Learn · Scoring Model

The AI Visibility Score — 5 layers, never collapsed.

By Antonia Sevilla · Founder of HoldThatLast updated June 2026
Short answer
The HoldThat AI Visibility Score is a 5-layer model: Technical Access, Semantic Understanding, Entity & Trust, AI Answer-Surface Presence, and Business Impact. Each layer is scored and reported separately — we never collapse them into a single number, because doing so would hide where the actual problem is.

Why 5 layers, not one score

A single AI visibility score sounds tidy, but it hides where the actual problem is. A brand can be technically perfect and still invisible because no AI engine has reason to cite it. Another brand can be cited widely but losing pipeline because the downstream funnel leaks.

HoldThat keeps the five layers separate so the diagnosis stays honest. Every audit reports them independently — and every recommendation is tied to a specific layer.

The 5 layers

01

Technical Access

Can AI crawlers — GPTBot, PerplexityBot, Google-Extended, ClaudeBot, CCBot — actually reach, fetch, and render your content. Measures crawl access, server vs. client rendering, structured data coverage, and canonical hygiene.

02

Semantic Understanding

Is your content structured around the questions buyers actually ask. Measures topic-cluster coverage vs. buyer intent, answer-readiness of priority pages, and internal linking from pillars to supporting content.

03

Entity & Trust

Is your brand recognized as a real entity with credible third-party signals. Measures entity-graph completeness, author bios, third-party citations, and reputational mentions across sources AI engines treat as authoritative.

04

AI Answer-Surface Presence

Do real prompts surface your brand inside real AI answers. Measures citation rate, citation position, and competitor displacement across ChatGPT, Perplexity, Gemini, Claude, and Google AI Overviews.

05

Business Impact

Is AI-driven discovery translating into measurable downstream outcomes. Measures AI-assisted sessions, assisted conversions, and pipeline contribution from AI-referred traffic.

How each layer is scored

Each layer is scored against the metrics defined for it — for example, schema coverage for Technical Access, intent-mapped topics for Semantic Understanding, and cited prompts per engine for AI Answer-Surface Presence. Scores are reported per layer alongside the underlying evidence.

The methodology is documented end-to-end on How It Works.

What this is not

  • It is not a single number you can put on a dashboard and call done.
  • It is not a black-box AI scorer with no underlying evidence.
  • It is not a repurposed SEO metric pretending to measure AI visibility.

See it in practice

The fastest way to understand the model is to see one layer applied to your brand. The free teardown shows you AI Answer-Surface Presence in action with one real prompt. The full audit runs all five layers across all five engines, with evidence and a prioritized roadmap.

Frequently asked

Free · 48-hour delivery

See your own AI visibility gap.

One real prompt. One competitor comparison. One short Loom. No sales call required.