Monitoring

Mentioned and cited are different results, so they score differently

80 vs 40

points for a citation against a bare mention, per question per engine

datapointValue in geo/score.go: 0 if absent, 40 mentioned, 80 cited, plus up to 20 for organic rank

Being named in an answer and being cited in one are not the same outcome. A name-drop puts you in a sentence; a citation puts your domain in the source list, where it gets clicked and where the next answer is likely to retrieve you again. Most tools in this category collapse both into one "visibility" percentage.

We score them apart. A datapoint is one question on one engine: nothing if you are absent, 40 if the answer named you, 80 if it cited your domain, plus a bonus for organic rank that tops out at 100.

Where it lives: Visibility and Analytics in the dashboard, and the weekly digest email.

What it looks like

Illustrative
One question

“Best product analytics for small teams?”

Asked on every engine you bought, on your cadence.

  • ChatGPT80Cited
  • Perplexity40Mentioned
  • Gemini0Absent

0 absent, 40 mentioned, 80 cited, plus up to 20 for organic rank. The score on the dashboard is the average across every question and every engine.

One question fanned across three engines. Illustrative scores, on the real 0/40/80 scale from geo/score.go.

What it does

01

Fans out across engines

The starting configuration checks Perplexity, ChatGPT, and Gemini, and each of the three is selectable on its own. Perplexity usually returns up to ten cited sources per answer while Gemini annotates five, so engine coverage changes both the check count and the citation corpus. Grok is sold separately, as a fixed 20 prompts a week: it reads X, but on our own 36-question run it cited X only on sentiment questions and never on a buying or fact-check one. Claude and Copilot are not measured and are not sold.

02

Excludes ungrounded answers from the citation rate

An engine answering from its weights cannot cite anyone. Counting it in the denominator means a product cited by every grounded engine reads as 50% cited the moment an ungrounded one joins the fan-out. The mention rate keeps the full denominator, because being named from memory is a real and separately interesting outcome.

03

Records sentiment and the snippet

Every stored check keeps what the engine actually said about you. An answer that names you unfavourably is a different problem from one that omits you, and it shows up as unfavourable rather than as a win.

How it works

  1. 01Checks follow the purchased prompt count and cadence, with a rolling abuse guard. The surface tells you when a slot frees up instead of failing with "try again later".
  2. 02Every run stores the engine, whether it searched the web, the sources it returned, and the timestamp.
  3. 03A run that changes something can leave the app. The outgoing webhook posts to Slack, Discord, Zapier, Make or n8n, so a score moving does not depend on anyone opening the dashboard.
  4. 04The rank bonus reads the same cached organic position the rank overlap check uses, so the two never disagree.

What it does not do

  • Answers vary run to run. A single check is weak evidence, which is why checks run on a repeating cycle and the score is meant to be read as a trend.
  • We check the engines through their APIs. That is close to but not identical to what a logged-in consumer sees in the chat UI, especially with memory and personalisation on.
  • A score going up does not prove your publishing caused it. It proves the answers changed.

What you get

Engines are picked one at a time in the configurator. See the engines row on pricing.

Read the full plan comparison.

Try it free

One question, one engine, no signup: does ChatGPT name you, cite you, or recommend someone else.

Free ChatGPT Visibility Checker

Live demo

See AI visibility monitoring on a real account, with no signup and nothing to configure.

Open the demo

Frequently asked questions

How is the score calculated?

Average of all datapoints, where each datapoint is one question on one engine: 0 absent, 40 mentioned, 80 cited, plus round(20/rank) when the page ranks, capped at 100. A citation is worth roughly double a mention and top rank adds the most.

Why is my score zero on day one?

Because most products are absent from the answers in their category on day one. That is the common outcome of a first baseline, and it is the number the rest of the product exists to change.

Do you check Google AI Overviews?

Not directly. The rank overlap check reads Brave Search, and the AI checks run against Perplexity and ChatGPT. AI Overviews would need a different collection method and we would rather say so than imply coverage we do not have.

Related features

Monitoring

Prompt library

Auto-generated buyer questions, editable by hand, regenerated when your positioning changes, rotated by staleness.

Monitoring

The mirror

The first-run screen: one buyer-intent question, the competitors the answer named, and whether you were in it.

Citation corpus

Rank overlap

Per cited page, the Google position for the question it was cited for — and a count of the ones that have no position at all.

Every feature we publish

Run it on your own product

Start with the free visibility check to see whether the engines name you today, then run the scan that shows which pages they used instead.

Start your first scanSee the live demo