Everything CiteGraph does, one page per capability.
Each page describes one capability the way we’d want it described to us: what it does, how it works, and — in a section most feature pages don’t have — what it deliberately won’t claim.
Measure
Prompt tracking
Buyer questions mined from your own site, run 3–5 times on each of 4 engines weekly — score ring, per-engine record, and a margin of error on every question.
Answers inspector
Every AI answer stored verbatim: engine, full text with brand mentions highlighted, and each cited source. The evidence under every CiteGraph number.
Share of voice
How often AI answers name you versus each competitor, per engine, trended weekly — with a margin of error on every rate, so real moves never blur into noise.
Error bars
AI engines answer differently run to run. CiteGraph asks each question 3–5 times on each of 4 engines and prints a margin of error on every rate — no single-run tool can.
What moved
When your AI visibility moves, CiteGraph names the cause: the questions that flipped, rank movement, a top-mover card — and only speaks past the printed margin.
Diagnose
Blind spots
Every buyer question where AI names a competitor instead of you: the full verbatim answer, the winner, the cited page, and the one sentence missing from yours.
Citation tracking
Every page AI engines cite for your category, deduplicated, graded for quality and gettability — plus citation targets: the pages answering the questions you lose.
Competitor discovery
Every scan reads its own answers and surfaces the brands AI actually recommends in your category — alias-folded, counted, and one click from tracked.
Framing analysis
Every mention classified — recommended pick, neutral listing, or caveat — charted as stacked bars over AI's verbatim phrases. Auditable counts, no fake 0–100 score.
Act & report
Plays
Every diagnosed loss becomes a play: the move, the target, the mechanism, and an AI-drafted artifact — paragraph, row, or pitch — ready to edit and ship.
Fact sheets
A sheet of the concrete, checkable claims AI answers are built from — pricing, limits, comparisons — drafted from your scan so your pages say the citable sentence.
Dossiers
Every scan produces a dossier: headline rates with margins, verbatim lost answers, cited sources, and plays — shareable by link, white-label on Scale.
Free layer
Per-engine trackers
The engines disagree — in our measured boards their top-5s can share as little as 2 of 5 brands. One page per engine on what each one does differently and how we track it.
ChatGPT visibility tracker
Track whether ChatGPT-family models name your product for real buyer questions — repeated runs, verbatim answers, cited sources, a margin on every rate.
Gemini visibility tracker
Track whether Google's Gemini names your product for real buyer questions — repeated grounded runs, verbatim answers, cited sources, margins on every rate.
Perplexity visibility tracker
Track whether Perplexity names your product and cites your pages — repeated runs, verbatim answers, resolved sources, and margins on every rate.
Claude visibility tracker
Track whether Anthropic's Claude names your product for real buyer questions — repeated grounded runs, verbatim answers, cited sources, and a margin of error on every rate.
See all of it on your own domain in about a minute.
Check your site →