, and Nike topped the list. After filtering out a significant amount of gibberish and machine errors, they used these primary brands as seeds to build a massive association network. They asked the AI to name brands associated with the seeds, and then brands associated with those, eventually mapping a web of nearly three million names.
By applying a customized version of the famous PageRank algorithm, the researchers calculated a brand authority score. This algorithm simulates a journey through the network, prioritizing the brands the AI recalls most frequently.
The final scores do not just measure simple popularity. They measure how deeply embedded a brand is within the AI's neural connections. For instance, the luxury fashion label Maison Margiela was never recalled unprompted, but it ranked incredibly high because it sits at the dense intersection of so many other high-profile fashion brands. In the age of AI, this methodology offers a brand-new way to map digital influence.