How we measure
AI visibility.

Noted checks whether AI answers mention your firm, recommend competitors, or cite your website. We save the evidence behind each answer and use it to decide what to improve. A report is a record of what happened in those checks—not a universal AI ranking.

We research 25 questions for your market

We start with your services, location, and target client. The list includes provider-selection, service, location, and comparison questions. You review the focus; we write the questions. A law firm’s client questions differ from the employer questions a legal recruiter wants to appear for.

We keep the exact wording stable for comparisons. Questions added later get a new version. A question naming your business tests whether AI understands you; it is reported separately from an unsolicited recommendation.

We compare five relevant competitors

We select businesses serving the same need and market, including names found in the initial answers. The report shows the questions where they appear and you do not. We inspect cited pages for useful differences, such as clearer service descriptions or relevant experience. Those differences suggest changes to test; they do not prove why an AI system picked a competitor.

We record the question, answer, sources, and date

The monthly scope is 25 questions across four providers: 100 scheduled checks. For each check, we record the exact question, answer, provider and model, time, search evidence, and whether it completed. We identify failures instead of treating them as zero visibility.

Our automated checks use search-enabled developer APIs from OpenAI, Anthropic, Google, and Perplexity. These are not recordings of the consumer ChatGPT, Claude, Gemini, or Perplexity applications. Models, tools, account context, and personalization can differ. Reports label the method rather than treating API results as identical to what every user sees.

Three different signals

• A mention: the answer names the correct business. Similar names and unrelated businesses need review.
• A recommendation: the answer presents the business as an option for the requested service. A negative mention or passing reference is not a recommendation.
• A citation: the answer links to your website. A link in search results is recorded separately from a citation supporting the final answer.

We can also record the first recommended business, but its position applies to that answer only. We do not subtract an AI mention count from a Google position to invent a ranking gap.

What does “30 out of 100” mean?

If your firm appears in 30 of 100 completed, search-backed answers, its observed mention rate is 30%. If only 80 checks complete, we report 30 out of 80, identify the missing 20, and compare only matching questions and providers over time. This is not a measure of market share, search volume, or lost clients.

How we choose the two monthly page improvements

We connect missing answers to a specific reader need on your existing site. A change might clarify who a service is for, explain a process, answer a recurring buyer question, or make supported experience easier to find. We write or revise two pages per month and publish after your designated reviewer approves factual and legal accuracy.

The monthly report records what changed, the evidence behind the decision, and three next priorities. We preserve earlier answers and page versions so the next report has a real comparison.

How this relates to Google rankings

Search Console tells us about Google indexing and search performance. A Google Maps observation depends on the query, map location, and time. Google Places discovery order is not a measured Google ranking. Our older public market snapshots and newer search-backed pilot runs identify their different methods; we do not combine them into one score.

What counts as progress?

Correct business identification, more relevant recommendations, and citations to useful pages are signals we can observe. We repeat the same checks and record changes in coverage. A single new mention—or a same-day repeat—is not proof that a page edit caused an improvement. Relevant visits and enquiries are separate outcomes.

We are testing this process on Noted itself. Internal testing helps us improve delivery; it is not independent client evidence or a guarantee of results.

First report within five business days

After completed onboarding, you receive the starting report within five business days. It shows the tested questions, competitor appearances, missing coverage, and proposed page priorities. Monthly monitoring then gives us a consistent way to revisit those decisions.

See exactly what the C$1,500/month service includes or discuss your firm’s coverage.