High Clarity

AI Visibility Index Methodology

How we measure, score, and report AI citation visibility for businesses in a metro and vertical combination.

How the measurement works

The AI Visibility Index measures how often businesses in a specific metro and vertical are cited by AI models when users ask common discovery and recommendation questions.

For each combination, we build a fixed question set, run every question multiple times per model, record the responses, extract cited business names, and compute standardized visibility metrics. The result is a ranked table that shows which businesses appear most often in AI-generated answers.

All measurements are conducted against live model endpoints using neutral, realistic prompts. We do not prompt models to favor any business, geography, or outcome.

The 25-question template

Each combination uses a standardized template of 25 questions designed to mirror how real users ask for recommendations in that vertical. Questions cover discovery, comparison, urgency, and local intent.

Placeholder — question content pending

The finalized list of 25 questions for this vertical will be inserted here. Each question will be shown in plain text so the methodology is transparent to readers and crawlers.

Models and surfaces queried

We query the major consumer AI models that users rely on for recommendations and local discovery. The exact set may change as the market evolves; each combination page lists the models included in that measurement.

Current models include ChatGPT, Claude, Gemini, and Perplexity. We query each model through its standard chat interface or API using identical prompts and no additional context.

Runs per question and why repeat runs are necessary

AI model responses are non-deterministic. The same question can produce different answers on different runs because of sampling, context windows, and model updates.

To reduce noise and capture a stable signal, we run each question multiple times per model. Citation counts are aggregated across all runs before scores are computed. This repeat-run design makes the index more reliable than a single-query snapshot.

Scoring formulas

Citation rate

The percentage of all queries in the measurement for which a business was cited at least once.

citation_rate = citations_received / total_queries

Weighted score

A business receives partial credit when it is cited alongside competitors in the same response. A solo citation earns more weight than a shared citation.

weighted_score = sum(1 / businesses_cited_in_response) across all citations

Share of voice

A business's share of all weighted citations recorded across every business measured in the combination.

share_of_voice = weighted_score_business / sum(weighted_score_all_businesses)

Inclusion criteria

A business is included in a combination if it meets the following criteria:

  • It operates in the measured vertical.
  • It serves the measured metro area.
  • It has a discoverable online presence, such as a website, directory listing, or public profile.
  • It is not excluded at the request of the business owner or for legal reasons.

We do not charge for inclusion. Rankings are determined entirely by measured citation behavior.

Correction process

We aim for accuracy in business names, websites, and classification. If a business believes its name, website, or vertical classification is incorrect, we will review the claim and update the index if warranted.

Corrections do not change citation counts or rankings. They only fix how a business is represented. To request a correction, use the contact form on this site and include the combination URL, the business name as shown, and the requested change.

Disclosure

High Clarity sells AI consulting, digital marketing, and related services to businesses. Some businesses that appear in the AI Visibility Index may be current or prospective clients. Measurement and ranking are independent of any commercial relationship.

Rankings are derived from model responses, not from client status, payment, or preference. This methodology is published so readers can evaluate the index on its own merits.