Grow your share of voice on every AI engine, and see what earns it.
Every month it tracks how much of the conversation is yours across the five engines your buyers use, which competitors appear beside you, and the sources the engines cited to put them there. One number you can take to a board, every answer kept as proof, and a briefing saying what changed and what to do next.
Live in three steps. One hour of your time.
You get the right questions
Not a keyword list. We draft the questions your buyers, investors and journalists actually put to an engine, including the comparison questions where you are measured against named competitors and the awkward ones somebody asks when you are in the news. You correct the set in an hour, and the setup is done.
We ask them, everywhere, on a schedule
Every question goes to Google AI Overviews, ChatGPT, Gemini, Copilot and Perplexity, and every answer comes back whole: the wording, who is named, who is cited, the sources behind it. There is nothing to install and no tag to add, because we read the engines rather than your website.
You can finally answer the question
How are we doing in AI? One number, broken down by engine and by question, with every figure a click from the answer that produced it. You hear about movement the week it happens, and when somebody asks what the machines were saying in March, you have it in front of you.
One number your board understands. Every answer behind it.
Most teams have a feeling about AI and no evidence for it. The dashboard replaces the feeling with a number you can put on a slide, and the number is never a black box: every point of it traces back to an answer you can sit and read.
That is what makes it usable in a meeting. You can say where you stand, show exactly why, and say what would move it, without anybody having to take your word for any of it.
From the dashboard // one sample month
On the dashboard // six views
The number you report upwards
One score for the whole question set, built as the exact average of the per-engine scores beneath it, so it can be checked rather than believed. It answers the question a board actually asks, which is not whether AI matters but whether you are winning or losing in it, and whether that is better than last month.
Where you are missing
Every tracked question in one list, with who was named and who was cited on each. The quickest route to the questions you are absent from entirely, which are usually the ones worth the most.
What was actually said
The complete text of every capture, not a summary of it. You read what your buyer read, in the engine's own words, including the parts about you that are out of date or simply wrong.
Proof, months later
Every capture kept and dated, with the sources the engine showed beside it. When a board member, a journalist or a counterparty asks what the machines were saying at the time, you produce the record instead of a recollection. AI answers keep no history of their own; this is the history.
Bad news, early
Citations gained, citations lost, answers quietly reworded after a model update. On the Dashboard tier you hear about a change in the week it happens, which is the difference between correcting it and explaining it.
The reading, done for you
A short written read in plain English: what moved, what it means and what is worth doing about it. It lands on both tiers, so the thinking still arrives in a month when nobody has time to log in.
What each answer tells you.
Every capture is read for eight things. Each one settles a question you would otherwise be guessing at, and each one points at something you can do.
-
01
Whether you are even in the room
Being mentioned at all is the first threshold, and most firms find they are below it. You see whether you appear in each answer, whereabouts you appear, because the order of mention is itself a ranking, and which competitors are standing where you are not. Variants of your name are matched too, so a mention that drops your Ltd or your hyphen still counts for you.
- named or absent
- position in answer
- competitors named
- name variants
-
02
Whether the mention sends anyone to you
Naming is credit. Citation is traffic. An answer that names you without linking to you sends nobody anywhere, so you get the sources each engine showed, which of them are your own pages, and which are somebody else's page talking about you. That split tells you whether to publish or to get published.
- source links
- your pages
- third-party pages
- citation share
-
03
How you stand against the firms you lose to
Your share of voice is scored against the same figure for the competitors you named at the start, on the same questions, in the same week. It turns a general anxiety into a specific one: not that AI is a threat, but that on the four questions that decide your shortlist, two named rivals appear and you do not.
- overall score
- by engine
- by question
- competitor comparison
-
04
Whether what it says about you is true
An answer can name you and still be wrong about you. Every capture is read for the claims made: what you do, whom you work for, where you operate, and anything stated as fact that is years out of date or was never true. An error in an AI answer is not a one-off; it is repeated to everyone who asks until something displaces it.
- full answer text
- claims about you
- out-of-date facts
- stated errors
-
05
What changed since last time
AI answers move without notice and nothing on your side tells you. Each capture is compared against the one before it, so a citation you quietly lost becomes something you find out about in days rather than something you never learn. This is the item most clients say earns the subscription on its own.
- new citations
- lost citations
- reworded answers
- model updates
-
06
Which pages the engines are actually reading
Every answer is assembled out of pages, and the pages are the part you can influence. You get the sources the engines lean on across the whole question set, which shows you two things at once: the publications that carry weight in your sector, and the pages of your own that are already doing the work and deserve more of it.
- cited domains
- your citable pages
- sector publications
- source overlap
-
07
Where the risk is, before it is a story
The question set includes the ones nobody enjoys writing down: the disputes, the incidents, the things a journalist would ask. Watching how the engines answer those is an early warning you cannot get any other way, because it shows you what a stranger is told about you on the day they go looking.
- risk questions
- tone of answer
- repeated claims
- flagged captures
-
08
Whether any of it can be proved
A number nobody can check is an opinion. Every capture is stored dated and in full, with its sources, so any line in the briefing can be traced back to the answer that produced it and put in front of a board, a journalist or a counterparty without anybody having to trust the reporter.
