AI Optimisation Metrics: the 20 measurements that determine your AI visibility
A comprehensive guide to the metrics SCANPIRE uses to analyse how AI systems discover, understand, and interact with your website content, organised across five thematic categories.
As AI systems increasingly determine which websites get surfaced, cited, and acted upon, understanding the specific metrics that drive AI visibility has become a strategic imperative. SCANPIRE analyses your website across 20 key optimisation metrics, each derived from real scan data and mapped to one or more of five optimisation types: AEO, GEO, LLMO, AAIO, and AISEO. This guide breaks down every metric, explains what it measures, why it matters, and how it is calculated.
The five optimisation types
Before diving into the individual metrics, it is important to understand the five optimisation types that form the foundation of AI readiness measurement.
- AEO. Answer Engine Optimisation.
- Be selected as the direct answer by AI platforms like Google AI Overviews and Perplexity.
- GEO. Generative Engine Optimisation.
- Be cited and summarised in LLM-generated responses from ChatGPT, Claude, and Gemini.
- LLMO. Large Language Model Optimisation.
- Ensure content is interpretable, discoverable, and parseable by AI systems.
- AAIO. AI Agent Interoperability Optimisation.
- Enable machine-actionable interaction by autonomous AI agents.
- AISEO. AI-Enhanced SEO.
- Bridge traditional search optimisation with AI-specific ranking and visibility requirements.
AI discovery and visibility
These four metrics measure how effectively AI systems find and surface your content. They answer the fundamental question: when someone asks an AI system about your industry, products, or services, does your website appear in the response?
Prompt Share
The likelihood of your content being surfaced when users submit prompts related to your topic area. It maps directly to the AI Content Citation Pillar and is driven by both AEO and GEO optimisation. A website with strong Prompt Share appears frequently in AI-generated responses, while a low score means AI systems are citing competitors instead.
Algorithmic Preference Score
Quantifies the selection bias that large language models exhibit toward your content. When multiple sources contain similar information, LLMs develop preferences based on content structure, authority signals, and semantic clarity. Maps to the Generative LLMO Pillar.
AI Indexation Rate
Goes further than traditional indexation. It evaluates whether AI crawlers can effectively discover, parse, and understand your content. Includes your robots.txt policies for AI bots, structured data availability, and content accessibility for AI-specific crawlers like GPTBot and ClaudeBot.
Citation Inclusion Rate
The percentage of AI-generated answers that would cite your domain. A composite metric calculated as (AEO + GEO + AISEO) / 3, combining your performance across answer engines, generative engines, and AI-enhanced search. One of the most actionable metrics because improvements in any of the three contributing categories directly boost your citation rate.
Content authority and trust
These four metrics assess how AI systems perceive the quality and credibility of your content. Even if AI discovers your website, it will not cite content it does not trust.
Answer Coverage
Your topical authority for direct answers, mapping directly to your AEO category score. Websites with high Answer Coverage have comprehensive, well-structured content that directly addresses user questions across their domain. Critical for appearing in featured snippets, AI Overviews, and answer engine results.
Entity Authority Score
The strength of your entity representation across AI systems. Calculated as (AAIO + GEO) / 2, it evaluates whether AI systems recognise your brand, products, or services as distinct, authoritative entities. Strong entity authority means AI platforms are more likely to surface your content when users ask about your specific domain.
AI Trust Signal Density
AI systems evaluate trustworthiness through signals like authorship attribution, source citations, content provenance, and E-E-A-T (Experience, Expertise, Authoritativeness, Trustworthiness) markers. Calculated from (E-E-A-T Signals + Source Credibility) / 2.
Semantic Coherence Score
Measures whether your headings, body content, schema markup, and metadata all tell a consistent story. Calculated as (Content Structure + Structured Data) / 2. Essential because AI systems that detect semantic inconsistencies are less likely to trust and cite your content.
AI agent interoperability
As AI agents increasingly handle tasks on behalf of users, from booking appointments to completing purchases, these four metrics measure how well your website supports automated interactions.
Conversion via AI Mediation
Task completion through AI agents, mapping directly to the Agentic and MCP Interoperability Pillar. As more users delegate tasks to AI assistants, websites that support AI-mediated conversions will capture revenue that others miss entirely.
