AI visibility monitoring for B2B brands
Know when your brand drops out of AI answers.
Your buyers ask ChatGPT, Perplexity, Gemini and Claude which companies to consider. CitePulse asks the same questions every day and tells you when your brand is present in the answers, when it drops out, and when a competitor appears in its place.
Baseline on day one. Daily runs from then on. From $49 a month, 7 to 10 day trial with a card, cancel any month.
Your AI visibility changes even when your website does not.
AI assistants change their answers, their sources and the companies they recommend without telling anyone. A brand named in every answer this month can be gone next month. A competitor can start showing up on the exact questions where you used to appear. Nothing on your side has to change for that to happen. Without a fixed question set measured on a schedule, you find out from a lost deal.
You don't know where you appear
Which buyer questions, in which assistants, name your brand today? Most teams have never asked the same question twice.
You don't know what changed
Did your visibility drop, did competitor presence rise, or did the assistant just wobble? A single check cannot tell those apart.
You don't know why
Which pages and sources appear behind the answers AI gives your buyers? Those sources can influence which brands appear in the answer.
A single audit shows one day. Monitoring shows the direction.
An audit is a photograph. It tells you where you stood on the day it ran, and the engines will have moved by the following week. CitePulse starts with a baseline, then repeats the same questions on the same engines every day and reads each one in a window of the last three measurements, so a real change stands out from the normal day-to-day movement of the answers. We measure that movement continuously and use it as the noise band behind your alerts; until an engine has 14 days of data, a default band applies. Either way a 3-point wobble is never sold to you as a win or a loss.
| One-time snapshot | CitePulse monitoring |
|---|---|
| Where you stood on one day | The same questions every day, with history and direction |
| Competitors as of that day | Competitor presence and share movement, run by run |
| Hard to prove a fix worked | Before/after per change, read against the noise band |
| You react after the damage | An alert when your visibility drops or competitor presence rises |
One view of how AI sees your market.
Every number is measured on real AI answers to your buyers' questions, shown per engine, with the detection rule published and every answer excerpt and cited URL kept as the receipt.
AI Visibility Score
0 to 100: how often your brand is named in the answer or cited as a source, per engine and composite, this period against the last.
Question coverage
Five baseline questions plus up to 30 of your own. Share of tracked questions where your brand appears, and the list where it does not.
Competitor presence
Which competitors appear on your questions, their share against yours with the full denominator shown, and who gained or lost visibility since the last run.
Platform breakdown
OpenAI, Perplexity and Gemini on every plan; Grok on Protect and Scale; Google AI Overviews and AI Mode read as sources. One platform can drop you while the others do not.
Sources behind the answers
The domains and pages that appear as sources for your questions: your own, a competitor's, or a third-party comparison. Marked by whether your own page can win that question.
Biggest moves, with the noise shown
The questions with the largest gain or loss since the last run, each read against the engines' own week-to-week movement.
See the change before it costs you a deal.
Each daily run is compared with the previous ones. When something moves beyond the engines' published noise band (a default band until 14 days of panel data exist), you get an alert in the dashboard and by e-mail. The alert types the product raises today:
Competitor intelligence
CitePulse never measures your brand alone. For every question it records which competitor domains were named or cited, so you see visibility share, question overlap, who gained and who lost since the last run, and which sources support their presence and not yours.
Source intelligence
Most of what an assistant says about your category comes from pages you do not own: comparison articles, review sites, community threads, vendor blogs. CitePulse lists the sources that keep appearing behind your questions and marks where a third-party page, not your own, is the one being cited. Across 95 audited companies, Reddit sat behind the answers for 91% of them and YouTube for 89%: the monthly ranking of the sources AI cites most.
From baseline to continuous monitoring.
1 · Define your monitoring baseline
Add your domain, competitors and the buyer questions you want CitePulse to track. Five questions are generated for your category; add up to 30 of your own.
2 · Get your baseline
The day you subscribe, a full monitoring run establishes the starting point. Nothing to install.
3 · Monitor daily
The same questions, the same platforms, every day, each read in a window of the last three measurements. Frozen sets keep every comparison like-for-like.
4 · See what moved
Changes by question, competitor, platform and source, each read against the engines' noise band.
5 · Act and measure again
Log what you changed in the Action Center; the next runs show the before/after per question and platform.
Built on measured AI answers, not assumptions.
CitePulse records actual answers from the assistants to a controlled, frozen question set, every day, with live web search where the engine offers it, and scores them with a published rule: brand named in the text, or domain in the sources. Each question is read in a window of the last three measurements. We do not estimate AI visibility from search rankings, keyword volumes or third-party data; that is also why a measured answer costs more to produce than an estimate, and why an alert from it is worth acting on.
What the measurements have shown so far: only 1.5% of the sources behind AI answers point at the company itself (93 audits), and the sources behind the same question turn over by about two thirds day to day. Wrong prices and facts in the answers are caught separately by Fact Integrity; agencies run all of this per client with white-label reports on the Scale plan.
Public methodology
The detection rule, the disclosed share-of-voice denominator, the exact models and access methods, and the honest limits. Read it
See it on a sample company
The client panel with alerts, before/after and sources, on example data. Open the demo
Choose the level of visibility you need to monitor.
Every plan measures daily, starts with a 7 to 10 day trial (card at the start, nothing charged if you cancel in time) and includes the baseline run, alerts, question suggestions, the Action Center and cancellation any month. Annual billing: 23% off, price locked for 12 months. Prices in USD; your card issuer converts. Full limits and the comparison with other tools on the pricing page.
Annual prices are per month, billed once a year and locked for 12 months.
Starter
$49 / month
One brand, first daily monitoring. 7 day trial.
