AI visibility monitoring, run as a standing service.
AI visibility monitoring tracks whether, how often and how accurately AI engines cite you when the questions that matter are asked. A fixed set of those questions runs through ChatGPT, Gemini, Perplexity, Microsoft Copilot and Google's AI Overviews on a standing cadence, and every cycle is read against the same baseline. We run it as a managed monthly service, for organisations directly and white-label for the firms that serve them.
If you have an in-house search team with room to own the cadence, a self-serve tracking tool may be all you need; our guides below compare them honestly. This page is for the other case: the measurement has to run without anyone in-house carrying it, hold up in front of a board or a client, and hand its findings into the work that actually moves the answers. The one-off version is the free AI visibility audit; this is what it looks like as a standing capability.
Five readings, one baseline.
There is no public ranking to look up, so every reading is an observation on a fixed set of questions, compared with the cycle before it. The full method is in our measurement guide; the numbers worth reporting, and the vanity metrics to refuse, are in the metrics guide.
- Citation rate On a fixed set of the questions your buyers and stakeholders actually ask, how often each engine cites you in the answer. The core selection metric: when the question is asked, are you part of the evidence or not.
- Share of citations Your citations as a proportion of all citations on those prompts, against the competitors that matter. The nearest thing AI search has to a rank, and it moves for reasons you can inspect.
- Accuracy What the engines actually say when you do appear: correct and current, or stale, conflated and confidently wrong. Presence without accuracy is a liability, so it is logged, not assumed.
- Sources cited Which pages each answer was built from, yours or someone else's. This is where the readings become a work list, because visibility moves at the source layer the engines read.
- The trend Every reading against the baseline, month over month, on an unchanged prompt set. The trend is the product: it is what turns four observations into evidence that the work is landing.
Frozen questions, standing cadence, per-engine readout.
- The prompt set The questions that matter to the brief, written down and frozen: buyer questions, stakeholder questions, the queries where being absent costs you. The set stays fixed so every later reading is comparable with the first.
- The cadence The set runs through each engine on a standing schedule, frequently enough that run-to-run noise averages out. Single snapshots flatter or panic; aggregated runs tell the truth.
- The readout A per-engine report a board or a client can read in a minute: citation rate, share of citations and accuracy for each engine, beside the baseline. Never one blended score, because the engines are not one channel.
- The response Monitoring observes; it does not change what the engines say. When a reading moves the wrong way, the findings hand straight into source-layer correction work, ours or your team's, and the next cycle shows whether the fix landed.
The measurement layer, sold on its own.
Monitoring runs as a flat monthly engagement, sized to scope: the engines tracked, the size of the prompt set, the cadence, and the depth of reporting. It stands alone for organisations that want the scoreboard before committing to correction work, and it runs as the measurement layer inside a wider AI search visibility programme. For communications and government-relations firms it runs white-label across client brands, with the reporting issued in the firm's name.
We quote on a call, once we understand whose answers matter to you. The honest starting point costs nothing: the free AI visibility audit runs the baseline read, a one-per-organisation snapshot of how the engines describe you today, and who they name instead.
- Running cost A flat monthly fee, sized to scope and quoted on a call. No per-seat licences, no usage tiers.
- What arrives each month The per-engine readout against baseline, the sources driving each answer, and a prioritised note on what the readings say should happen next.
- What it deliberately is not A software licence, or a substitute for the source-layer work that moves the answers. Monitoring observes; the correction work is scoped separately, by us or by your team.
The complete guide set.
- How to measure AI search visibility: metrics, tools and method The measurement pillar: observation, baselines and the four things to track.
- AI visibility metrics that matter (and the vanity metrics to ignore) The five metrics worth reporting, and the numbers that only look like evidence.
- The best AI visibility tracking tools, and when you need a managed service The tool landscape, compared: where the named names live.
- Reporting AI search visibility to your board or members The one-page format a non-specialist can read, and the honesty rules behind it.