- dated captures
- sources shown
- full text stored
- exportable record
None of this touches your website. We measure the engines rather than your pages, so there is nothing to install, nothing to maintain, and the tracking carries on unbroken through a redesign or a move to a new platform.
Prove what the machines said. Months later.
AI answers are written fresh every time somebody asks. Wording shifts, sources come and go, and a model update can drop you out of an answer you were in last week. Nothing on your side changed, and nothing tells you it happened.
That is the difference from search. A ranking can be looked up months later; an answer is generated, read and gone. Keep both captures and a change becomes evidence you can show a board, a regulator or a journalist. Keep neither and it is one person's recollection against another's, which is not a position you want to be in on the day it matters.
Two ways to run it.
Both track the same questions across the same five engines and both include the monthly briefing. The only question is how close to it you need to be.
Answer Report
The same tracking, delivered as a monthly briefing. Nothing to log into, and nobody has to remember to go and look.
- The full question set, put to all five engines
- Share of voice, overall and by engine
- What changed since last month, in plain English
- The answers behind every claim, quoted in the report
- Delivered by email, monthly
For teams where AI answers are a standing agenda item.
Request the Answer ReportAI Visibility Dashboard
Everything in the Answer Report, plus the tracking live as it happens and the whole archive sitting behind it.
- Live share of voice, by engine and by question
- Every captured answer, dated and searchable
- Movement flagged when it happens, not held for the briefing
- The evidence archive, exportable for board and press use
- Competitor visibility on the same question set
For teams where AI answers touch deals, reputation or a risk that is already live.
Request the DashboardThe answers are being given right now.
Today, buyers, journalists and investors will ask an engine about your sector and be given an answer. It names some firms and not others, and it is given whether or not anybody on your side is watching.
Absence is not a neutral position. AI answers lean on the pages that already rank and are already cited, so the firms named today are the ones most likely to be named tomorrow, and measuring is what puts you among them.
Unlike a search result, an AI answer leaves no trace: generated, read, gone. The month you start looking is the month your record begins, and everything before it is unrecoverable. That is the whole argument for starting now rather than at the next review.
You cannot correct an answer you never saw. Monitoring is what turns AI from a rumour into something a team can act on.
01What is an AI visibility dashboard?
An AI visibility dashboard tracks how AI engines answer the questions that matter to your organisation: who gets named, who gets cited and whose pages the answers are built on. Ours follows an agreed set of buyer questions across Google AI Overviews, ChatGPT, Google Gemini, Microsoft Copilot and Perplexity, scores your share of voice, keeps every captured answer in a dated evidence archive, and reports movement in a monthly plain-English briefing.
02Which AI engines does it cover?
Google AI Overviews, ChatGPT, Google Gemini, Microsoft Copilot and Perplexity. Claude does not currently allow reliable capture of its answers, so we treat it as a stated gap rather than pretend coverage. When that changes, it joins the set.
03What is share of voice in AI answers?
The proportion of tracked answers in which you are named or cited, measured across the whole question set and broken down by engine and by question. It turns a vague worry about AI into one number that can be watched, compared and moved. The dashboard shows yours overall, per engine, and over time.
04How is this different from the free AI Visibility Audit?
The free audit is a snapshot: an instant, automated read of how AI engines see you today. The dashboard is continuous: the same questions asked on a schedule, every answer captured and compared, so you see movement, not a moment. Most clients run the free audit first to see where they stand, then move to monitoring.
05What is the Answer Report, and which tier do we need?
The Answer Report is the entry tier: the same tracking, delivered as a monthly emailed briefing with no login and no dashboard. The AI Visibility Dashboard is the upgrade: live access, the full evidence archive and movement alerts between briefings. If AI answers are a quarterly agenda item, the Report is enough. If they touch deals, reputation or a live risk, take the Dashboard.
06What do you need from us?
An hour at the start. We draft the question set from what your buyers actually ask, you correct and approve it, and you tell us which competitors matter. No website access, no code and no tags are needed, because the dashboard reads the engines, not your site.
07How often are answers captured?
On a schedule agreed with you when the question set is fixed, and the briefing lands monthly on both tiers. When something moves between captures on the Dashboard tier, it is flagged rather than held for the briefing.
08Can it prove what an engine said after the fact?
Yes. Every captured answer is stored dated and in full, with the sources the engine showed. When a board member, a journalist or a counterparty asks what the machines were saying, you produce the record rather than a recollection. AI answers change and leave no history; the archive is the history.
09Will it improve our visibility, or only measure it?
The dashboard is the measurement. Moving the numbers is the work of the GEO and AEO Programme, and the dashboard is what proves that work is paying back: you watch share of voice respond to what gets published and fixed. Measurement without the Programme still earns its keep, because it catches drift and misstatements early.
10Do we need the SEO and Website Audit first?
Not for monitoring itself. But AI engines build answers from pages they can crawl and trust, so if the dashboard shows you absent, the audit is usually where the reason is found. The two work as diagnosis and instrument: the audit finds what to fix, the dashboard shows the answers responding.
Start growing your share of the answer.
Tell us who you are and where to reach you. We draft the question set within days, and the first captures land as soon as you have approved them.
