LLM Visibility Playbook: How to Rank in ChatGPT, Claude, Perplexity & Gemini

If someone asks ChatGPT "what's the best tool for X" and your brand isn't mentioned, you have a visibility problem — and it's not a Google problem. It's an LLM problem, and fixing it requires a different playbook than traditional SEO.

This guide gives you the full step-by-step framework for improving your brand's visibility across ChatGPT, Claude, Perplexity, and Gemini. It's structured as a playbook: sequential steps, each building on the last, with measurable outcomes.

Why LLM Visibility Matters Now

These numbers matter because the usage pattern is different from traditional search. When someone asks an LLM a commercial question — "what CRM should I use?", "best accounting software for freelancers?" — they're not getting 10 blue links. They're getting a single narrative answer with 2–5 sources. If you're not in those sources, you don't exist for that query.

Our data across 500+ brands shows that 74% score below 40/100 on the AIS Index — meaning most brands are nearly invisible in AI search despite being well-established in Google results. The LLM citation landscape is still early, which means there's significant upside for brands that move now.

How LLMs Decide What to Cite

LLMs don't crawl the web at query time. They operate from a knowledge base built during training (updated on different schedules per engine) and, for Perplexity specifically, from live web retrieval. Understanding this distinction is foundational to this playbook.

What earns you a citation in LLM responses:

• Training data inclusion — Your content needs to be indexed and included in training crawls. This means being on authoritative domains (or getting cited by them).
• Authority signal accumulation — LLMs weight content from Wikipedia, Crunchbase, G2, industry publications, and high-domain-authority sources. Getting your brand represented there is not optional.
• Semantic relevance — Content that clearly, factually, and directly answers the types of questions your audience asks. LLMs cite answers, not landing pages.
• Structured accessibility — Content that's easy to parse: clear headings, concise answers, structured data markup, and explicit metadata about who you are and what you do.
• Retrieval eligibility — For Perplexity (and increasingly other engines with live retrieval), your content needs to be indexed, accessible to crawlers, and recently updated.

Step 1: Get Your Baseline Score

You can't optimize what you can't measure. Before doing any of the tactical work below, run a scan to get your current AIS Index score across all four engines.

The scan returns:

• An overall AIS Index score (0–100)
• Per-engine scores broken down by Visibility, Authority, Sentiment, and Adversarial resilience
• Top citation gaps — the specific areas where competitors are being cited and you're not
• Quick wins — the 5 highest-leverage fixes for your specific score profile

Run your free baseline scan →

This baseline matters because the tactics in this playbook have different ROI depending on your weakest dimension. If your Visibility score is low (you're not being mentioned at all), the fix is different from a low Authority score (you're mentioned but not trusted as a source). The scan tells you where to start.

Step 2: Set Up llms.txt

The llms.txt file is to LLMs what robots.txt is to traditional search crawlers — but instead of restricting access, it's a machine-readable file that tells AI crawlers exactly what your brand is, what you do, and which pages to prioritize for ingestion.

Place this file at https://yourdomain.com/llms.txt. Structure it in Markdown:

# [Your Company Name]

> [One sentence: what you do and who you serve]

[2-3 paragraph factual description. No marketing language. Think Wikipedia-style:
what the company does, when it was founded, what problems it solves, key differentiators.]

## Key Pages

- [Homepage](https://yourdomain.com): [Brief description]
- [Product](https://yourdomain.com/product): [Brief description]
- [Pricing](https://yourdomain.com/pricing): [Brief description]
- [Documentation](https://docs.yourdomain.com): [Brief description]

## What We Do

[Bulleted list of core use cases — written as factual capabilities, not marketing claims]

## Company Facts

- Founded: [Year]
- Headquarters: [City, Country]
- Category: [Industry/category]
- Customers: [Scale / key use cases]

The key principles for an effective llms.txt:

• Factual, not promotional — Write like an encyclopedia, not a sales deck. Claims like "leading solution" or "best platform" are noise to LLMs. Factual statements about what you do are signal.
• Complete sentences about your category — Include the words that define your category clearly. If you're a "project management tool for engineering teams," say that explicitly.
• Link to your most authoritative pages — The pages most likely to answer questions about your brand, product, or use cases.
• Keep it under 1,000 words — Long, dense files are less reliably parsed. Clarity over completeness.

Step 3: Add Structured Data

Schema.org structured data doesn't just help Google — it provides LLMs with unambiguous, machine-readable facts about your organization. At minimum, add Organization schema to your homepage and product pages.

<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "Organization",
  "name": "[Your Company]",
  "url": "https://yourdomain.com",
  "description": "[One-sentence factual description]",
  "foundingDate": "[Year]",
  "sameAs": [
    "https://en.wikipedia.org/wiki/[YourCompany]",
    "https://www.crunchbase.com/organization/[slug]",
    "https://www.linkedin.com/company/[slug]",
    "https://g2.com/products/[slug]"
  ],
  "areaServed": "Worldwide",
  "knowsAbout": ["[Topic 1]", "[Topic 2]", "[Topic 3]"]
}
</script>

The sameAs property is critical. It tells LLMs that your domain is the same entity as your Wikipedia page, Crunchbase listing, G2 profile, and LinkedIn presence. This entity resolution is how LLMs consolidate authority signals — without it, a mention on G2 and a mention on your website might be treated as different entities.