- How to rank in ChatGPT, Gemini, Perplexity and AI Overviews Per-engine retrieval and citation mechanics, and where to focus by audience.
- How to get your company mentioned in ChatGPT The levers that move whether a brand appears, and how to track them.
- How ChatGPT decides what to say about brands The retrieval and citation mechanics behind AI answers, in plain English.
- Who fixes wrong AI answers about your company? The provider landscape when an answer is wrong: ORM, specialists, counsel, DIY.
- What to do when ChatGPT gets your company wrong The calm, practical correction sequence, from documentation to re-measurement.
- How to fix AI giving wrong information about your brand Why engines get brands wrong, and the source-layer fix that holds.
You cannot manage what nobody is measuring. The baseline read is free — the trend is the service.
Get your free auditAI visibility monitoring, answered.
Straight answers on what monitoring measures, what it costs, and what it can and cannot do on its own.
What is AI visibility monitoring?
AI visibility monitoring is the standing measurement of whether, how often and how accurately AI engines cite an organisation when the questions that matter are asked. A fixed set of priority questions is run through engines such as ChatGPT, Gemini, Perplexity, Microsoft Copilot and Google's AI Overviews on a regular cadence, and each cycle records citation rate, share of citations against competitors, accuracy of what is said, and the sources each answer was built from, all compared against a baseline. There is no public ranking to look up, so monitoring by observation is the only honest way to know how you are represented.
Which AI visibility metrics matter, and which are vanity metrics?
Five metrics carry decisions: citation rate on a fixed prompt set, share of citations against competitors on those same prompts, accuracy of what the engines say, referral traffic from AI surfaces, and the month-over-month trend on an unchanged prompt set. The common vanity numbers fail at least one honesty test: raw mention counts move when the prompt mix changes rather than when visibility does, one-off screenshots capture a single run of a single engine, and blended visibility scores with undisclosed methodology cannot tell you what to fix.
How often should AI visibility be measured?
Frequently enough that run-to-run variation averages out, reported monthly against the baseline. The same prompt on the same engine can answer differently an hour later, so any single reading is noise; the discipline is frequent runs aggregated into a stable monthly figure, with closer attention around model updates, launches and contested moments. A quarterly screenshot is not monitoring, it is a souvenir.
Do you offer AI visibility monitoring as a standalone monthly service?
Yes. The measurement layer is sold on its own: the prompt set built with you, the standing per-engine cadence, and monthly reporting a board or a client can read, as a flat monthly engagement. It runs standalone for organisations that want the scoreboard before committing to correction work, and as the measurement layer inside a wider AI search visibility retainer. For communications and government-relations firms it runs white-label across client brands, with reporting issued in the firm's name.
What is the difference between an AI visibility tool and a managed monitoring service?
A tool is a scoreboard you operate: you choose the prompts, run the cadence, interpret the data and do the fixes it points to. A managed service does that operating for you: the prompt set is built from the questions your buyers actually ask, the cadence runs without anyone in-house owning it, the reporting is written for a board or a client rather than a dashboard, and the findings hand into correction work rather than sitting in a chart. Our tools guide compares the named options and sets out the honest test for which model fits.
How much does AI visibility monitoring cost?
It is a flat monthly fee, sized to scope rather than sold as a package: the number of engines tracked, the size of the prompt set, the cadence, and the depth of reporting all move it. Most engagements begin with a baseline assessment that establishes where you stand before any standing cadence starts, and the free audit is the no-obligation version of that first read. We quote on a call, after we understand whose answers matter to you.
Can monitoring alone improve my AI visibility?
No, and a provider who implies otherwise is selling a dashboard. Monitoring observes what the engines say; it does not change it. Visibility moves at the source layer the engines read: clearer entity signals, structured and quotable content, corroborating third-party coverage, corrections at the sources feeding an error. What monitoring does is make that work honest, by showing where you stand, which sources drive each answer, and whether the fixes are landing. The two together are the programme; either alone is half of one.