Actionability Score
How effectively AI agents can complete tasks on your website, mapping directly to the AAIO category. This includes form structures, action buttons, machine-readable workflows, and API-accessible functionality. A low Actionability Score means AI agents cannot reliably interact with your site on behalf of users.
Agent Failure Rate
The percentage of AI agent attempts that would fail on your website. Calculated as 100 - ((Error Handling + API Access) / 2). Every failed agent interaction represents a lost conversion opportunity. Reducing your Agent Failure Rate requires improving error handling, providing clear API access points, and designing agent-safe interaction flows.
Task Completion Latency
The actual time it takes for AI agents to complete actions on your website. Currently marked as Not Tracked because it requires runtime testing with real AI agents, something that cannot be determined through static analysis alone. As AI agent testing capabilities mature, this metric will become available in future SCANPIRE releases.
Technical AI readiness
These three metrics assess the technical infrastructure that enables AI systems to efficiently crawl, interpret, and interact with your website.
Crawl Efficiency Ratio
How much useful content is indexed compared to the total content crawled. Calculated as (Robots Policy + Feeds/Sitemaps + Tech SEO Infrastructure) / 3. A low ratio means AI crawlers are wasting their crawl budget on low-value pages, navigation elements, or encountering barriers that prevent them from reaching your important content.
Temporal Relevance Score
AI systems need to determine content freshness and temporal relevance for time-sensitive queries. Calculated as Content Structure x 0.6 + Feeds/Sitemaps x 0.4, with more weight given to content structure because well-organised content with clear date signals is more useful to AI than updated sitemaps alone.
Multimodal Eligibility Rate
As AI interactions expand beyond text to voice assistants, smart displays, and visual search, this measures what percentage of your content is usable across these channels. Calculated as (Voice Optimisation + Responsive Design + Smart Display Surfaces) / 3.
Overall health indicators
These five aggregate metrics provide a high-level view of your AI optimisation posture across all pillars and categories. They are especially useful for executive reporting and strategic planning.
Remediation Urgency Index
A weighted priority score calculated as (High Priority x 3 + Medium Priority x 2 + Low Priority x 1) / Total Recommendations. A higher index means your site has proportionally more critical issues. Helps teams prioritise their optimisation efforts by focusing on high-impact fixes first.
Pillar Balance Ratio
How evenly distributed your scores are across all five AI readiness pillars, calculated as 1 - (Standard Deviation / Mean) x 100. A score of 100 means perfectly balanced performance. An imbalanced profile suggests that while you may excel in some areas, critical weaknesses in others are undermining your overall AI visibility.
Category Coverage Score
The percentage of scan categories that exceed a 60-point threshold, calculated as Count(Categories > 60) / 7 x 100. A quick snapshot of how many areas of your AI readiness are adequately covered.
Critical Gap Count
The raw number of scan categories scoring below 50. Each critical gap represents a significant weakness that could substantially impact AI visibility. The goal is to reduce this count to zero through targeted optimisation of the lowest-scoring categories first.
AI Optimisation Velocity
The average performance across the four AI-specific categories, calculated as (AEO + GEO + AISEO + AAIO) / 4. A single number that represents your overall AI optimisation momentum, excluding traditional SEO and accessibility metrics. Particularly useful for tracking progress over time and benchmarking against industry standards.
How to use these metrics
- 01Start with Overall Health Indicators. Check your Category Coverage Score and Critical Gap Count to understand the big picture. If you have critical gaps, address those first.
- 02Focus on Discovery. Without AI Discovery and Visibility, the other metrics become irrelevant. Ensure your Prompt Share and AI Indexation Rate are at least 'Good' before optimising deeper metrics.
- 03Build Trust. Work on Content Authority and Trust metrics by improving E-E-A-T signals, entity authority, and semantic coherence.
- 04Prepare for Agents. AI Agent Interoperability is the fastest-growing category. Reducing your Agent Failure Rate now positions you ahead of competitors.
- 05Optimise the Technical Foundation. Crawl Efficiency and Temporal Relevance ensure AI systems can efficiently access and prioritise your content.
Scott King
Scott King is the Growth & Innovation Principal for Asia Pacific within Adobe's Digital Strategy Group, and a leading AI subject matter expert across the region. Founder of Scanpire.com, the AI readiness analytics platform. Previously, Scott founded the customer experience consultancy Accordant before its acquisition by Merkle Dentsu, where he served as Vice President, Enterprise Solutions.