If you’ve been trying to hold a specific AI citation position, Ahrefs has some uncomfortable news: the volatility you’re seeing isn’t a ranking signal you can optimize away. It’s a structural feature of how AI answers work.
Ahrefs researcher Mateusz Makosiewicz tracked 37,500 AI citations and pulled out 10 findings. The headline takeaway is direct: treat citation volatility as normal, and aim to appear more often over time rather than trying to hold every citation without interruption.
What the data says
The study found that AI citation instability appears to be a built-in trait of AI answers. Citations can shift significantly with model updates, meaning a source that gets cited consistently today may drop after the next model version ships. No specific percentage drop was shared for individual citations, but the pattern was consistent enough across 37,500 tracked citations to treat it as the default behavior rather than the exception.
The operator implication
For solopreneurs and content operators building for AI search visibility, this changes the measurement question. Instead of asking “am I cited on this query today,” the better question is “how frequently do I appear across this topic cluster over a rolling period.” Frequency and breadth of coverage matter more than holding a single citation slot.
Model updates are the single biggest disruption factor the research identified. If your citation count drops after a model update, the data suggests that’s expected behavior, not a signal that your content degraded.
The full 10-point breakdown is available on the Ahrefs blog.
