LinkedIn
YH
Yuliya Halavachova · Founder & Principal Data Scientist at UltraScout AI

Yuliya developed the AI Citation Velocity Index to help brands identify momentum shifts in AI visibility before they become crises — or miss opportunities. 16+ years in Data Science and NLP applied to brand intelligence.

A brand with 200 citations today is not necessarily winning. A brand that had 100 citations last month and 200 this month is accelerating. Citation count tells you where you are. Citation Velocity tells you where you're going.

The Core Insight

AI citation velocity is the leading indicator of future visibility. Brands with accelerating velocity outperform static leaders within 90 days in 73% of cases tracked by UltraScout AI.

1. What Is the AI Citation Velocity Index?

The AI Citation Velocity Index (CVI) measures the rate of change in a brand's citation count across AI platforms over a rolling 30-day window. It answers the question every brand needs to ask: Is my AI presence getting stronger or weaker right now?

Unlike static citation counts, CVI captures momentum — the direction and speed of change. This makes it a predictive metric, not just a descriptive one.

AI Citation Velocity Index =

CVI = (Ct − Ct−30) ÷ Ct−30 × 100

Where Ct = citations this period · Ct−30 = citations 30 days prior

2. CVI Score Ranges

+20% or above Accelerating: Rapid citation growth — strong momentum, content and authority signals firing correctly.
+5% to +20% Growing: Healthy upward trend — maintain current strategy and scale what's working.
−5% to +5% Stable: Plateau — visibility is holding but not growing. Time to identify new citation opportunities.
−5% to −20% Declining: Early warning signal — investigate immediately. Competitor activity or content staleness likely.
Below −20% Collapsing: Crisis — significant algorithm change, competitor displacement, or technical block. Requires urgent action.

3. Why Velocity Beats Count

Brand A: Static Leader

Citations: 850 → 850

CVI: 0% (Stable)

Trend: Plateau — competitors catching up

No Momentum

Brand B: Rising Challenger

Citations: 320 → 480

CVI: +50% (Accelerating)

Trend: Will overtake Brand A in 60 days at this rate

High Momentum

Brand B has 44% fewer citations today — but is the strategically superior position. Any competitive intelligence tool that only shows citation count would flag Brand A as the winner. CVI shows the real picture.

4. Platform-Specific Velocity Patterns

Platform Velocity Sensitivity Avg Response Lag Primary Velocity Driver
ChatGPT Low-Medium 4–8 weeks Training data refresh cycles
Gemini High 1–2 weeks Real-time web index integration
Perplexity Very High 24–72 hours Live web crawl, citation freshness
Claude Low 6–12 weeks Model version updates
Copilot High 2–5 days Bing index freshness

Strategic Implication

Velocity-focused brands should prioritise Perplexity and Gemini optimisation first — both respond to new content within days. ChatGPT and Claude velocity improvements require longer lead times and should be planned 8–12 weeks in advance.

5. Measuring CVI: Data Requirements

Minimum Data Requirements

  • 30-day citation count baseline per platform
  • Consistent query set (same queries both periods)
  • Platform-level breakdown (not aggregate only)
  • Minimum 50 queries for statistical significance

Advanced CVI: Weighted by Query Intent

Weight citations by the commercial intent of each query:

Weighted CVI = Σ(Intent Weight × Platform Weight × CVIquery)

Transactional queries carry 3× weight vs informational queries. This prevents high velocity on low-value informational queries from masking decline on purchase-intent queries.

6. Case Study: SaaS Brand Velocity Turnaround

UK B2B SaaS Platform — 90-Day CVI Recovery

Situation (Month 1):

Total Citations:1,240
CVI:−18% Declining
Root Cause:3 competitor comparison pages published, stale case studies

Interventions Applied:

  • Published 4 new comparison articles targeting specific competitor queries
  • Refreshed 12 case studies with 2026 data and updated schema
  • Added FAQ schema to 8 high-intent product pages
  • Submitted updated sitemap to Perplexity and Bing

Results (Month 3):

Total Citations:1,890 (↑ 52%)
CVI:+41% Accelerating
Time to recovery:47 days from first intervention

7. Improving Your CVI

Fastest Velocity Gains (1–2 weeks)

  • Publish fresh content on Perplexity-indexed domains
  • Update existing high-authority pages with current data
  • Add structured FAQ schema to product and service pages
  • Submit updated sitemaps to accelerate re-crawl

Medium-Term Velocity Gains (2–6 weeks)

  • Earn mentions on high-authority third-party sites (G2, Trustpilot, industry press)
  • Publish original research with citable data points
  • Build comparison content targeting competitor brand queries
  • Optimise llms.txt with updated product and service descriptions

Long-Term Velocity Foundations (6–12 weeks)

  • Build topical authority clusters around your core category
  • Establish entity recognition in AI knowledge graphs
  • Develop proprietary data and indices (AI models cite original research)
  • Implement UltraScout AI platform monitoring for early velocity signals

Key Takeaway

Citation count tells you today's score. Citation Velocity tells you tomorrow's winner. Track CVI weekly — a single month of declining velocity, if caught early, costs one content refresh. Missed for a quarter, it costs six months of recovery.

Track your Citation Velocity Index

UltraScout AI tracks CVI weekly across ChatGPT, Gemini, Claude, Perplexity and Copilot — with automated alerts when velocity shifts.

Explore the Platform → See Pricing →

References

  • Halavachova, Y. (2026). "AI Citation Velocity Index: Measuring Citation Growth Rate in AI Search." UltraScout AI Research.
  • Aggarwal, P. et al. (2024). "GEO: Generative Engine Optimization." ACM SIGKDD 2024. arXiv:2311.09735