The 4 AEO Signals Most Websites Miss in 2026

Most website audits check the same 30 or so signals and call it done. VisRank now checks 49 — including four that no other free scanner covers. These aren't edge cases. They're the signals AI engines actually look for when deciding which source gets cited. If they're missing from your site, you're invisible to ChatGPT, Perplexity, and Google AI Overviews — regardless of your Google ranking.

What are the 49 signals VisRank checks?

VisRank runs 49 automated checks across four categories. Each contributes to a weighted overall score: SEO (35%), AEO (25%), Security (25%), Local (15%). The AEO category expanded most — it now includes 18 checks covering not just schema markup but active AI discoverability signals.

The rest of this article covers the four new AEO checks in detail — what they measure, why they matter, and what a failing result looks like in your scan.

What is llms.txt — and why do AI engines look for it?

llms.txt is a plain-text file at /llms.txt that tells AI crawlers what content on your site is most important. It works like robots.txt — but instead of telling crawlers what not to index, it gives large language models a structured summary of your site so they can read the right pages first.

GPTBot (OpenAI), ClaudeBot (Anthropic), PerplexityBot, and Google's AIO crawler all support or are moving toward llms.txt as a discovery signal. A site without one forces each crawler to figure out its content structure independently — a slower, less reliable process. According to the llmstxt.org specification, the file format is intentionally simple: Markdown, with a short description, a list of your key URLs, and an optional summary of what each page covers.

A minimal llms.txt looks like this:

# YourSite — AI Crawler Index
# https://yoursite.com

## About
YourSite is a [brief description of what you do].

## Key pages
- Homepage: https://yoursite.com
- Services: https://yoursite.com/services
- Blog: https://yoursite.com/blog

## Content summary
[2–3 sentences describing your primary topics and audience.]

VisRank checks for the file with a HEAD request to {url}/llms.txt and flags it as a HIGH severity issue if missing — because this is one of the simplest, highest-leverage additions you can make to improve AI discoverability.

What is an AI Citability Score?

AI Citability measures how easily AI engines can extract and reproduce your content as a cited answer. It's distinct from readability for humans — it's about the structural patterns that large language models are trained to identify as "citable source material."

VisRank's citability check scans your page for ten patterns:

• Definition blocks — phrases like "X is a..." or "X refers to..." that give AI a clean entity definition to extract
• Statistics and percentages — numbered facts that AI engines treat as citable data points
• TL;DR summaries — explicit "In short" or "Key takeaway" blocks that AI Overviews pull directly
• Numbered steps — ordered lists that AI extracts as actionable procedures
• Comparison patterns — "vs." language and comparison tables that Perplexity and ChatGPT prioritise for head-to-head queries

A page with none of these patterns — just flowing prose paragraphs — scores low. Adding even two or three of them measurably increases your chance of appearing in an AI-generated answer. This is the same principle that makes answer-first content structure so effective for AEO generally.

Why content freshness matters for AI citation

AI engines — especially Perplexity and ChatGPT Search — strongly prefer recently updated content when multiple sources cover the same topic. A page with a visible 2026 date and a dateModified field in its schema will be chosen over an otherwise identical page with a 2023 date.

VisRank checks three freshness signals:

1. Current year present in text — does your page mention 2026? This is the most direct freshness signal for AI crawlers that parse visible content.
2. datePublished / dateModified in schema — JSON-LD Article or BlogPosting schema should include both properties. dateModified is the one AI engines weight most heavily.
3. Recency of the date value — if dateModified is more than 12 months old, VisRank flags it as a MEDIUM issue — a reminder to review and refresh the page.

This doesn't mean rewriting everything constantly. It means keeping your key pages genuinely current and making that freshness machine-readable. According to Google's Article structured data documentation, dateModified is a recommended property on all Article and BlogPosting schemas.

What is Speakable schema?

Speakable schema marks specific sections of your content as optimised for voice and audio playback. It uses "@type": "SpeakableSpecification" in JSON-LD (or data-speakable HTML attributes) to tell AI assistants and Google AI Overviews exactly which text to read aloud when a user asks a voice query.

Without Speakable markup, AI tools guess which parts of your page to surface for audio. They usually guess the meta description — which is rarely your best answer. A Speakable block lets you nominate your strongest, most concise answer paragraph as the one that gets read. Google officially supports Speakable for news content and is extending support to general web content as voice and AI Overview traffic grows.

A minimal Speakable JSON-LD block:

{
  "@context": "https://schema.org",
  "@type": "WebPage",
  "speakable": {
    "@type": "SpeakableSpecification",
    "cssSelector": ["h1", ".article-summary", "h2"]
  }
}

VisRank checks for Speakable schema in your JSON-LD blocks and for data-speakable attributes on HTML elements. It's flagged as LOW severity — worth adding, but not a blocker. We cover the full AEO signal landscape in Why ChatGPT Cites Your Competitors.

How to fix your AEO signals in 5 steps

1. Create /llms.txt. Add a plain Markdown file at your domain root. Describe your site in 2–3 sentences, list your 5–10 most important URLs with labels, and include a brief content summary. This takes about 15 minutes and has HIGH impact.
2. Add citability patterns to key pages. Review your homepage and top-traffic pages. Add at least one definition block, one numbered list, and one statistics reference. These additions also improve human readability — it's not an SEO-only change.
3. Update dateModified in your Article schema. Every time you revise a page, update the dateModified property in your JSON-LD. Add the current year to the visible text if it isn't already there. Set a recurring reminder to review key pages annually.
4. Add Speakable schema to pillar pages. Identify the 1–2 sentence summary that best answers your page's primary question. Wrap it in a SpeakableSpecification JSON-LD block targeting that element's CSS selector.
5. Verify with VisRank. Run a free scan → after each fix to confirm the check passes. New scans reflect changes immediately.

Key takeaways

• VisRank checks 49 signals across SEO, AEO, Security, and Local — more than any other free tool. The full AI search optimization guide explains all four categories.
• llms.txt — a 15-minute addition that signals your site's content structure to every major AI crawler. Flagged HIGH if missing.
• AI Citability — pages with definition blocks, statistics, TL;DRs, and comparison patterns get cited more. Prose-only pages get skipped.
• Content Freshness — dateModified schema + the current year in visible text are the fastest wins for improving AI's preference for your content over competitor pages.
• Speakable schema — lets you nominate exactly which text AI Overviews and voice assistants should read aloud, instead of letting them guess.
• These four checks appear in every VisRank scan under the AEO section — with severity levels, what passed and what failed, and a specific fix for each issue.