Loading...
Loading...
Methodology
The AAA audit produces an AI Visibility Score out of 100 for every site it scans. This page explains what goes into that number, the research it's built on, and why each weight is set where it is. If you want the score for your own site, the audit runs free and returns a PDF in about 10 minutes.
Book a CallLast reviewed .
The stakes
of Google queries now trigger an AI Overview. CTR on the underlying organic result set drops by up to 61% when the Overview appears. Seer Interactive, 25.1M impressions (2025).
higher organic CTR for brands cited in AI Overviews vs competitors who don't appear in the cited sources. Same Seer study, segmented across 1,800 keywords.
visibility lift in generative engines when pages add named citations to credible sources. Aggarwal et al., Princeton, KDD 2024.
Framework
Presence asks whether AI crawlers can reach you. Readiness asks whether the content is structured to be cited. Business Impact weights findings by what they actually cost you. A site can pass Layer 1, fail Layer 2, and still be invisible.
Layer 1
Can AI crawlers reach and parse the page in the first place? GPTBot, ClaudeBot, and PerplexityBot do not execute JavaScript when crawling — pages that depend on client-side rendering for headings, copy, or JSON-LD return effectively empty to the crawler. The audit runs each page through a server-side render test simulating these user agents.
Layer 2
Once parsed, is the page structured to be cited? Statistics, named-source quotations, and outbound citations to credible sources are the three highest-leverage tactics in the Princeton GEO paper (+37%, +31%, and +40% lift respectively). Structured data — Article, HowTo, BreadcrumbList — is read by the citation extractors. FAQPage JSON-LD is no longer rewarded as of Google's 7 May 2026 deprecation.
Layer 3
What does AI visibility actually return? Seer Interactive's 25.1M-impression study found brands cited in AI Overviews see +35% higher organic CTR vs. competitors who don't appear. AI Overviews now run on roughly 48% of Google queries, with measured CTR drops of up to 61% on the AI-served result set. Layer 3 weights findings by commercial impact: a missing meta description on a service page beats a missing meta description on a privacy page.
Domains
The free audit scores all seven. SEO and Content remain the two largest weights (22% and 20%), with AI Visibility close behind at 18% — reflecting the v2.0 rebalance toward generative-engine readiness while keeping classical search fundamentals load-bearing.
Research
Adding statistics, citing credible sources, and including named quotations raises a page's visibility in generative search engines by up to 40 percent. Citation density is the single largest source of lift across all the tactics studied.
Adapted from Aggarwal, Murahari, Rajpurohit, Kalyan, Narasimhan & Deshpande, GEO: Generative Engine Optimization, ACM KDD 2024.
The audit's GEO domain operationalises this finding directly. Pages are scored on citation density (statistics + attributed quotes + outbound citations per 1,500 words), citable content blocks, and structured data. The thresholds are calibrated against the Princeton paper's Table 5 and Aleyda Solis's AI Search Optimization Checklist — specifically the “standalone-extractable” principle: every page should stand on its own, with the citation hooks an AI engine needs to pull it into an answer.
Methodology details
Vercel's December 2024 joint study with MERJ analysed crawler behaviour across the major AI engines and found that GPTBot, ClaudeBot, and PerplexityBot do not execute JavaScript. Pages built as a client-rendered shell — where the H1, main body, and JSON-LD only appear after a JS bundle loads — return an empty document to those crawlers. The audit replicates the GPTBot user agent and checks that the same critical content present in a default fetch is also present in the raw HTML response.
Google formally deprecated FAQPage rich results on 7 May 2026. Emitting FAQPage JSON-LD now risks being treated as legacy noise by both Google and the LLM crawlers that increasingly favour Article and HowTo schema. The audit flags FAQPage schema as a remediation finding, not a positive signal. Visible Q&A content on the page is still valuable — the change is in the structured-data wrapper, not the content.
The v2.0 weights are SEO 22%, Content 20%, AI Visibility (GEO) 18%, Performance 15%, Security 12%, UX & Design 10%, Technical 3%. SEO and Content together carry 42% — classical search fundamentals are still the largest single source of qualified traffic in 2026. AI Visibility climbed from 10% in v1.7 to 18% in v2.0, reflecting where the leverage is moving for South African mid-market businesses. The full weight table and per-check severity matrix live in the audit engine's geo-readiness-score-rubric — available on request.
Tell me about your site and the score you’re aiming for. 30 minutes, no pitch — just the next two or three things worth fixing.
Book a Call