- AI search still relies on crawlable, trustworthy pages, so fix technical basics first and make your business facts consistent everywhere.
- Publish original information, such as real pricing ranges, processes and photos, because pages that only repeat other sources give AI no reason to cite you.
- Check robots.txt and firewall rules for AI crawlers, and treat llms.txt as optional because no assistant has promised to use it.
- Measure with GA4 referrals from AI assistants, a monthly prompt log and a 'how did you find us' form question, and expect small numbers at first.
An AI SEO strategy is your plan for being found, understood and cited when people get answers from AI-powered search, not only when they click a list of blue links. It builds on classic SEO, because the systems behind AI answers still lean on pages that are crawlable, trustworthy and clear. What it adds are new priorities: being unmistakably identifiable as a business, publishing information other sources do not have, and tracking mentions and referrals that do not look like normal rankings.
This guide explains how AI SEO works in plain terms, how AI is transforming SEO in practice, and how to build a strategy step by step. It covers AI Overviews and AI Mode, assistants that cite sources, technical access for AI crawlers, measuring AI referrals in GA4, and using AI tools in your own content without publishing junk. Where details are not publicly known, we say so instead of guessing.
How AI SEO works, in plain terms
Most AI search experiences combine two parts. A language model writes the response, and a retrieval step finds web pages for the model to draw from. When you ask a complex question, the answer often appears as a short written summary with links or citations to the pages behind it. These systems also appear to break a complicated question into several smaller searches and then combine what they find.
That has a practical meaning. To be cited, a page first has to be retrievable: crawlable, indexed, relevant to the question and available to the tool. Then it has to be worth quoting: specific, clear and trustworthy. Nobody outside these companies knows the exact selection rules. Anyone selling a precise formula, a list of "AI ranking factors" or a magic setting is guessing. The observable principles are enough to act on, and that is what the rest of this guide focuses on.
How AI is transforming SEO, and what stays the same
The fundamentals have not been replaced. What has changed is how much weight falls on clarity, original information and presence beyond your own website. The table compares the two.
| Area | What stays the same | What changes in practice |
|---|---|---|
| Foundation | Crawlable, indexable, fast, secure pages | More tools fetch your pages, so access rules matter beyond one search engine |
| Content | Helpful, accurate and written for people | Direct, quotable answers and original details stand out when answers are summarized |
| Authority | Reputation, links, reviews and mentions | Third-party mentions and consistent facts help systems confirm who you are |
| Intent | Match the page to what the searcher wants | Questions get longer and more conversational, with follow-ups |
| Measurement | Search Console, analytics and conversions | Visibility includes being named in answers, which needs prompt testing and referral tracking |
| Clicks | Higher placement usually earns more visits | Simple questions may be answered without a click, so impressions and clicks can drift apart |
The biggest shift is at the top of the funnel for simple informational questions. Commercial and local searches still end with someone choosing a business, so brand recognition, reviews and clear service pages matter more, not less.
AI Overviews and AI Mode: what site owners can do
AI Overviews are written summaries that Google shows above the usual results for some searches, with links to supporting pages. AI Mode is a more conversational way to search in Google, built for complex or multi-part questions and follow-ups, and it also links out to sources. Both point people to ordinary web pages, which in practice means a page needs to be indexed and able to appear in regular results before it can be cited.
Sensible steps look a lot like good editing:
- Put a direct answer in the first lines under a heading that mirrors a natural question.
- Follow with the detail: steps, conditions, costs and exceptions.
- Cover likely follow-up questions on the same page, often in a short FAQ.
- Use lists and tables where they make comparisons or processes easier to scan.
- Keep facts current, and remove outdated pages instead of leaving them live.
No special markup is known to be required for these features. Structured data helps machines understand a page, but treat it as clarity, not a switch you flip.
Assistants that cite sources: ChatGPT, Perplexity and Gemini
ChatGPT with search, Perplexity and Gemini can retrieve current web pages and show citations. When you test them, the cited sources are often a mix: business websites, industry publications, directories, review platforms and forums. The implication for a business is that your own site is only one of several places feeding an answer, so reviews, profiles and mentions elsewhere count.