- ChatGPT, Perplexity, Gemini, DeepSeek; Google AI Overviews and AI Mode as sources
- 5 baseline questions + up to 15 of your own
- Daily measurement, 3-measurement window
- About 2,400 measured answers a month
- Alerts, competitor and source tracking, 3 experiments
Shield
$89 / month
One brand that wants Claude and more questions. 7 day trial.
- Adds Claude, weekly with live web search; DeepSeek twice a day
- Up to 30 of your own questions
- About 5,400 measured answers a month
- Fact Integrity (10 facts), 5 active experiments
- AI bot-log totals per crawler
Protect · most chosen
$169 / month
A brand actively working on its AI visibility. 10 day trial.
- Adds Grok and Claude, twice a week each, Mistral monthly; DeepSeek three times a day
- Up to 50 of your own questions
- About 10,900 measured answers a month
- Priority alerts, Fact Integrity (20 facts), 15 active experiments
- Full AI bot logs: which pages each crawler reads
Scale
$399 / month
Agencies and multi-brand teams. 10 day trial.
- Protect engines for every client domain, measured every 2 days
- Up to 10 client domains, $30 each above that
- White-label reports, agency dashboard, API, badges
- Up to 25 own questions and 30 experiments per client
- Partner programme with commission
Not ready to subscribe? Check your current baseline free, no card. Prefer a walkthrough first? Book a demo. Enterprise (many domains, SSO, SLA): contact us.
Questions before you start
How do I audit how often my brand is cited by AI tools like ChatGPT?
Run the free audit. It asks ChatGPT, Perplexity and Gemini five real buying questions from your category, each one three times, and reads Google AI Overviews as well: 45 real AI queries, results in about 30 seconds, no card. You see on which questions your brand was named or cited and which competitors were named instead. Monitoring then repeats the same questions every day.
What is the best tool to check if my company is showing up in AI-generated answers?
Any tool that asks the engines real buying questions and lets you read the raw answers is a fair start; a score alone hides too much. CitePulse does that free: 45 real AI queries across ChatGPT, Perplexity, Gemini and Google AI Overviews, 10 technical checks of your site, an AI Reputation Score from 0 to 100, and the list of competitors the engines name instead of you.
How can I see if AI assistants are mentioning my competitors more than me?
The audit shows which competitors the AI names instead of you on each question. Monitoring records competitor domains on every question, every day, with the share-of-voice denominator, so you see who gains presence when you lose it. Connect Google Search Console (read-only) and you also see where Google ranks pages that AI does not cite.
Is there a way to audit my visibility in large language model responses?
Yes. The free audit sends 15 queries per engine to ChatGPT, Perplexity and Gemini and reads Google AI Overviews. The one-time Full Audit adds Claude and Mistral; monitoring adds DeepSeek on every plan, Claude from Shield, and Grok plus Mistral on Protect and Scale. Every run keeps the answers, so you read what each model said instead of trusting a number.
How do I find out what AI chatbots are saying about my business?
The audit and the dashboard show verbatim quotes: fragments of the AI answers about your brand, per engine, next to the brands named and the sources cited. Fact Integrity then compares those answers with the facts you approved, so a wrong price or a mixed-up company shows up as a flagged line, not as a surprise from a customer.
Which AI visibility monitoring tool fits a small B2B company?
For one domain, Starter at $49 a month runs daily monitoring across ChatGPT, Perplexity, Gemini and DeepSeek with up to 15 of your own buying questions; Shield at $89 adds Claude, 30 questions and 10 approved facts. Protect at $169 adds Grok and Claude twice a week and up to 50 questions. Every plan asks each question once a day per engine, reads the last three measurements, starts with a 7 to 10 day trial and with the free audit as the baseline, so you check the numbers before paying.
Is CitePulse an SEO tool?
No. It monitors how AI assistants answer your buyers' questions: whether your brand is named or cited, how competitor presence moves, and which sources appear behind the answers. Search rankings are a separate signal; you can connect Google Search Console to see the two side by side.
Is this a one-time AI visibility audit?
No. Monitoring starts with a baseline run on the day you subscribe, then repeats the same questions every day and shows what changed against the engines' own noise.
Which AI platforms do you monitor?
OpenAI (the ChatGPT model family through the API with web search), Perplexity, Gemini and DeepSeek on every plan; Claude from Shield (weekly, twice a week on Protect); Grok and Mistral on Protect and Scale. Google AI Overviews and Google AI Mode are read as sources. The exact models and access methods, including how API models relate to the consumer apps, are on the methodology page.
Does "present" mean recommended?
No. Present means your brand was named in the answer text or your domain appeared in the answer’s sources, by a published string rule. Tone is not scored; a negative mention counts as presence. Wrong prices or facts in the answer are caught separately by Fact Integrity, which compares answers with the facts you approved.
Can I monitor competitors?
Yes. Competitor domains are recorded on every question, on every run, with the share-of-voice denominator shown.
Can CitePulse tell me why visibility changed?
It shows what changed: which questions, which platform, which competitor gained presence, which source appeared or disappeared, and whether the move is bigger than the engines' normal wobble. It does not claim causes it cannot see.
Does CitePulse guarantee higher AI visibility?
No. Nobody can. We measure, we show the levers, and every fix you mark is read before and after against the engines' normal wobble, so you see what moved and what did not.
How is this billed?
Monthly by card through Stripe in USD, cancel any month from the billing portal. Every payment comes with a VAT invoice. Details on the pricing page.
If AI is influencing your buyers, you should know what it is telling them.
Monitor your brand, competitor presence and the sources behind the answers, on the AI platforms shaping purchase decisions.
Baseline included. Daily monitoring from day one, 7 to 10 day trial.