Research · Q3 2026 · Research baseline
AI Answer Volatility Index
How much do AI answers and citations shift between runs? This public index summarizes engine-level volatility so you know when a single scan is noise — and when repeated sampling is mandatory.
Composite volatility
44/100
moderateWhy it matters
Brands that react to one-off scan deltas waste budget. Volatility-aware teams set minimum sample sizes, use confidence intervals, and monitor source portfolios — not just mention counts.
Engine breakdown
| Engine | Answer vol. | Source vol. | Risk |
|---|---|---|---|
ChatGPT Answer phrasing shifts frequently; citation sets moderately stable on evergreen queries. | 42% | 38% | moderate |
Perplexity Real-time retrieval drives higher source churn between runs. | 55% | 61% | volatile |
Gemini Grounding varies by query class; commercial prompts show higher variance. | 48% | 44% | moderate |
Claude Lower citation surface; mention presence more stable on branded queries. | 36% | 29% | stable |
Key insights
- Single-run AI visibility checks are directional — repeated sampling with confidence intervals is required before claiming movement.
- Source volatility exceeds answer volatility on retrieval-heavy engines (Perplexity, grounded Gemini).
- Branded queries are more stable than category/commercial prompts — prioritize high-intent commercial clusters for monitoring.
- Volatility spikes often precede competitor citation shifts, not just mention drops — watch source portfolios, not mentions alone.
Composite score averages answer + source volatility baselines from AEOArc sampling methodology (bootstrap CI, multi-run aggregation). Public index uses anonymized research baselines; customer dashboards use project-specific volatility from scan history. This index is a market signal, not a guarantee of future engine behavior. AI answers change without notice.