Answers also vary. The same question can produce different results depending on wording, location, the assistant and the day. Do not treat a single answer as a ranking. Treat repeated testing over weeks as evidence. For businesses that serve a local area, our guide to GEO for local businesses shows how these assistants appear to choose who to recommend.
Building an AI SEO strategy step by step
Here is a practical sequence you can start this month. Each step is deliberately basic, because the basics are where most sites fall short.
- Fix the foundation. Make sure key pages are crawlable, indexed, fast and mobile-friendly, and clean up duplicates and broken pages. Our technical SEO checklist covers the details.
- Make your entity unmistakable. Align your name, services, location and description across your website and profiles.
- List the questions customers ask. Pull them from calls, your sales team, Search Console and People Also Ask, and test them in the assistants yourself.
- Build or upgrade the pages that answer them. Add original information that other pages lack.
- Strengthen corroboration off your site. Reviews, directory listings, partner pages and industry mentions all confirm your claims.
- Allow appropriate access. Review robots.txt, firewall and bot-protection rules, and keep important text in plain HTML.
- Set up measurement. Start a prompt log and an AI channel in GA4.
- Review monthly. Update stale pages and repeat what gets cited.
Entity clarity and original information: what AI systems need from your site
Entity clarity
An entity is a clearly identifiable thing: your business, a person, a product or a place. Search engines and AI assistants both need to be confident that the business on your website is the same one on your profiles and reviews. Use the same business name, phone number, address, hours and service names everywhere: your website, Google Business Profile, Bing Places, Apple Business Connect, Yelp and industry directories.
Build an About page that says who runs the business, where it operates, what it is qualified to do and who it serves. Give expert articles real author pages. Add Organization or LocalBusiness structured data with links to your official profiles, as covered in our schema markup guide. Avoid vague labels like "solutions" and say plainly what you do.
Original information
If your page says what ten other pages already say, an AI summary has no reason to cite you, because it can state the same thing without you. Content that deserves a citation includes things only you know: your typical price ranges and what drives them, how your process works step by step, realistic timelines, photos of real work, local knowledge such as permits or climate quirks, and honest comparisons that admit trade-offs.
If you publish your own numbers, make sure they are real and clearly yours. Never invent data. Then format the page for extraction: lead with the answer, keep paragraphs short, use descriptive headings, and put steps and comparisons in lists and tables.
Technical access for AI crawlers: robots.txt and llms.txt
Different bots do different jobs. Search crawlers build the indexes behind Google and Bing. AI companies also run their own crawlers, some to collect training data, some to build a search index, and some to fetch a page when a user asks. Examples include OpenAI's GPTBot, OAI-SearchBot and ChatGPT-User, and Perplexity's PerplexityBot. Google also offers a separate robots.txt token called Google-Extended to control some uses of your content in its generative AI products. Names and behaviors change, so check each company's current documentation.
Blocking a crawler can reduce your chances of appearing in that product, while allowing it means your content may be used under that company's policies. For most businesses that want to be recommended, allowing search-related crawlers is sensible, while training permission is a business decision. A common hidden problem is that firewall, CDN or bot-protection settings block AI crawlers without anyone noticing, so check your server logs and security rules.
Also make sure the content itself is reachable. Keep key information in plain HTML text, not only in images, PDFs, pop-ups or pages behind logins.
You may also see llms.txt mentioned. It is an emerging convention: a plain text or Markdown file at the root of your site that lists and summarizes your most important pages for language models. No major assistant has publicly committed to relying on it, so treat it as optional housekeeping that is cheap to add, not a strategy and not a promise of citations.
Measuring AI referrals in GA4
You cannot measure AI visibility with a single report. Use three layers together.
