Prompt volatility tracking across AI answer engines
Prompt volatility quantifies how unstable an AI engine's answer to a given prompt is. High-volatility prompts need larger n before any reported number means anything, and low-volatility prompts can be sampled cheaply.
How it differs by engine
Derived across the sample set — there is nothing on the page to parse.
Google AI Overviews prompt volatility
Derived across the sample set.
google_ai_overviews
Google AI Mode prompt volatility
Derived across the sample set.
google_ai_mode
ChatGPT prompt volatility
Derived across the sample set.
chatgpt
Perplexity prompt volatility
Derived across the sample set.
perplexity
Microsoft Copilot prompt volatility
Derived across the sample set.
copilot
Gemini app prompt volatility
Derived across the sample set, with grounding_rate reported alongside.
gemini
DeepSeek prompt volatility
Derived across the sample set, with search_rate alongside.
deepseek
Qwen prompt volatility
Derived across the sample set, reported per intent bucket.
qwen
Doubao prompt volatility
Derived across the sample set and across runs over time.
doubao
Kimi prompt volatility
Derived across the sample set.
kimi
Baidu ERNIE prompt volatility
Derived across the sample set.
baidu_ernie
Naver prompt volatility
Derived across the sample set.
naver
Sampling
What the API returns
{
"engine": "google_ai_overviews",
"field": "prompt_volatility",
"prompt": "best ai visibility tracking tool",
"country": "GB",
"language": "en",
"n": 30,
"volatility": {
"jaccard": 0.42,
"mention_variance": 0.11,
"recommended_n": 75,
"band": "moderate"
}
}Response fields
| Key | Type | Notes |
|---|---|---|
| jaccard | float | Mean pairwise citation-set overlap across samples. |
| mention_variance | float | Variance in brand presence across samples. |
| recommended_n | integer | Samples needed for a ±10 point interval on this prompt. |
| band | string | stable | moderate | volatile. |
Prompt volatility FAQ
- What is prompt volatility tracking?
- Prompt volatility quantifies how unstable an AI engine's answer to a given prompt is. High-volatility prompts need larger n before any reported number means anything, and low-volatility prompts can be sampled cheaply.
- How is prompt volatility sampled?
- Volatility is what tells you the right n for every other field. We recommend using it to allocate credits: sample noisy prompts deeply, stable prompts lightly.
- What does prompt volatility cost?
- Derived from samples you already pay for. No separate charge.
Get prompt volatility on your prompt set
Track it in the dashboard or pull it in through the API. Send us the prompts, the brands and the markets, and we'll come back with a per-engine coverage plan and the credit arithmetic behind it.