AEO 2026: How to Get Cited by ChatGPT, Claude, and Perplexity

Your Site Is Probably Invisible to AI Engines Right Now

When a buyer asks ChatGPT "what's the best B2B CRM for Southeast Asia" or asks Perplexity "top marketing automation tools in Singapore," your site either appears in the answer or it doesn't. There is no page two. There is no paid slot. There is no second chance in that session.

We built the Cyberg7 AEO audit tool specifically to measure this problem. Since launching in May 2026, we have run 17 audits across 14 distinct domains, and the average score is 23 out of 100 — a flat Grade F (data extracted 2026-05-26). These are not obscure hobby blogs. They are active B2B SaaS and agency sites that rank on Google, run paid campaigns, and have real traffic. They are simply invisible to the AI engines their buyers now use first.

What "Being Cited" Actually Means in 2026

Answer Engine Optimization — AEO — is the practice of structuring your content so that generative AI systems surface and attribute it when answering user queries. It is distinct from traditional SEO in one critical way: the goal is not a ranked blue link, it is a sentence or paragraph that the AI quotes or paraphrases, usually with a source link attached.

The academic framing comes from a peer-reviewed paper accepted at KDD 2024. Pranjal Aggarwal and co-authors at arXiv:2311.09735 write that "we introduce Generative Engine Optimization (GEO), the first novel paradigm to aid content creators in improving their content visibility in generative engine responses" (Aggarwal et al., arXiv, 2024). That paper is the first rigorous treatment of the concept, and its core finding — that certain content signals reliably increase the probability of being cited — is the foundation of everything that follows.

The three AI engines that matter most for B2B buyers right now are ChatGPT (GPT-4o and above), Claude (Sonnet and Opus), and Perplexity. Each indexes the web differently, but all three reward the same underlying content properties: factual density, structural clarity, and crawlability.

The Five Signals That Determine Whether You Get Cited

Grounding in the GEO research and in our own audit data, we have identified five signals that determine citation probability.

Signal	What AI engines look for	Common failure mode
Crawlability	GPTBot and similar agents can fetch the page	robots.txt blocks AI crawlers
Factual density	Specific numbers, dates, named entities	Vague brand copy ("we help you grow")
Structured answers	Headers, lists, tables that answer a single question	Wall-of-text about-us pages
Named authorship	A real team or individual with credentials	Anonymous "the team" with no detail
Citation-worthy claims	Original data, research references, primary sources	Unsourced assertions

These five signals map directly onto the scoring dimensions in our audit tool. Sites that fail on all five — the majority we have seen — score below 30. Sites that fix even two of them reliably cross 50.

The Evidence: What the Research and Our Audits Show

Crawlability is the baseline, and many sites fail it

OpenAI publishes its crawler documentation openly. According to OpenAI's GPTBot documentation, site owners can disallow GPTBot via robots.txt, and many do so accidentally — inheriting a blanket Disallow: / rule from an old security template or a misconfigured CDN. If GPTBot cannot fetch your page, it cannot cite it. This is the single fastest fix available: check your robots.txt today.

Factual density is the differentiator

The GEO paper (Aggarwal et al., arXiv, 2024) tested which content modifications produced the largest lift in generative engine visibility across 10,000 queries. Adding statistics and citing authoritative sources produced significantly higher citation rates than adding fluent prose alone. In plain terms: a page that says "our platform reduces onboarding time by 40% based on data from 200 customer accounts" is more citable than one that says "our platform dramatically speeds up onboarding."

In our own audit data (17 audits, 14 domains, extracted 2026-05-26), the single most common failure we see is brand copy written for human emotional persuasion rather than factual retrieval. Marketing teams optimize for conversion rate on a landing page. AI engines optimize for factual precision in a corpus. These are different objectives, and most sites have only pursued the first one.

Structure signals answer intent

Perplexity, in particular, tends to cite pages where the answer to a specific question appears within the first two paragraphs or under a clear H2 header. A detailed breakdown of how AEO differs from SEO and what signals Perplexity prioritizes is covered in the Cyberg7 AEO resource at aivisibility.cyberg7.com.sg, which tracks citation patterns across multiple AI engines. The core pattern: pages that answer one question completely, with headers that match the question phrasing, get cited more than pages that cover many topics loosely.

A broader practitioner summary of these patterns, with worked examples from real domains, is documented in the AEO 2026 guide on dev.to, which consolidates testing observations from practitioners across multiple markets.

Five Steps You Can Execute This Week

The following steps are ordered by effort-to-impact ratio, easiest first.

1. Audit your robots.txt for AI crawler blocks. Fetch yourdomain.com/robots.txt and check for User-agent: GPTBot followed by Disallow: /. If it exists, remove the disallow or replace it with a specific path exclusion. Do the same for ClaudeBot and PerplexityBot. This takes under 30 minutes and unblocks everything downstream.

2. Add one data-backed claim per key page. Pick your three highest-traffic pages. On each, replace one piece of vague brand language with a specific, sourced claim. Format: "[Specific number or outcome] based on [source or sample size] as of [date]." This directly increases factual density, the highest-impact signal in the GEO research.

3. Rewrite your H2 headers as questions or direct answers. Instead of "Our Approach," write "How We Reduce Onboarding Time for B2B SaaS Teams." AI engines pattern-match headers to query intent. Headers that read like search queries or direct answers get pulled into responses more often than generic navigation labels.

4. Add a named author block with credentials to every article or resource page. Include name, role, and one verifiable credential (years of experience, number of clients, a named publication). This addresses the authorship signal and also improves E-E-A-T signals that influence which sources AI engines treat as authoritative.

5. Publish at least one page with original data. It does not need to be a research report. A single table summarizing your own customer data, audit results, or platform metrics — with a methodology note and extraction date — gives AI engines something to cite that no competitor can replicate. This is the step most B2B sites skip because it requires internal data access, but it produces outsized citation lift.

Each of these steps is reversible, testable, and measurable. Run an AEO audit before and after each change to confirm the score movement.

Related Reading

• What is AEO? How to Get ChatGPT, Perplexity & AI Search Engines to Cite Your Website

Run Your Own Audit

If you want to see exactly where your site scores across all five signals, our free audit at aivisibility.cyberg7.com.sg/audit returns a scored report in under two minutes — no login required.