Referral traffic in GA4
In GA4, open the traffic acquisition report, switch the dimension to session source and medium, and look for sources such as chatgpt.com, perplexity.ai and gemini.google.com. Some assistants add their own source tag to links they send, and others arrive as ordinary referrals or even as direct traffic, so check what your own reports reveal instead of assuming. Create a custom channel group called "AI assistants" that matches those sources, so they roll up in one line.
For links you control, such as your email signature, social profiles and shared resources, use the UTM builder so AI-influenced journeys are not confused with other sources. Add a "how did you find us?" field to forms, with an option for an AI assistant. If you want help with the setup, our analytics tracking service can build it.
Prompt testing and indirect signals
Keep a simple log. Every month, run the same ten to twenty questions your customers would ask across the main assistants, and record whether you were named, which pages or sources were cited and who else appeared. Repeat each prompt a few times, because answers vary. Search Console will not give you a clean AI-only filter as far as we can tell, so also watch for question-style queries where impressions rise while clicks stay flat, and watch branded searches and direct visits, which often rise when people discover you in an assistant first.
Expect small numbers at first. What matters is whether those visitors call, book or buy.
Using AI tools responsibly in content production
AI tools are useful for summarizing call transcripts, outlining pages, drafting FAQs, tidying prose and spotting gaps. They are poor at knowing your business. They invent facts, produce generic text, get local details wrong and repeat what is already online, which is the opposite of original information.
A responsible workflow keeps people in charge:
- A person with real expertise chooses the angle and supplies the original material: experience, data, photos and examples.
- The tool helps with structure and first drafts.
- An expert reviews and edits every page, and checks every number, name, law and price.
- The page carries a real author and a date, and gets updated when facts change.
Google's spam policies are aimed at content produced at scale mainly to manipulate rankings, whoever or whatever wrote it. The safe path is the same either way: publish fewer pages that are genuinely useful, accurate and reviewed, and never fabricate reviews, figures or quotes.
Where to start this month
Pick three actions. First, audit your business facts across your website and profiles and fix every mismatch. Second, rewrite your five most important pages so each opens with a direct answer and adds something original. Third, start a prompt log and an AI channel in GA4, so in ninety days you can see whether anything moved.
If you want a team to do this for you, Growth Asteroid's AI search optimization service starts from $2,900 per month. If you serve a local market, read the Google Business Profile guide as well, because those facts feed many AI recommendations. Strong fundamentals, clear facts and original information are the dependable part of an AI SEO strategy, and no one can promise more than that.
Frequently asked questions
What is an AI SEO strategy?
An AI SEO strategy is a plan to be found, understood and cited in AI-powered search, including AI Overviews, AI Mode and assistants such as ChatGPT, Perplexity and Gemini. It builds on classic SEO and adds clear business facts across the web, original information worth citing, appropriate crawler access, and new measurement such as prompt testing and AI referral tracking.
Is AI search replacing traditional SEO?
No. The systems behind AI answers still rely on crawlable, indexed, trustworthy pages, so technical health, helpful content and reputation remain essential. What changes is the weight on clarity, original information and mentions beyond your own site. Simple questions may be answered without a click, while commercial and local searches still end with someone choosing a business.
How can I get my business cited by ChatGPT, Gemini or Perplexity?
Nobody can promise a citation. What helps is being easy to retrieve and worth quoting: crawlable pages, consistent business details across your website and profiles, reviews, third-party mentions, and pages that answer real questions with original information. Test the questions your customers ask every month, record who gets named, and improve the pages that are missing.
Should I block AI crawlers or create an llms.txt file?
That depends on your goals. Blocking a crawler can reduce your chances of appearing in that product, so businesses that want recommendations usually allow search-related crawlers and decide separately about training use. An llms.txt file is an emerging convention that is cheap to add, but no major assistant has promised to use it, so treat it as optional.
How do I track traffic from ChatGPT and other AI tools in GA4?
Open the traffic acquisition report, view session source and medium, and look for sources like chatgpt.com, perplexity.ai and gemini.google.com. Create a custom channel group so they roll up together. Some visits may arrive as direct traffic, so also add a 'how did you find us' form question and keep a monthly log of prompt tests.
