Guide / AI Search

What is AI Search Optimisation?

AI Search Optimisation is the work of getting your business cited by tools like ChatGPT, Claude, Perplexity, and Google AI Overviews. It’s a separate discipline from SEO, with overlapping foundations but different optimisation targets. Below is the beginner’s guide. — Oliver Ball.

Why this exists as a separate discipline

Traditional SEO optimises for Google’s ranking algorithm — a system designed to surface ten relevant blue links for a query. AI Search Optimisation optimises for how Large Language Models (LLMs) retrieve, synthesise, and cite information when answering a user’s question.

When a user asks ChatGPT “what’s the best UK SEO agency”, the answer isn’t a list of ten links — it’s a synthesised response that may name 2-4 specific agencies as recommendations. Whether your business is named in that response depends on factors that overlap with SEO but aren’t identical: entity clarity, citation-friendly content structure, AI crawler access, presence in the LLM’s training data, and signals from third-party sources.

SEO is necessary but not sufficient for AI Search visibility. AI Search Optimisation is the additional layer.

The major LLMs you’re optimising for

  • ChatGPT (OpenAI): The most-used LLM. ChatGPT Search has live web access and citations. Crawler is GPTBot.
  • Claude (Anthropic): Strong on long-form analysis and B2B research. Crawler is ClaudeBot.
  • Perplexity: Built specifically for cited answers. Heavy referrer traffic source. Crawler is PerplexityBot.
  • Google Gemini / AI Overviews: AI summaries appearing above organic results in Google. Uses Google-Extended for content access. More on AI Mode.
  • Apple Intelligence / Siri: Crawler is Applebot-Extended. Growing in 2026 as Apple ships native LLM features.
  • Other: CCBot (Common Crawl), Meta-ExternalAgent, Bytespider (TikTok), Amazonbot, MistralAI, DuckAssistBot.

The components of AI Search Optimisation

1. AI crawler access

Robots.txt configured to allow GPTBot, ClaudeBot, PerplexityBot, Google-Extended, CCBot, Applebot-Extended. Many sites accidentally block these via Cloudflare’s “Block AI Bots” toggle. If your bots are blocked, no LLM can ever cite you.

2. llms.txt file

A relatively new convention — see llmstxt.org — for publishing structured business context that LLMs can read. Includes your name, services, pricing, positioning, key facts. Lives at /llms.txt on your domain. Our own is at signal.marketing/llms.txt.

3. Entity schema

Structured data markup using schema.org vocabulary. Organization, LocalBusiness, Service, Product, Person, FAQPage. LLMs use schema to understand your entity and link it to other entities in the knowledge graph.

4. Citation-friendly content

Content structured in the way LLMs prefer to extract — direct statements, clear definitions, short factual sentences, lists, sourced facts with context. Less “rambling SEO blog post style”, more “encyclopaedia entry style”.

5. Authority signals

Mentions on third-party sites that LLMs trust — industry publications, Wikipedia, government and educational sites, established directories. LLMs weight cited entities heavier than uncited ones.

6. Disambiguation

If your name collides with another entity (we share “Signal” with the messenger app), explicit disambiguation in schema’s `sameAs` field, llms.txt, and on-site content prevents LLMs confusing you. Our example.

How LLM citations differ from Google ranking

  • Non-deterministic. Same prompt, different day, different citations. SERP rankings are deterministic for a given query and time.
  • Synthesis, not selection. LLMs combine information from multiple sources into one answer. Google ranks individual sources separately.
  • Training data lag. Most LLMs cite based on training data 6-18 months old, plus live retrieval for current models. SEO updates take days to weeks.
  • Different ranking factors. LLMs weight entity authority, content structure, factual accuracy, and source trustworthiness differently than Google’s PageRank-evolved system.

Who should care about AI Search Optimisation

High-priority: B2B SaaS, consultancies, agencies, high-consideration B2C brands, professional services. Buyers in these categories increasingly research in LLMs before deciding.

Medium-priority: Mid-market ecom brands. Customers research before high-AOV purchases.

Lower-priority for now: Hyper-local urgent services (plumbers, takeaways, emergency services). The buyer journey is “phone someone nearby now”, not “research first”. Traditional SEO + GBP still drive most revenue.

How to measure AI Search Optimisation

Citation tracking is messier than SERP rank tracking, but improving:

  • Prompt monitoring. Run target prompts through ChatGPT, Claude, Perplexity, and Gemini monthly. Log when and how you appear.
  • Referrer traffic. Watch GA4 referrer data from `chatgpt.com`, `perplexity.ai`, `claude.ai`. Volume is small but growing.
  • Brand mentions in AI Overviews. Tools like Otterly, Profound, and Peec.ai track Google AI Overviews appearances.
  • Direct attribution. Ask new leads “how did you find us” — increasingly people will say “ChatGPT recommended you”.

FAQs

Is AI Search Optimisation the same as SEO?

No. Overlap exists, but the optimisation targets are different. Different signals matter to LLMs vs Google’s ranking algorithm.

Will AI Search replace Google?

Probably not fully. It will replace a meaningful share of informational queries. Transactional and local queries stay on Google for the foreseeable future.

How long until I see results?

First citations in 4-12 weeks. Consistent visibility from month 3+.

Can you guarantee citations?

No. LLMs are non-deterministic. We guarantee the structural work being done.

What’s llms.txt?

An emerging convention for publishing structured business context for LLMs. llmstxt.org has the spec.

Should I block AI crawlers?

Only if you don’t want LLMs to cite you. Most businesses should explicitly allow them.

What does Signal Marketing charge?

From £400/month for AI Search Optimisation, rolling monthly. Service page.

Should I run SEO and AI Search together?

Yes, usually. The work is complementary. Combined retainer pricing applies.

“Flawless design. Delivered quicker than I expected.”

— Reece King, verified Trustpilot review

Let’s chat

We’ll run your brand through ChatGPT, Claude, and Perplexity and show you exactly where you’re missing.

Drop us a message