AI citation tracking
Your customers are asking AI about you.
Citation shows you the answers.
Citation asks the questions your customers ask, checks what ChatGPT, Claude, and Perplexity answer, and turns it into one AI visibility score with the cited pages sitting underneath it.
From $49 a month. You choose when to check. Cancel whenever. Accounts are set up by our team.
Overall AI visibility
▲ 6 this month
Across ChatGPT, Claude and Perplexity
Claude · “best hosting for nonprofit organizations”
Sample runFor nonprofit organizations, Acme Hosting is a strong fit: it offers discounted nonprofit plans, SOC 2 certified infrastructure, and migration support for organizations moving off legacy hosts.
Brand named Page cited Score 6/6
Most of the buying journey now happens inside an AI answer, before anyone reaches your site, and none of that shows up in your analytics. When a model lists three vendors in your category and you aren’t one of them, there’s no impression count, no lost-click report, and no ranking drop to investigate. The prospect simply never arrives, and nothing in your dashboard registers that anything happened.
It doesn't appear in GA4. Analytics records the visit that happens and stays silent about the answer that decided it.
You can't spot-check your way out of it. Ask a model the same question twice and you can get two different answers, so one manual check tells you very little about what a thousand prospects are seeing.
You can't fix what you can't attribute. Improving your position means knowing which of your pages get pulled, which competitors get named instead, and which model is responsible.
SCORING
One score, every time you check
Every answer gets scored on three things: whether the exact page you want was cited, whether any page on your domain was cited, and whether your brand was named in the text. Those roll up into a single 0 to 100 AI visibility score, and the trend chart splits it by check and by model, so you can tell a real shift from ordinary model variance.
CITATIONS
The pages AI actually pulls from
Top Cited Pages ranks the URLs the models reach for most, and it's rarely the page you'd have guessed. Most accounts find one or two pages carrying nearly all the AI visibility, which is useful in both directions: it tells you what to build more of, and it tells you which single page you can't afford to break.
PROMPTS
Where you win, and where you don't
Each prompt keeps its own score history, so you can see which questions reliably surface you and which ones hand the answer to somebody else. The bottom of that list is a content roadmap that came from real model behavior instead of a keyword tool.
EVIDENCE
Read the whole answer, not a summary
Open any run to see exactly what the model said, with your brand mentions highlighted in context and every citation listed. When a competitor gets recommended instead of you, the sentence that did it is right there, and every run stays in the record so you can compare this month against last.
How it works
Three steps, then it runs when you say so
Add your prompts and brand keywords. List the questions your customers ask AI, add your domains, and say which names count as a mention of you. Most accounts track 10 to 20 questions, enough to cover a category without spending an allowance on near-duplicates.
Check whenever it matters. Each prompt goes to ChatGPT, Claude, and Perplexity, and the full answers and citations get stored. You choose when to check: every few days to watch a trend, or the moment you publish something you want tested.
Read the score, then read what's under it. The score tells you whether you're moving. The cited pages, brand mentions, and per-model breakdown tell you what to do about it.
Act on it
A number on its own is a report card
Citation is built so that every part of the score points at a specific piece of work. Four patterns come up constantly, and each one has a different fix.
Named, but never cited. The models know who you are and aren't pulling your pages. That's usually a structure problem rather than an authority problem: content that isn't shaped like an answer, missing FAQ or schema markup, or weak entity signals tying your name to your category.
Everything riding on one page. One URL is carrying your whole category. Build siblings around it so a single change to that page, or to one model's index, doesn't take your AI visibility with it.
A competitor named every time. Look at where their citations come from. Third-party pages, directories, and roundups tend to move this number more than anything you publish on your own site.
Score dropped, nothing changed. A model updated, or a source you were relying on did. The per-model chart shows which one moved and when, which is the difference between a real problem and a week of guessing.
Want the fixes handled rather than listed? MOSAIC turns your Citation data into a prioritized AI-readiness plan and implements it.
Talk to MOSAICBuilt for
Marketing and brand teams
Track your own AI visibility, show the trend next to your other channels, and find out which pages are doing the work.
