EN
English✓ ENSvenskaSV
Start free trial
ProductAI Visibility IndexPricingLog in
← Back to the Formula 1 reportAI Visibility Index · Methodology · v1.0

How we measure who AI picks

Every edition of the AI Visibility Index uses the same method, so results can be compared across industries and over time. This page explains exactly how it works.

01

What we measure

Visibility scoreHow often a brand is mentioned in answers, weighted by how early it appears. 0–100.
Share of voiceA brand’s mentions as a share of all brand mentions in the category.
Average positionWhere a brand appears in an answer when it is mentioned. 1 is first.
SentimentHow positively the answer describes the brand. 0–100, where 50 is neutral.
AccuracyThe share of factual claims about a brand that match verified facts.
SourcesThe web pages an engine cites or links to when it answers.
02

The questions

Each edition uses 300–500 questions that real people ask about the category. We build the set in three steps:

  1. Collect common questions from our prompt volume estimates and public search data.
  2. Group them into intents: rankings and “best of”, comparisons, history, recommendations and buying.
  3. Neutralise the wording. No question names a brand unless the question is a direct comparison, and every brand in a comparison gets the same number of questions.
EXAMPLE QUESTIONS · FORMULA 1“Which F1 team is the most successful right now?”“What is the most iconic Formula 1 team?”“Which teams are on the 2026 F1 grid?”“Ferrari or McLaren: which team has the better car?”

The full question list for each edition is available on request to journalists and researchers.

03

Running the scan

ChatGPTPerplexityGeminiClaudeGoogle AI OverviewsDeepSeekMeta AI
Fresh sessionsEvery question is asked in a new session with no history, so earlier answers can’t influence later ones.
Repeated runsEach question is asked [N] times over [D] days. AI answers vary, so we average across runs.
Regions and languageRun from [regions] in [language]. Results can differ in other markets.
Default settingsEach engine’s default public model and settings at the time of the scan. Model versions are logged.
DatesEach report states its exact scan window.
04

Scoring

VISIBILITY SCOREvisibility = Σ (mention × position weight) ÷ number of answers × 100 position weight: 1st = 1.0 · 2nd = 0.8 · 3rd = 0.6 · 4th+ = 0.4A brand mentioned first in every answer scores 100. A brand never mentioned scores 0.
SHARE OF VOICEshare of voice = brand mentions ÷ mentions of all tracked brandsAdds up to 100% across the brands in the edition.
SENTIMENTeach mention classified as positive, neutral or negative sentiment = 50 + 50 × (positive − negative) ÷ all mentionsClassified automatically, with [X]% of mentions checked by a person.
ACCURACYaccuracy = correct factual claims ÷ all factual claims checkedClaims are compared with a fact sheet built from official sources and verified before the scan.
05

Checks and limits

Answers varyThe same question can get different answers. Repeated runs reduce, but don’t remove, this. Treat differences under [±N] points as a tie.
PersonalisationSigned-in users with history may see different answers than our fresh sessions.
Snapshot in timeEngines update often. A ranking describes the scan window, not the future.
Not real-world successThe Index measures how AI talks about brands, not sales, results or quality.
Human reviewA person reviews a random [X]% of answers and every claim marked as inaccurate.
06

Independence and corrections

No brand pays to be included, excluded or ranked. Brands can’t see results before publication. PromptedBrand customers are scored exactly like everyone else.

If you believe something in a report is wrong, email [[email protected]]. We review every request, and publish corrections with a note and date at the top of the report.

07

Using our data

You’re free to quote and share Index findings. Please credit “PromptedBrand AI Visibility Index” and link to the report.

SUGGESTED CITATIONPromptedBrand (2026). Formula 1 AI Visibility Index, Edition 1. promptedbrand.com/index/formula-1

Journalists and researchers can request the full dataset and question list at [[email protected]].

Run the same method on your brandPromptedBrand tracks your visibility with this methodology every day.
Run a free scan