AI Visibility: Measure How AI Sees Your Brand

Buyers ask ChatGPT, Gemini, Claude and Perplexity the questions you used to win on Google. The answer typically names about three brands. Prompt Intelligence tells you whether you are one of them, what the AI says about you, and what to fix when the answer is wrong.

The search results page is disappearing. Over half of Google searches already end without a click, and hundreds of millions of people ask ChatGPT questions every week. Buying journeys now start with a prompt, and the model answers with a shortlist. That creates two questions most brands cannot answer today.

When your buyers prompt, are you mentioned at all? We call this your Share of Model.

When you are mentioned, are you described correctly: your positioning, your audience, your claims, your pricing? We call this Positioning Accuracy.

Together they place every brand on our Visibility and Representation matrix: Invisible Brand, Hidden Expert, Misrepresented Brand or Category Leader. Most AI-visibility tools chase visibility alone. We measure both, because being recommended for the wrong reasons is not a win.

Measuring AI visibility is a discipline, not a screenshot habit. We track six metrics across the full engine panel: ChatGPT, Gemini, Google AI Mode, Google AI Overviews, Claude, Copilot, Meta AI, Perplexity and Grok.

The percentage of runs in which your brand is named.

How often your own pages are cited as a source in the answer.

Your share of mentions against a defined competitor set.

Where you appear in the answer, and whether you are recommended, merely mentioned, compared or caveated.

How positively the answer frames your brand, tracked over time.

Whether what the AI claims about you is actually true.

LLM answers vary between runs. A single answer proves nothing, so every metric is a rate across a stated number of runs in a stated window. A brand that appears in 7 of 10 runs has 70% presence, not "presence".

There is no reliable "prompt search volume", and we will not invent one. We prioritise prompts with transparent demand proxies: keyword volume of the underlying intent, People Also Ask frequency, and how often the question appears in communities and sales conversations.

Everything starts with your Prompt Universe: a versioned database of 150+ prompts, tagged by funnel stage, persona and topic cluster, built the way an engineer would build it.

We still do complete classic keyword research, but through agentic workflows with live connections to keyword and search volume tools. The full demand dataset for your category is pulled, clustered and intent-classified in hours, not billed by the day.

We extract real phrasing from sales calls, support tickets, reviews and CRM notes, so the prompt set reflects how your customers actually speak, not how a keyword tool guesses they speak.

We mine a large-scale PAA dataset to find the question patterns search engines have already validated for your category.

We scrape and index the threads where buying questions are asked in the wild. These platforms do double duty: they seed prompts and they are among the sources AI engines cite most.

Your existing SEO keyword set is translated into conversational prompts, carrying its volume along as a demand proxy.

LLM-assisted expansion fills the gaps per persona and funnel stage, and every synthetic prompt is tagged as synthetic. Observed and generated prompts are never silently mixed.

We probe what models believe your brand is, does and is known for, and compare it against ground truth.

Wrong pricing, discontinued products, misattributed claims: every error is logged with the prompt, engine and severity, and mapped to the upstream source that needs fixing.

The same panel runs against 3 to 10 competitors, producing a gap matrix per topic cluster: exactly which clusters they own, and where you can win.

Which domains, publications and communities the engines cite in your category, ranked by frequency. This is where authority is built, and it directs the off-page roadmap.

For every key prompt we diagnose whether the answer is grounded in live retrieval, fixable in weeks, or baked into the model's weights, a longer entity and authority play. That single diagnosis decides where effort goes.

Visibility metrics only matter if they connect to revenue. The AI Influence Funnel is our model for making that chain measurable, stage by stage.

Machine Access

Model Visibility

AI Search traffic

Declared Discovery

Qualitative Journey Intelligence

Attribution to business results pipeline · revenue · demand

Can AI crawlers reach and retrieve your content at all? Verified at log level, this is the prerequisite for everything below.

Are you present, recommended and correctly represented in the answers? This is where the six metrics live.

Verified visits arriving on your site from ChatGPT, Perplexity, Copilot and AI search surfaces, separated from spoofed bot noise.

