𝗔𝗜 𝗥𝗲𝗰𝗼𝗺𝗺𝗲𝗻𝗱𝘀 𝗯𝘆 𝗙𝗮𝗺𝗲. 𝗕𝘂𝘁 𝗙𝗮𝗺𝗲 𝗗𝗼𝗲𝘀𝗻'𝘁 𝗘𝘅𝗽𝗹𝗮𝗶𝗻 𝗘𝘃𝗲𝗿𝘆𝘁𝗵𝗶𝗻𝗴.

Most AI conversations focus on visibility. People ask if AI can crawl a site or cite a brand. These questions matter. But they do not explain why AI chooses one brand over another.

We analyzed 20,000 AI product recommendations across five categories:

  • Beauty
  • Supplements
  • Coffee
  • Pets
  • Home & Living

The data included 1,490 brands. We wanted to know what drives these choices.

First, we tested store quality. We used our AI Commerce Score to measure machine readability and technical structure. We expected high quality to lead to more recommendations. The data proved us wrong. Store quality explains only 2.1% of recommendation frequency.

Next, we tested public fame. We looked at Wikipedia readership, media coverage, and brand mentions. Fame explains 24.9% of recommendation frequency. Fame matters much more than store quality. But it still does not explain most results.

We compared the top 50 most recommended brands against the bottom 50. The recommendation gap was massive. The top brands got six times more recommendations. Yet, their store quality scores were nearly identical to the bottom brands.

Is AI behavior random? We ran every query twenty times to check for stability. The results were not random. The same brands appeared in the same spots in 78% to 91% of runs. The system is stable. It is just a mystery.

The findings show a clear split:

  • Visibility asks: Can AI see you?
  • Recommendation asks: Will AI choose you?

Store quality explains 2.1%. Fame explains 24.9%. What explains the other 73%?

It might be trust signals, training data exposure, or semantic authority. We do not know yet. We are investigating it now.

Source: https://dev.to/atom_foundry/-ai-recommends-by-fame-but-fame-doesnt-explain-most-recommendations-3dgh