Beyond Organization schema, add per-page schema types that match your content:

• Product on product/pricing pages with offers, price ranges, and reviews
• HowTo on tutorial and guide pages
• FAQPage on FAQ sections (see Step 4)
• Article on blog posts and guides (with datePublished, dateModified)
• BreadcrumbList on all pages for navigation context

Step 4: FAQ Schema for Answer Eligibility

FAQ schema is your most direct path to getting cited in AI search. When LLMs look for answers to specific questions, they prioritize content that clearly states both the question and the answer — and FAQ schema makes that structure machine-readable.

The pattern that works:

<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "FAQPage",
  "mainEntity": [
    {
      "@type": "Question",
      "name": "What is [your product category]?",
      "acceptedAnswer": {
        "@type": "Answer",
        "text": "[Your company name] is a [category] tool that [specific functionality]. It [what it does for users]. Key features include [feature 1], [feature 2], and [feature 3]."
      }
    },
    {
      "@type": "Question",
      "name": "How does [your product] compare to [competitor]?",
      "acceptedAnswer": {
        "@type": "Answer",
        "text": "[Factual comparison. Be accurate. LLMs will cross-reference claims.]"
      }
    }
  ]
}
</script>

FAQ pages to create (each targeting a question your prospects actually ask LLMs):

• "What is [product name]?" — The entity definition question
• "What does [product name] do?" — The functionality question
• "How much does [product name] cost?" — Pricing queries
• "Who uses [product name]?" — Customer profile queries
• "What are alternatives to [competitor]?" — Comparison queries where you want to appear
• "How to [use case you solve]?" — Workflow and how-to queries

Step 5: Create Citation-Worthy Content

LLMs don't cite landing pages. They cite content that authoritatively answers specific questions. The gap between what most brands publish (feature descriptions, use cases, customer stories) and what LLMs cite (research, data, frameworks, direct answers) is the core content gap to close.

Content formats LLMs cite most

The format matters as much as the content. Structure your citeable assets for machine parsing:

• Lead with a direct answer — Answer the question in the first sentence. LLMs sample early in the document.
• Use numbered and bulleted lists — Lists are structured data, not prose. They're easier to extract and cite.
• State numbers explicitly — "74% of brands score below 40" is citable. "Most brands score poorly" is not.
• Use explicit headings — H2 and H3 tags that mirror the questions your audience asks LLMs.
• Keep paragraphs under 100 words — Dense paragraphs are parsed less reliably than tight, focused ones.

Step 6: Build Authority Signals

Authority is the second-largest component of the AIS Index (30%), and it's determined by where your brand appears across the web — not just your own domain. LLMs heavily weight signals from specific high-authority sources:

Priority authority sources

• Wikipedia — The single highest-authority source for LLMs. If your company is notable enough, get a Wikipedia page or ensure your company is accurately described in relevant Wikipedia articles. This is not optional for brands aiming for a score above 70.
• Crunchbase — Particularly for B2B software. Ensure your company profile is complete, accurate, and up-to-date with a clear description, industry tags, and founding date.
• G2 and Capterra — Review platforms that LLMs use as authority signals for software products. A complete profile with reviews signals legitimacy and citability.
• Industry publications — Guest articles, bylines, and coverage in industry publications (TechCrunch, Forbes, vertical trade publications) directly feed LLM training data.
• Podcast mentions and transcripts — Podcast content is increasingly part of LLM training corpora. Industry podcast appearances are an underrated citation asset.
• LinkedIn profile completeness — LinkedIn is indexed and included in several LLM training sets. A complete, accurate company page with regular posts contributes to authority.

The principle: citation moats are built on the open web, not your own domain. Your content is the seed; third-party citations are the compounding asset.

Step 7: Measure and Iterate

LLM visibility is not a one-time optimization. Knowledge bases update, competitors act, and your score reflects the current competitive landscape — not just your absolute effort.

A sustainable measurement cadence:

• Baseline scan — Before you start (Step 1)
• Post-implementation scan — 4–6 weeks after each major change (new content, structured data, authority campaign)
• Monthly monitoring scan — Track score trends and catch drops before they compound
• Competitor comparison quarterly — Check if competitors are gaining relative to your absolute improvements

What to look for in your score trends:

• Visibility score improving? Your content is being ingested. Keep creating.
• Authority score flat? You need more third-party coverage. Prioritize the authority source list above.
• Sentiment score declining? Check for hallucinations — AI engines making false or negative claims about your brand. This requires active monitoring and correction via official channels.
• Adversarial score dropping? Your brand has weak defenses against negative associations. Content clarifying what you are (and aren't) helps here.

Engine-Specific Tactics

The four major LLMs have meaningfully different behaviors that affect what works:

ChatGPT (OpenAI)

Training data cutoff creates a lag between publishing content and appearing in responses. Prioritize content that's been indexed and available for a full training cycle. ChatGPT heavily weights Wikipedia and Crunchbase for entity information. For product queries, G2 and review platforms matter.

Perplexity

Perplexity performs live retrieval — it actually fetches pages at query time, not just from training data. This means your SEO fundamentals matter here more than with other engines. Fresh content, fast page loads, accessible crawling, and recent publish dates all influence Perplexity citations directly.

Claude (Anthropic)

Claude shows a stronger preference for authoritative, well-structured content. The "first 150 words" pattern matters more here — Claude often cites content that opens with a direct, comprehensive answer rather than context-setting or narrative buildup. Technical documentation and specification-style content performs well.

Gemini (Google)

Gemini has access to Google's full index and knows your domain's existing authority score. This makes Gemini the engine where traditional SEO strength (domain authority, backlinks, engagement signals) carries the most weight. However, it also has a bias toward Google-native properties (YouTube, Google Business, Google Workspace docs), so presence there matters.