What is AI SEO? A practical guide to search in the age of AI answers

AI SEO is search engine optimisation adapted for a world where people get answers from AI, not only from a list of links. It has two halves. The first is using AI tools to research, produce and measure SEO work faster. The second, and more important, is optimising your brand so that AI systems such as Google AI Overviews, ChatGPT, Perplexity and Claude cite and recommend you when they answer your buyers' questions.

This guide explains what changed, how AI SEO differs from the SEO you already know, and the strategy we use with clients to grow both rankings and AI visibility.

Why search changed

For two decades the unit of search was the ranked result. You optimised a page, it ranked, people clicked. Three shifts broke that model at once.

• Google now answers first. AI Overviews and AI Mode compose an answer above the classic results for a large share of queries, citing a few sources.
• Buyers ask assistants. A growing part of product research happens inside ChatGPT, Perplexity, Gemini and Copilot, where the response is a shortlist of three or four brands rather than ten blue links.
• Production got cheaper. AI tooling lets anyone publish at volume, which raised the bar on what earns a ranking or a citation.

The consequence is that visibility is now split between two surfaces. You can rank well and still be invisible where the decision is made.

AI SEO versus traditional SEO

Dimension	Traditional SEO	AI SEO
Goal	Rank pages for keywords	Rank pages and be cited or recommended in AI answers
Unit of research	Keywords	Keywords plus buyer prompts
Content	Pages that satisfy a query	Pages structured so a passage can be lifted as an answer, with evidence and attribution
Authority	Backlinks	Backlinks plus entity signals and third-party mentions that models learn from
Measurement	Rankings, traffic	Rankings, citations, share of model, revenue

Everything in the left column still applies. AI engines retrieve from the same web index that Google ranks, so a technically sound, authoritative site is the foundation. AI SEO adds a layer on top rather than replacing the base.

The three surfaces you are optimising for

1. Classic search results

Organic listings still send the majority of search traffic for most sites. Technical health, intent-matched content and links remain the levers. Our technical SEO service starts here because nothing else works if pages cannot be crawled and indexed.

2. Google AI Overviews and AI Mode

These generative answers cite pages that state a clear answer, back it with evidence and carry strong entity signals. Optimising for them is called generative engine optimization, or GEO.

3. Conversational assistants

ChatGPT, Perplexity, Claude, Gemini and Copilot answer from a mix of training data and live retrieval. Whether they name your brand depends on how consistently the web describes you as a credible option. This is answer engine optimization, or AEO.

An AI SEO strategy in six steps

Step 1: Benchmark across every engine

Before changing anything, record where you stand. Track rankings for your commercial keywords as usual, then run a fixed set of buyer prompts through each assistant and record which brands and sources appear. Our free AI Visibility Checker does this for up to five prompts.

Step 2: Map demand as prompts, not only keywords

People phrase questions to assistants differently from search boxes. "Best CRM for a ten-person law firm" is a prompt; "law firm CRM" is a keyword. Build your demand model from both, using sales calls, support tickets and search data.

Step 3: Fix the foundation

Crawlability, indexing, rendering and Core Web Vitals. Allow the AI crawlers you want in robots.txt, publish an llms.txt file, and make sure primary content is in the initial HTML rather than injected by JavaScript.

Step 4: Build citable content

For each priority query, publish or rewrite a page so that the answer appears in the first paragraph, is supported by data or named sources, and is structured with headings that mirror the question. Add Organization, Article, FAQ and Person schema so machines can attribute the content. See our SEO content writing service for how we brief these pages.

Step 5: Earn authority the models trust

Language models learn who is credible from the sources they train on and retrieve from: industry publications, review platforms, directories, communities and news. A digital PR and link building programme now serves both rankings and AI recommendations.

Step 6: Measure share of model and revenue

Report rankings, but also report the percentage of AI answers that mention you. We call this share of model. Pair it with organic pipeline and revenue so leadership sees cause and effect.

Using AI to do SEO faster

The other half of AI SEO is tooling. Used well, AI assists rather than replaces expertise.

• Keyword and prompt clustering at a scale no analyst can match by hand.
• Brief generation from SERP and citation analysis, then edited by a strategist.
• First drafts that senior writers rework and subject-matter experts review.
• Technical QA that flags regressions across thousands of URLs.
• Reporting that turns raw ranking and citation data into plain-language insight.

The failure mode is publishing unedited AI output. Google's helpful content systems and the assistants themselves reward accuracy and expertise, and both are quick to ignore thin, generic pages.

Common mistakes

1. Treating AI visibility as a separate project. It is the same content, authority and technical work with different measurement.
2. Blocking AI crawlers by default. If GPTBot, ClaudeBot and PerplexityBot cannot read your site, assistants rely on what others say about you.
3. Chasing volume. Ten excellent, citable pages beat a hundred average ones.
4. Reporting only rankings. A stable ranking can hide a collapsing click-through rate under an AI Overview.

Where to start this week

Pick ten commercial queries. Search each one and note whether an AI Overview appears and who it cites. Run the same ten as prompts in ChatGPT and Perplexity. The gaps you find are your first roadmap. If you would like us to do this for you, request a free AI SEO audit and a senior strategist will send a written report within 48 hours.

Frequently asked questions

Is AI SEO different from traditional SEO?

It extends it. Traditional SEO optimises for ranked lists of links. AI SEO also optimises for AI-generated answers that cite a handful of sources, and it uses AI tooling to research, produce and measure at greater speed.

Does AI-written content rank in Google?

Google evaluates content on helpfulness and expertise, not on how it was produced. AI-assisted content that is accurate, original and reviewed by an expert can rank; thin, unedited AI output usually does not.

How long does AI SEO take to work?

Technical and citation fixes often show within six to twelve weeks. Compounding organic revenue growth typically becomes visible from month four onwards.

Do I need an AI SEO agency?

Not necessarily. Teams with in-house SEO capacity can adopt the framework in this guide. An agency helps when you need senior strategy, production velocity and AI visibility tracking without hiring for all three.