New research: I analyzed 70K+ AI responses over 4 weeks to see how often the same brands are named for the same prompts.
This is my second time analyzing AI volatility, where I was originally inspired by
@randfish's call for more data on the topic.
Some highlights:
- The "core" of a response barely moves, but the tail turns over daily
- The usual leading brand is named first about half of the time, but when it isn't, it's often missing entirely
- While cited URLs change frequently, the mix of page types remains fairly stable
- Finding exact brand or citation list matches was rarer than you might expect, but it's a deliberately strict test
This isn't the biggest study out there, but I also think its size is closer to what a typical brand would track for themselves, which was the main idea.
The link is on the Detailed homepage, or you can also find it in the replies below.
Cheers! 🙌