What buyers tell you themselves: "found you via ChatGPT" in lead forms, sales calls and onboarding questions.

How AI actually shaped the buying journey, surfaced from interviews and sales conversations.

Pipeline, revenue and demand, connected back up the funnel.

The funnel is also where correlation becomes evidence: we track machine access and model visibility against the AI search traffic that follows, so you can see visibility gains turn into visits, and visits into pipeline. Every engagement instruments as many stages as your data allows, and tells you honestly which stages it cannot yet see.

Per engine and per cluster: mention rate, citation rate, position, sentiment and share of voice against competitors.

Every cause rolled up into four scores, benchmarked against competitors: Discoverability (can engines reach you), Clarity (do they understand you), Authority (do trusted sources cite you) and Trust (do they recommend you).

Every finding paired with a fix, one owner, a KPI and a 30/60/90 horizon. Discoverability lands with development, Clarity with content, Authority with PR, Trust with brand. AI visibility is not an SEO-team task, and the scorecard makes that visible to a CMO in one glance.

Trends, wins and losses, root causes and the ranked next actions. Interpretation, not a data dump.

Visibility drops, new competitors entering answers, and new factual errors, flagged as they happen.

We practise what we preach. Prompt research, answer parsing, QA and reporting run on our proprietary agentic delivery engine, improved weekly by our LLM Visibility Engineer. A full client panel takes days, not weeks, and a deterministic QA gate checks every dataset before a consultant ever interprets it. You pay senior consultants for judgment, not for copy-paste work.

AI Visibility is the Understand package: from EUR 2,000 per month, including continuous tracking, bi-monthly deep dives and guidance readouts, for up to three markets. It is the evidence layer that our GEO and ChatGPT Ads services build on, and it is where nearly every client starts.

Request your baseline audit

What is AI visibility?

AI visibility is how findable and how well represented a brand is inside AI engines such as ChatGPT, Gemini, Claude and Perplexity: which prompts buyers ask, whether the brand appears in the answers, how it is described, and which sources the engines rely on. We call the discipline of measuring it Prompt Intelligence.

What is Share of Model?

Share of Model is how often and how prominently a brand appears in LLM answers for the prompts that matter in its category. Think of it as market share inside AI answers, measured as a rate across repeated runs.

How is this different from an AI visibility tool?

Tools report scores; we deliver diagnosis and a plan. Our prompt sets are built from your customers' real language rather than keyword lists, every metric is reproducible, and every report ends in prioritised actions with owners. The tooling is our leverage, the service is the product.

Which AI engines do you cover?

ChatGPT, Gemini, Google AI Mode, Google AI Overviews, Claude, Copilot, Meta AI, Perplexity and Grok, each measured in both parametric and live-search modes where the engine supports them.

How long does a baseline take?

Days, not weeks. Our agentic delivery engine builds the Prompt Universe, runs the panel and parses the answers, so senior time goes into interpretation and the action plan.

Do you measure in Dutch as well as English?

Yes, natively. We build a separate prompt universe per language market rather than translating one, because competitors, citation sources and even category framing differ per language.

Know where you stand in AI answers

Get a baseline of your visibility, representation and share of voice across nine AI engines, with a scorecard and a 90-day action plan.

Request your baseline audit

Visibility

Representation

Mention rate

Citation rate

Share of voice

Position & role

Sentiment

Accuracy

Rates, not snapshots

No fake volume

Full keyword research, run AI-native

Customer language mining

A proprietary People Also Ask database

Reddit, forums and community indexing

Keyword-to-prompt translation

Synthetic expansion, honestly labelled

Entity and perception audit

Hallucination and error detection

Competitive benchmark

Citation source landscape

Retrieval or weights

01. Machine Access

02. Model Visibility

03. AI Search traffic

04. Declared Discovery

05. Qualitative Journey Intelligence

06. Attribution to business results

Baseline report

Four-Layer Scorecard

90-day action plan

Live dashboard and monthly insight report

Alerts