AI Search in 2026: How UK Businesses Should Optimise for ChatGPT, Perplexity, and Google AI Overviews

The Search Landscape Has Changed

In 2023, 0% of Google searches produced an AI-generated overview. In 2026, AI Overviews appear for an estimated 30–40% of informational queries in the UK — and that number is growing.

Meanwhile, a significant portion of B2B buyers now start their vendor research on ChatGPT or Perplexity before touching Google. They ask things like "What's the best AI agency for a UK SME?" and read the AI's answer directly, often without clicking through to a website.

This has created a new category of SEO problem: your content may be ranking in Google but not being cited by AI systems, leaving you invisible in the fastest-growing discovery channel.

This guide covers what's actually working in 2026 for UK businesses wanting to be found via AI-powered search.

How AI Search Systems Decide What to Cite

To optimise for AI search, you need to understand how these systems work at a basic level:

Google AI Overviews are built on Google's search index plus their Gemini models. They pull content from pages that rank in the top results for a query, then summarise or synthesise that content. If you rank in the top 5–10 for a query, you have a reasonable chance of being cited in the AI Overview.

ChatGPT (browsing mode) uses Bing's index when browsing is enabled. For non-browsing responses, it uses training data up to its knowledge cutoff. To be cited in ChatGPT answers, you need: Bing indexation, authority signals on Bing, and for high-stakes queries, brand mentions in sources that were in ChatGPT's training data.

Perplexity uses a combination of web search (Google and Bing results) plus its own crawl. It's particularly citation-heavy — it shows sources for almost every claim. Being cited in Perplexity means being highly relevant for the query, having a clean robots.txt that allows their crawler (PerplexityBot), and having clear, factual, well-structured content.

Claude (Anthropic AI, when browsing) similar to ChatGPT browsing mode but generally uses Brave Search results. Optimising for Brave means similar signals to Google: authority, content quality, structured data.

GEO: Generative Engine Optimisation

GEO is the emerging discipline of optimising content to be cited in AI-generated answers. It builds on traditional SEO but with some specific additions:

1. Clear, factual, citable statements

AI systems are designed to cite specific claims. Content that makes clear, specific, factual assertions (not vague brand claims) is more likely to be cited.

Good: "Google's AI Overviews appear for an estimated 30–40% of informational queries in the UK as of 2026." Bad: "AI is transforming the search landscape in exciting and unpredictable ways."

The more specific and citable your content, the better.

2. Question-and-answer structure

AI systems are responding to questions. Content structured as direct answers to specific questions performs better in AI-generated responses.

This is why FAQs, how-to guides, and "what is X" explainers punch above their weight in AI search. They match the query structure of how people use AI search tools.

3. E-E-A-T signals

Google's E-E-A-T framework (Experience, Expertise, Authoritativeness, Trustworthiness) is now baked into both traditional ranking and AI citation decisions. For UK businesses:

• -Experience: Show real work, real clients, real results (our case studies on WeDispose, DynaHomes, etc.)
• -Expertise: Demonstrate deep knowledge in your content (this article is an example)
• -Authoritativeness: Get cited by other credible sources (PR mentions, industry publications, directory listings)
• -Trustworthiness: Legal pages, transparent pricing, secure site, genuine contact information

4. llms.txt

A relatively new standard (proposed 2024, adopted by some AI crawlers). A llms.txt file at your domain root tells AI systems what your site is about and how to use it.

If you want AI crawlers to use your content, create a /llms.txt file that summarises your business, key facts, services, and links to your most important pages.

5. Clean crawler access

Most AI crawlers identify themselves via their User-Agent:

• -Google: Googlebot, Google-Extended
• -OpenAI: GPTBot
• -Anthropic: ClaudeBot
• -Perplexity: PerplexityBot
• -Meta: facebookexternalhit

Your robots.txt should allow (not block) these crawlers. If you're blocking them inadvertently with a wildcard disallow, you're invisible to AI search.

What to Create for AI Search in 2026

Detailed FAQ content — not the 5-question FAQ on your contact page, but genuinely comprehensive question-and-answer content on topics you want to be known for. "How much does a website cost in the UK in 2026?" "What's the difference between a chatbot and an AI calling agent?" Genuine answers, 200–500 words each.

Comparison and alternative content — "AI chatbot vs live chat: which is better for UK SMEs?" "Scale IT UK vs typical agency" type pages. These match high-intent AI search queries.

Structured data (especially FAQPage and HowTo... wait, HowTo is deprecated) — FAQPage schema remains valuable for AI Overviews even though it no longer generates FAQ rich results in standard Google Search. AI systems use structured data to understand content relationships.

Local authority content — For UK location-based queries ("AI agency in Leeds"), being the most authoritative, well-cited source in your local market gives you a significant AI citation advantage. This means local blog content, local case studies, GBP optimisation, and local directory citations.

Specific numbers and data — AI systems prefer content with specific, verifiable numbers over vague claims. "Response time dropped from 14 hours to under 2 minutes" is more citable than "dramatically improved response time."

The Technical Foundation

For AI search, the technical requirements overlap heavily with traditional SEO, with a few additions:

• -Fast load time — AI crawlers have crawl budgets; slow sites get crawled less
• -Clean HTML structure — AI systems parse your HTML to understand content hierarchy
• -Semantic markup — Use heading tags correctly (one H1, logical H2/H3 hierarchy)
• -Schema.org markup — Structured data helps AI systems understand what your content is about
• -Stable URLs — AI training datasets freeze at points in time; stable, permanent URLs are more likely to be cited
• -Correct canonical tags — Prevent content duplication that confuses AI crawlers

Measuring AI Search Performance

This is the hardest part — there's no "AI Overview Console" yet. Proxies:

1. -GSC impressions for featured snippet positions — AI Overviews often use content that previously appeared in featured snippets
2. -Direct traffic from brand queries — If people are finding you via AI assistants and then Googling you, branded search volume increases
3. -Perplexity.ai manual testing — Search for your target queries on Perplexity and see if you're cited
4. -ChatGPT manual testing — Ask ChatGPT (with browsing enabled) about your target queries and see if you appear

Tools like SE Ranking, BrightEdge, and Semrush have started adding AI visibility tracking, though coverage is still limited in the UK market.

The UK Opportunity

For UK B2B service businesses, AI search is still relatively uncrowded. Most UK SMEs and many agencies haven't started optimising for it. The businesses that establish themselves as cited authorities in AI search now will have a significant head start when it becomes the primary discovery channel.

Get a free SEO and AI visibility audit — we'll check your current AI crawler access, analyse your content structure for AI citability, and create a prioritised plan for improving your AI search visibility.

Related: Local SEO for UK Businesses | Full-Funnel Digital Marketing