Citation Velocity Score
Also known as: Score de Vélocité de Citation
The Citation Velocity Score (CVS) is a proprietary Rankeo metric that measures the rate at which a domain accumulates AI citations compared to its historical baseline. Unlike volume metrics that measure total citations, CVS measures acceleration — the derivative of citation growth.
Definition
The formula is elegant: CVS = (citations in the last 30 days) / (average monthly citations over the previous 90 days). A score above 2.0 indicates a Rising domain (algorithmic amplification in progress). Between 1.0 and 2.0 indicates Steady state. Below 1.0 indicates Declining trajectory with risk of sustained delisting if repeated across consecutive months.
CVS is a leading indicator, not a lagging one. A domain's CVS today predicts its citation volume 60-90 days from now, making it the first metric in SEO history that enables predictive rather than reactive optimization.
How It Works
AI engines like ChatGPT, Perplexity, Gemini, Claude and Grok operate on a "rising sources" bias since early 2026. When an algorithm detects that a domain is accumulating citations faster than its historical rate, it amplifies the signal — citing that domain more often in similar queries. This creates a self-reinforcing loop that lasts 30-90 days, called the Rising Window.
The mechanism was introduced to solve the "stale authority" problem: established sources (Semrush, Ahrefs) have enormous citation volumes but stable rates. Rising domains, even with lower volumes, signal freshness and relevance that users expect. Citation Velocity Score quantifies this dynamic and makes it actionable.
Practical Example
In March 2026, a Rankeo client had 14, 18, and 23 AI citations across their 20 reference prompts over three months. Their baseline was 18.3 citations/month. In April, they reached 47 citations.
CVS calculation: 47 / 18.3 = 2.57x
This placed them in the Rising Zone. The predictive read: "Your May volume will be approximately 90-120 citations due to algorithmic amplification compounding."
Actual May result: 108 citations — within the predicted range.