Mention
The company is named in the answer.
/ AI Search Visibility Measurement
See how AI answers describe, cite, and recommend your company.
Create a dated, repeatable benchmark for the buyer questions, AI platforms, competitors, recommendations, citations, and factual accuracy that matter to your market.
A scoped service, not an instant website scan. Scope and price are agreed first.
/ Overview
Decide from evidence, not screenshots.
Buyers ask AI systems to describe, compare, and shortlist companies. Without a structured view of those answers, it is easy to miss where your company is absent, where it is misrepresented, and which comparisons deserve investigation.
We run an agreed set of buying questions across the agreed platforms, record the answers, and classify what appears. Repeat runs show whether observations are stable or changing. They support comparison; they do not, by themselves, establish what caused a change.
Establish the baseline before judging your next round of changes.
/ Boundaries
What this measurement can and cannot tell you.
What it can do
Show observed recommendation, placement, citation, source, competitor, and accuracy patterns within the agreed sample.
What it cannot do
Represent every possible prompt, user context, geography, model version, or future answer.
/ Signals
Three signals, recorded separately.
A single answer can contain all three, so they overlap rather than forming exclusive states. None of them is, on its own, a buyer outcome. We define the denominator for each metric rather than treating mention rate and share of voice as interchangeable.
One recorded answer
Citation
The answer includes a reference to a company-owned page or a relevant independent source about the company. We record the exact URL and distinguish owned from independent sources.
Recommendation
The answer presents the company as a suitable option, judged against a rule agreed with you before collection.
/ Deliverables
What you receive.
If sentiment is included, the report explains its scoring rules and scale.
Agreed before collection
Agreed framework
The buying questions, platforms, markets, and competitors in scope, agreed with you before collection starts.
Recorded output
Dated scorecard
Counts, denominators, prompts, platforms, collection dates and conditions, with repeat comparisons where scoped. Model versions are recorded when available.
Competitor and source map
Which companies appear alongside you on the same questions, and which owned and independent sources sit behind the answers.
What you decide from
Review priorities
Observations separated from hypotheses, with the questions worth investigating next. Implementation work is scoped separately.
Illustrative scorecard view. Not client data, and not a fixed deliverable for every scope.
/ Reporting views
See the pattern from more than one angle.
A useful report should help your team see where a visibility issue is concentrated, not just produce one aggregate number.
- View 01
Question and prompt clusters
Group related buying questions so you can see whether a gap is concentrated around comparisons, fit, alternatives, pricing, or another decision area. The report states how each cluster was defined.
- View 02
Model by topic view
Compare the agreed platforms and question groups to see whether an observation appears on one platform or more widely. This is a reporting view, not a universal model ranking score.
- View 03
Competitor and source view
See which companies appear on the same questions and the owned or independent URLs observed alongside the answers.
Illustrative example of a reporting view. Not client data, and not a fixed deliverable for every scope.
/ Example
What a scorecard looks like.
Illustrative example using 20 fictional answers. No client data was collected and no AI platform was tested. The three signals can overlap.
| Signal | Count | Share of answers |
|---|---|---|
| Company named | 8 of 20 | 40% |
| Positively recommended | 5 of 20 | 25% |
| Owned page cited | 3 of 20 | 15% |
Scroll the table sideways to see every column.
/ Process
How a measurement round runs.
Platforms may include Google AI features, ChatGPT, Gemini, Perplexity, Copilot, and Claude. The platform set and collection approach are agreed before work starts.
- 01
Agree scope
The buying questions, markets, competitors, and platforms that matter, agreed with your team first.
- 02
Collect and check
Answers recorded and classified against the agreed rules, with manual review where appropriate.
- 03
Explain and repeat
Exceptions and model changes noted, and repeat runs compared like for like against the same framework.
/ Why now
Create the baseline while the answer is still changing.
AI models and answer systems are still changing what they retrieve and recommend. That creates an opening: you can improve the evidence they find before incomplete or inconsistent answers become familiar to buyers. Start with a benchmark now, then measure what changes.
/ Pricing & scope
Configure this scope around your priorities.
Adjust the quantities to create a planning estimate. You can use a service on its own or combine it with the full five-step system. We review every configuration before work begins.
Monthly · $100 per cluster per month · minimum 10
1 cluster = 10 related buyer-intent prompts. The controlled starting benchmark is 10 clusters / 100 prompts across 6-7 agreed LLMs.
Controlled baseline.
$1,000
10 buyer-intent clusters
- Controlled baseline:
- 10 buyer-intent clusters
- Derived prompts:
- 100 prompts
- Approximate prompt-platform runs per wave:
- 600–700
Estimate summary
- 10 buyer-intent clusters
- $1,000
- Estimated monthly total
- $1,000
- Estimated one-time total
- $0
- Estimated approved pass-through
- $0
- Estimated first month
- $1,000
Estimated first month $1,000. Estimated monthly total $1,000.
Planning estimate only. Final scope, cadence, dependencies, third-party costs, and commercial terms are confirmed in a written statement of work. AI rankings, citations, traffic, revenue, and editorial outcomes are not guaranteed.
Measurement describes a defined sample of prompts, platforms, and locations on the dates observed, and results vary between runs. Client factual approval is required before anything is published, and publishers and platforms keep editorial and ranking control.
We review the configuration for dependencies, sequencing, feasibility, and third-party costs before proposing a final scope.
/ System fit
Use this service on its own or connect it to the five-step system.
A focused scope works when the question is already clear. The connected system works best when measurement points to website, evidence, technical, and publishing gaps that need to move together.
/ FAQ
Common questions.
Is one run enough?
One run is a snapshot of that date and prompt set. Repeat runs show how much answers vary and whether an observation holds, but no number of runs is a complete census of every possible AI conversation.
Is the sample representative?
It is a defined set of agreed questions, not a statistically representative sample of all buyer conversations. Where a broader claim would need justified sampling, we say so rather than imply it.
Will monitoring fix our pages?
No. Measurement shows where you stand and what changed; improving the pages, content, and sources is implementation work, scoped separately under our GEO services.
Can you guarantee mentions or citations?
No. AI systems are outside our control and their answers vary by model, prompt, session, location, and date. We commit to an agreed framework, careful collection, and honest reporting of what we observe.
/ Next step
Stop judging visibility from a handful of screenshots.
Agree a measurement framework your team can use to make the next decision. Every request is reviewed manually by our team.