What Is an Authority Score and Why Should You Track It?
10 Mar 2026 · Bert Admin
A single number for a complex question
How well does AI know your brand? It's a simple question with a complicated answer. Your brand might be mentioned frequently but described negatively. You might be cited by authoritative sources but never actually recommended. You might appear in some AI models but be invisible in others.
Authority Score (0-100) is a composite metric that combines these dimensions into a single trackable number. It's not a vanity metric — it's a weighted calculation across five specific aspects of how AI models perceive your brand.
The five dimensions
Authority Score is built from five stages of the AI Perception Funnel, each measuring a different aspect of your brand's AI visibility:
- Presence (15%) — Does the model know your brand exists? Open-ended category queries reveal whether you appear at all when your industry is discussed.
- Perception (20%) — How does the model describe your brand? Descriptive queries uncover the adjectives, associations, and sentiment AI models attach to you.
- Positioning (25%) — Where does the model place you versus competitors? Comparative queries show your relative standing in the competitive set.
- Recommendation (25%) — Does the model actively recommend you? Decision queries test whether AI models suggest your brand when users ask for recommendations.
- Resilience (15%) — What risks or vulnerabilities does the model surface? Probing queries reveal negative associations, outdated information, or competitive threats.
How it's calculated
For each dimension, we analyse brand mentions, citation sources, sentiment signals, and competitive positioning across multiple AI models and prompts. The raw signals are normalised and weighted to produce the composite Authority Score.
Confidence tiers indicate how reliable the score is:
- Directional — from a Quick run with a single observation. Useful for a first look, but treat as indicative rather than definitive.
- Reliable — from a Standard run with 3 observations per prompt. Statistical noise is reduced and patterns are more stable.
- Verified — from a Deep run with 10+ observations per prompt. Statistically significant at the prompt level with confidence intervals.
Why track it over time?
A single Authority Score snapshot is useful, but the real value comes from tracking it monthly. AI models update their training data and behaviour regularly. Your competitors are publishing content, earning citations, and shifting the landscape.
A rising Authority Score confirms your strategy is working. A declining score is an early warning that something has changed — maybe a competitor has improved, maybe negative content has entered the training data, maybe a previously strong citation source has lost relevance.
After three months of tracking, you have trend lines that no competitor can replicate retroactively. That historical data becomes one of your most valuable strategic assets.