Agencies and consultants
Track several brands on Growth and above, each with its own prompt sets, competitors and scoring, all drawing on one pool of credits.
SEO and content leads
Report AI visibility alongside rankings, and test whether new content gets picked up instead of assuming it will.
Why not just ask ChatGPT yourself?
You should, once. It’s a useful thing to look at. The trouble is that one answer is one sample: responses vary between runs, they vary between models, and they shift as indexes and training updates roll through, so what you saw on Tuesday is not evidence about Thursday. Citation runs the same fixed prompt set against all three models every time you check, which turns anecdotes into a series you can trend, and keeps the raw answers so you still have last month’s version to compare against when something moves.
Simple pricing
One credit is one question, asked of every model. Spend them whenever you like: daily while you are making changes, weekly when you are not.
Starter
One brand, checked regularly
$49per month
- Credits a month
- 100
- Brands
- 1
- With 10 questions
- 10 checks
Growth
Daily tracking, or a few brands at once
$149per month
- Credits a month
- 300
- Brands
- 3
- With 10 questions
- 30 checks
Scale
Agencies and multi-brand teams
$369per month
- Credits a month
- 750
- Brands
- Unlimited
- With 10 questions
- 75 checks
Tracking more questions uses credits faster, so a 25-question brand on Growth gets 12 checks a month rather than 30. Need more in a given month? Credit packs are 10 for $9, 25 for $19, and 50 for $29.
Cancel any time. Your data stays available through the end of the period.
- ChatGPT, Claude and Perplexity on every prompt
- Full history, trend charts and per-model breakdowns
- Every cited page and brand mention behind every answer
- Competitor tracking and share of voice
- AI Readiness Audits for 5 credits
- CSV export, and as many people on the account as you need
Questions
Which AI models does Citation track?
ChatGPT, Claude, and Perplexity. Each prompt runs against all three on every run, and every score is broken out by model so you can see where you're strong and where you're absent.
How often does it check?
As often as your credits allow, whenever you choose to spend them. A check of a 10-question set costs 10 credits, so a 100-credit plan covers ten checks a month. Use them every few days to watch a trend, or several in a week while you are actively changing pages. Nothing runs on a schedule; a check happens when you ask for one.
What does the AI visibility score actually measure?
Each answer is checked for three things: an exact match on a page you care about, a match on any page on your domain, and a mention of your brand name in the response text. Those combine into a score per answer, and the average across a check is normalized to a 0 to 100 scale so it's comparable over time.
How is this different from SEO rank tracking?
Rank tracking tells you where a page sits on a results list. Citation tells you whether a model recommends you at all, which pages it pulls from when it does, and what it says about you in the process. The two are complementary, and Citation is not a replacement for your existing SEO tooling.
Can I track more than one brand or client?
Yes, on Growth and above. Each brand keeps its own domains, keywords, prompt sets, competitors and scoring, and they all draw on the same pool of credits, so three brands checked every few days costs the same as one brand checked daily.
What's a credit?
One credit is one question, asked of every model. A check of a 10-question set costs 10 credits, so a plan with 100 credits a month covers ten checks, or four if you track 25 questions instead. Credits refresh each month and do not carry over. If a month runs long, packs of 10, 25, and 50 are added to your account on request.
Can Citation improve my AI visibility, or only measure it?
Citation measures and diagnoses: it shows which pages get cited, which prompts you lose, and which model moved. The implementation work is a separate engagement, and MOSAIC can take it on if you'd rather not do it in-house.
How long before the trend means anything?
You get a score on your first check. Eight to ten checks is usually enough to separate a real movement from ordinary variance in model output. That is about a month on most plans, and sooner if you check more often while you are making changes.
Do you keep the full AI answers?
Yes. Every check stores the complete response text along with the citations, so you can go back and read exactly what a model said at the time.
Can I get the data out?
Yes, CSV export covers all runs and scores.
Do I need to install anything on my site?
No. Citation works from the outside by asking the models, so there's no script, plugin, or DNS change involved.
See what AI says about you today.
Tell us the brand you want tracked and we’ll set the account up for you. Your first score lands the same day.