One of the most disorienting aspects of AEO is that results are unstable. A content piece that's being cited consistently can drop out of rotation without any visible change to the content or the page. We set out to quantify this instability — and the results were striking.
Across 50,000 tracked prompts over a 90-day period, we found that citation sources changed by an average of 41% month-over-month. For some categories, the change was as high as 60%. We call this phenomenon "citation drift."
Sources of citation drift
- Model updates: AI platforms regularly retrain or update models, shifting which sources they prefer
- Content freshness: AI models increasingly weight recency, cycling out older content even when it's high quality
- Competitive dynamics: When a competitor gains authority in a category, it can displace existing citations
- Retrieval system changes: The retrieval layer that selects sources for RAG-based systems changes independently of the model itself
Implications for AEO practitioners
Citation drift means that AEO is not a one-time project. It requires continuous monitoring, continuous content refresh, and continuous competitive intelligence. Brands that treat AEO as a quarterly initiative will consistently underperform those that treat it as an ongoing discipline.
DigitAI's weekly citation tracking is designed specifically to catch citation drift early, giving brands time to respond before visibility losses compound.