Table of Contents
Layer 1 — Visibility
00
The Measurement Framework
The four-layer structure that ties every metric in this guide together: Visibility, Perception, Influence, and Methodology. Each layer builds on the one before it, so measuring out of order tends to produce numbers that are technically correct but strategically misleading. Start here to understand what you are measuring, why the sequence matters, and how the chapters that follow connect to each other.
01
AI Competitive Map
The AI’s picture of your competitive market is often not your picture. Retrieval systems build their own competitive set based on patterns in what they have read, and that set can be outdated, incomplete, or simply wrong. This chapter shows how to find out who an AI places you alongside, who it leaves out entirely, and what a mismatched competitive map actually implies about your positioning.
02
AI Share of Voice
The most widely reported metric in AI visibility, and the most widely misunderstood. Most teams calculate Share of Voice inconsistently, which makes the number impossible to compare month to month or against competitors. This chapter shows how to calculate it correctly, what a high or low score actually implies, and where the metric can mislead you if read on its own.
Layer 2 — Perception
03
AI Topical Presence
Share of Voice tells you how often you appear. Topical Presence tells you what you appear for, which is a different and often more useful question. This chapter shows how to map the topics an AI associates with your brand, compare that map against your competitors, and identify the specific gaps worth closing first.
04
Factual Accuracy Rate
How accurately does AI describe your brand? Not how positively. Accurately. This chapter walks through how to audit what an AI gets wrong about you, the difference between a training-time error and a retrieval-time error, and why that distinction changes what you can actually do to fix it.
Layer 3 — Influence
05
Entity Map
The first chapter in this guide that changes the situation rather than measuring it. EntityMap is a structured knowledge layer you publish alongside your existing content, giving retrieval systems a declared map of who you are, what you claim, and how your concepts connect. This chapter covers why it reduces hallucinations, why it fixes ghost citations, and includes real measurement data from installing one on waikay.io.
06
Citation Data
Which sources is AI actually drawing on when it talks about your brand? This chapter shows how to find the specific URLs and domains shaping your AI visibility, and just as importantly, where competitors are being cited in places you are completely absent from. Citation data is often the fastest way to identify exactly which sources to target next.
Layer 4 — Methodology
07
AI Visibility Channels
Training data and live retrieval are two completely different systems, with different update cycles, different mechanics, and different levers you can actually pull. A fix that works for one does nothing for the other. This chapter explains how to tell which channel is producing a given AI response, and why that distinction changes everything about what to do next.
08
Prompt Tracking
Your measurements are only as good as the prompts behind them, and most prompt tracking introduces bias without anyone realising it. This chapter covers how to design and run prompt sets that produce reliable, repeatable data over time. It also covers the common mistakes that quietly distort results: leading phrasing, inconsistent sampling, and prompt sets too small to detect real change.
09
NLP and Entity Analysis
The linguistic layer sitting beneath every other metric in this guide. This chapter explains how AI systems actually build conceptual associations between your brand and the ideas surrounding it, and how those associations shift as new content gets published and indexed. Understanding this mechanism is what makes the difference between guessing why a score moved and knowing why it moved.
10
Data Gathering Methods
API access, scraping, and manual testing each come with real tradeoffs in cost, speed, and accuracy, and no single method is right for every situation. This chapter covers how to collect AI responses at scale without introducing the sampling biases that quietly make a dataset useless. It also covers how to combine methods so the weaknesses of one are covered by the strengths of another.
