Find out why AI skips your page.

Vantage asks ChatGPT, Gemini and Perplexity your buyers' questions, counts who gets cited, and tells your coding agent what to change.

Check your site free Add to your agent See plans

30 checks a month free. No card. Works in Claude, ChatGPT, Cursor and VS Code.

A real Vantage reading from 10 November 2026: ChatGPT was asked 3 times "how to check if chatgpt cites your website". help.openai.com was cited in 3 of 3 answers, semrush.com 2 of 3, ahrefs.com 2 of 3, and vantagemcp.dev in none; on 5 November it was cited in 1 of 3. Gemini, 10 November 2026, same question, 3 answers: frase.io was cited in 3 of 3, zdnet.com 2 of 3, speaklocal.com 1 of 3, and vantagemcp.dev in none. Perplexity, 9 November 2026, same question, 3 answers: ten sites, including vercatus.com, searchable.com and citeflow.io, were each cited in 3 of 3, and vantagemcp.dev in none.

Press an engine on the meter to see which sites its answers cited, and whether ours was one of them. Real readings of our own guide, 3 answers each: ChatGPT and Gemini on 10 Nov 2026, Perplexity on 9 Nov 2026.

Add it where you already work.

Pick your app. The first time, you sign up with your email right inside it. No config files.

Cursorone click VS Codeone click Claudepaste one link ChatGPTBusiness and Enterprise

Using Claude Code? Copy this, paste it into your terminal and press Enter. Then start Claude Code and ask it where your site is cited.

claude mcp add --transport http vantage https://vantagemcp.dev/mcp

Any other MCP app: paste this block into its settings with your API key.

.mcp.json

{
  "mcpServers": {
    "vantage": {
      "url": "https://vantagemcp.dev/mcp",
      "headers": { "Authorization": "Bearer YOUR_API_KEY" }
    }
  }
}

This is a real, verifiable listing

Vantage is published to the Official MCP Registry under a domain-owned namespace, not a self-reported claim. This opens the registry's own record for it directly (raw JSON, not a styled page).

dev.vantagemcp/vantage

New to MCP? What this actually means

MCP stands for Model Context Protocol - an open standard, introduced by Anthropic in 2024, that lets an AI assistant connect directly to outside tools and pull in real, live information instead of only working from what it already knows. Vantage is one of those outside tools: once connected, your AI assistant can check your actual AI-search citations on request, the same way it can already browse the web or read a file for you.

Click your app's key above (Cursor, VS Code), or copy the link into Claude's or ChatGPT's settings
When it asks, sign up with your email (or paste your key). Other tools: paste the .mcp.json block above into their MCP settings with your API key
Just ask it things like "check if my site is cited by ChatGPT" - no separate dashboard to open

Needs an AI tool with MCP support - Claude, ChatGPT, Cursor and most others already have it, and more are adding it every month.

We ran it on ourselves first.

Our guide to checking ChatGPT citations wasn't cited by ChatGPT. This is what Vantage found, what our agent changed, and when we check again.

5 Nov

Check

0of 3 answers
cited the guide

Our free-check page was cited once, never the guide.

5 Nov

Fix brief

Put a number in the opening.
Cover what the cited answer covers.
Write it in your own words.

5 Nov

Agent edits the page

~~Ask ChatGPT your buyers' questions…~~+ As of November 2026, OpenAI gives no report…+ Method 4: check Google Analytics+ Method 5: check ChatGPT can read your site

12 Nov

Re-check

AI answers take days to change. The result goes here, cited or not.

The loop, run on our own site: 3 ChatGPT answers per check, United States.

Three jobs, one tool

Ask in plain words inside Claude or Cursor. Vantage answers there.

See where you stand

Which searches cite you, and if that is growing.

“Where do we show up in ChatGPT?”

See who wins, and why

The sites AI cites most, and how their answers are shaped.

“Who is winning AI search for this?”

Fix your page

A step-by-step list of what to change to get cited.

“What should I change on this page?”

See all ten tools

find_cited_questions

Starts from your domain, not a keyword: the questions where AI answers already cite your site, most asked first. No guessing which keywords to check.

“What does ChatGPT already cite us for?”

analyze_citation_coverage

Which of up to 10 keywords actually cite your domain in ChatGPT's, Gemini's or Perplexity's answers, across several answers if you ask, and what changed since your last check. Cited and named are reported separately.

“Out of everything we care about, where do we show up, and did our changes work?”

analyze_citation_gapDifferentiator

Diffs your own page against the winning answer for a keyword and returns a fix brief your agent can carry out: what to rewrite, which points to add, in order, then where to get mentioned off your site. As far as we found, the only tool on our comparison page that does this.

“What should I actually change on this page to get cited?”

find_citation_leaders

Who dominates AI-answer citations for a topic, whether a domain is among them, and how much of it is Reddit, YouTube and other community sites.

“Who's winning AI search for this?”

analyze_citation_trend

Month-by-month citation counts for a domain, so you can see whether visibility is growing or fading.

“Is our AI visibility improving?”

analyze_citation_structure

How the winning AI answer for one topic is actually shaped: list-led, sources cited, opening length.

“What does a winning answer look like?”

analyze_citation_structure_batch

The same shape-check across up to 10 related topics in a single call, so a content cluster gets analyzed together instead of one keyword at a time.

“What do winning answers look like across this whole topic cluster?”

get_check_history

Reads back your earlier coverage checks for a domain, so your agent can compare before and after a change without paying to ask again. Free.

“What did we find last month?”

get_supported_options

Which engines and markets each tool accepts, and the limits, so your agent picks valid arguments before it spends anything. Free.

“Can you check Gemini in Italy?”

get_usage

How many of this month's units are left, before your agent runs a bigger check. Free.

“How many checks do we have left?”

Plans, drawn to scale.

Checks a month, to scale. No seats, no annual lock-in, full refund within 14 days.

Plan	Checks	Per month	Choose
Free	30	$0	Start free
Pro	500	$29	Choose Pro
Scale	1,500	$49	Choose Scale

Start free Choose Pro Choose Scale

One question checked daily is about 30 checks a month, so Pro covers about 16 questions every day.

What a check costs.

Every plan is a monthly allowance of checks. Most things cost 1 check. Reading several answers for one keyword costs 1 check per answer. A leaderboard costs 10, because it ranks the sites AI answers cite most for a topic, not just yours. Your allowance starts fresh each month.

What each thing costs

Keyword checkanalyze_citation_coverage1 check per keyword
Answer structureanalyze_citation_structure1 check per keyword
Citation trendanalyze_citation_trend1 check
Page gapanalyze_citation_gap1 check
Leaderboardfind_citation_leaders10 checks
Cited questionsfind_cited_questions10 checks
Past checksget_check_historyfree
Engines and marketsget_supported_optionsfree
Checks leftget_usagefree

Using Vantage from an agent? It reports the same allowance in units:get_usage returns units_remaining. One unit is one check.