BrainScore Methodology — How We Measure AI Visibility

Overview

What BrainScore Measures

BrainScore measures how well AI search engines understand, find, and recommend your brand. We run your website through 5 diagnostic dimensions, each scored 0–20, totaling 0–100.

Unlike traditional SEO tools that measure how Google ranks you, BrainScore measures how AI models like ChatGPT, Claude, and Perplexity represent you when customers ask them for recommendations.

The 5 Dimensions

How We Score Each Area

A1 Structural Readiness 0–20 pts

"Can AI find and read your website?"

What We Check

robots.txt (AI crawler permissions)
llms.txt (AI-specific instructions)
schema.org markup
Meta descriptions
Open Graph tags
sitemap.xml

Scoring Rubric

17–20Comprehensive structured data, AI-specific instructions, rich schema markup
14–16Core attributes properly structured and indexed
11–13Basic structure exists but lacks differentiating attributes
8–10Minimal structure, AI struggles to parse
0–7AI crawlers blocked or no meaningful structure

A2 Semantic Clarity 0–20 pts

"Does AI understand what you do — in YOUR words?"

What We Test

We ask AI models to describe your brand in 3 sentences, then compare their vocabulary against your website's own positioning language.

Scoring Rubric

17–20AI uses your brand's own vocabulary and positioning unprompted
14–16Accurate positioning reflected, mostly uses your language
11–13AI describes you relationally ("like [competitor] but...")
8–10AI defines you primarily as an alternative to a competitor
0–7AI contradicts your brand positioning or can't describe you

A3 Synthetic Customer Test 0–20 pts

"Would AI recommend you to a customer?"

What We Test

We ask multiple AI models category-level questions WITHOUT mentioning your brand (e.g., "What are the best [your industry] companies?") and check if your brand appears.

Scoring Rubric

17–20Recommended 1st across multiple models and query types
14–16Recommended in 3–4 query types, top position in at least one
11–13Mentioned in 1–2 query types, secondary tier
8–10Only surfaces when category is named directly
0–7Not mentioned in any recommendation queries

Models tested: Claude (Anthropic) + GPT-4o (OpenAI)

A4 Emotional Residue 0–20 pts

"Does AI capture your brand's personality?"

What We Test

We ask AI "What cultural associations does [brand] have?" and score the specificity of its response.

Scoring Rubric

17–20AI names specific cultural communities, movements, or moments
14–16Clear emotional associations with at least one specific cultural reference
11–13Generic emotions only ("innovative," "fun") — no specific references
8–10Purely relational identity (compared to competitors, not independent)
0–7No emotional associations; purely functional description

A5 Voice & Archetype 0–20 pts

"Does your brand voice survive without visuals?"

What We Test

We ask AI to describe your brand's personality and identify its archetype (Hero, Sage, Outlaw, etc.). We check if the voice is distinctive and consistent.

Scoring Rubric

17–20Specific archetype named unprompted, consistent across pages, survives visual removal
14–16Recognizable voice with at least one strong verbal identity element
11–13Voice relies on visual/platform references
8–10No clear archetype; personality collapses without visuals
0–7Generic corporate language; interchangeable with competitors

Tier Thresholds

What Your Score Means

Score	Tier	What It Means
83–100	Awesome	AI accurately represents and actively recommends your brand
70–82	Strong	AI understands your brand well; reliable visibility with minor gaps
56–69	Average	AI has partial understanding; significant room for improvement
40–55	Weak	AI struggles to represent your brand accurately
0–39	Invisible	AI doesn't know your brand exists or actively misrepresents it

Our Process

How a Scan Works

1

Structural Scan

We crawl your website checking for robots.txt, llms.txt, schema.org, meta tags, and OG tags.

2

AI Model Queries

We ask Claude and GPT-4o about your brand using multiple prompts designed to test each dimension.

3

Gap Analysis

We compare what your website says vs what AI tells people — identifying misalignment and blind spots.

4

Scoring

Each dimension scored 0–20 based on the rubric above. Total score: 0–100.

5

Recommendations

Prioritized actions with estimated point improvement so you know exactly what to fix first.

Why BrainScore

What Makes BrainScore Different

Free and instant

Most AI visibility audits cost $500+ and take weeks. BrainScore is free and takes 30 seconds.

Transparent methodology

This page explains exactly how we score. No black boxes.

Real AI responses

We show you actual quotes from AI models about your brand.

Actionable

Every scan includes specific recommendations with estimated point impact.

Trackable

Rescan monthly to measure improvement over time. Watch your BrainScore climb as you implement changes.

FAQ

Common Questions

How often should I scan?▼

Monthly. AI models update their knowledge regularly. Your score will change as you implement recommendations.

Can I improve my score?▼

Yes. Most brands see a 20–40 point improvement within 90 days by implementing the recommendations.

What AI models do you use?▼

Currently Claude (Anthropic) and GPT-4o (OpenAI). We're adding Perplexity and Gemini.

Is my data private?▼

Your scan results are stored securely. We don't share individual brand data publicly without permission.

How is this different from SEO?▼

SEO measures how Google's algorithm ranks your website. BrainScore measures how AI language models understand and recommend your brand. They're complementary but different.