Monitoring & metrics
Tracked prompts run automatically against six enabled engines chosen from seven — ChatGPT, Claude, Gemini, Google AI Mode, AI Overviews, Perplexity and DeepSeek, with Claude on by default and no per-engine pricing. LLM engines are collected through official provider APIs — never scraped chat UIs — and Google's AI surfaces (no API exists) through licensed SERP data providers. Self-serve plans scan weekly; Enterprise scans daily with adaptive back-off.
For every answer, Volexi stores the full response and extracts mentions, competitor mentions, fan-outs, sourced pages and citations — repeat citations of a page counted separately.
Scans run themselves
- Your first analysis runs from onboarding (Get started); after that, every scan is scheduled — fixed cadence keeps trend data comparable week over week, so there's nothing to trigger by hand.
- New prompts — and prompts whose text, market or language you edit — are picked up by the next nightly scan, so fresh data lands within about a day.
- During a run, Overview shows “Analysis running — answers arriving live” and refreshes itself as answers land.
- AI answers are stochastic — the same prompt can answer differently run to run. For your highest-stakes prompts, set Sampling (on the prompt page) up to ×5: the scan collects that many answers per engine and Visibility shows a 95% confidence range (“40% ±12”) instead of a single noisy point. Samples draw from your monthly answer budget.
Read and slice the metrics
- On
Overview, read the four headline KPIs — Visibility, Share of Voice, Citation Rate, Average Position — each with its week-over-week delta and sparkline. - Use the filter bar (All engines · All markets · All personas · All tags · 12 weeks) to slice every metric by engine, region, language, persona, tag and date range.
- Open
Docs → Metric formulasin-app for the exact Visibility formula, engine weights and worked examples. - Click Export CSV on any table to download the underlying data.
| Metric | Definition |
|---|---|
| Visibility Score | AI answers mentioning you, weighted by prompt volume × engine weight × mention rate. |
| Share of Voice | Your mentions ÷ all tracked-brand mentions, per engine and overall. |
| Average Position | Your rank among mentioned brands, over answers that mention you. |
| Citation Rate / Share | How often answers cite your URLs, and your share of all citation occurrences. |
| Sentiment | Every mention scored positive / neutral / negative with a quote — trended and benchmarked vs rivals. |
Check sentiment and benchmark competitors
- Open
Competitorsto see Share of Voice, average position by engine, and the sentiment benchmark (you vs each rival). - Click Read the answers behind these scores → to open the exact answers a sentiment score came from — the metric is always evidenced.
- To change the competitor set, use Add competitor or Remove; Volexi automatically re-scores your answer history so trends stay comparable.
If an engine fails during a scan, its answers are excluded and flagged with a Retry on Overview — degraded data is marked, never silently dropped.