Guides · AI search

Generative Engine Optimization (GEO): A Practical Guide

By Tom Johnston, founder · Updated 2026-11-09 · 5 min read

Short answer: Generative engine optimization (GEO) is the work of making your business easy for AI assistants such as ChatGPT, Perplexity, Gemini and Google's AI features to find, understand and cite. It rests on solid SEO, then adds answer-first pages, a clear and consistent business identity, mentions on other trusted sites, and open access for AI crawlers.

You will see the same idea called AI SEO, answer engine optimization (AEO), LLM SEO or GEO SEO. "AI seo" alone is searched about 6,600 times a month in the US (Google search data via DataForSEO, US, November 2026). The labels differ; the work is mostly the same. Here is what it involves, in order of importance.

What is generative engine optimization?

A generative engine is any tool that writes an answer instead of just listing links: ChatGPT, Perplexity, Google's Gemini, Microsoft Copilot, and Google's own AI Overviews and AI Mode. GEO is the practice of being one of the sources those tools draw on, and one of the businesses they name.

The term came out of academic research in 2023 and spread through the marketing world quickly. Our glossary entry on GEO gives the one-line version.

How do AI assistants choose which sources to cite?

None of the major providers publishes a full recipe, so treat any confident formula with suspicion. What is reasonably well understood:

• Most use live web search for current questions. ChatGPT search, Perplexity, Gemini and Google's AI features look things up on the web, then summarize. If you are not findable in search, you are unlikely to be cited.
• They lean on pages that answer clearly. A direct, quotable passage is easier to use than a vague sales page.
• They combine many sources. Answers often blend several pages, directories and review sites. Being consistent across all of them matters.
• Training data plays a part too. What a model "knows" about your brand without searching comes from what was written about you across the web before its training cut-off.

The practical result: GEO is mostly good SEO, good writing and a consistent reputation, aimed at being quoted rather than only being clicked.

The core GEO practices

1. Write answer-first pages

Open every page with a two-to-three sentence answer to the question the page targets. Use question-shaped headings and answer each one in its first sentence. End with a short summary of the key facts. We go deeper in writing for AI search.

2. Make your entity clear

An "entity" is just the thing being described: your business. AI systems need to be confident who you are. Make it easy:

• Use the exact same business name, address and phone number everywhere.
• State plainly what you do, where, and for whom, on your home page and about page.
• Add accurate structured data (Organization or LocalBusiness) with your name, logo, contact details and profiles. Our guide to schema markup for local businesses shows how.
• Keep your Google Business Profile, Bing Places listing and major directories complete and matching.

3. Earn citations and mentions

AI answers often draw on third-party pages: local news, trade associations, supplier sites, review platforms and comparison articles. A business that is only described on its own website is harder for a model to trust. Real mentions on real sites help with both SEO and GEO.

4. Put facts in plain text

Prices, service areas, warranties, license numbers, hours and specs should be in HTML text. Content locked in images, PDFs or scripts that need clicking to load is harder for crawlers to read.

5. Keep crawlers allowed

Check your robots.txt. Unless you have a specific reason, do not block the crawlers AI search tools use. Common ones include:

Blocking a training crawler and blocking a search crawler are different decisions. If you want to be found in an assistant's live answers, its search crawler must be able to reach you. Also note that Google-Extended controls use of your content for Gemini training, not your appearance in Google Search.

6. Keep content current

Dated prices, old staff names and services you no longer offer get repeated back to customers. Review your main pages at least twice a year and show a last-updated date where it is honest.

A worked example

Say you run an HVAC company in Denver. A GEO-ready furnace repair page would open with a direct answer ("Most furnace repairs in the Denver area are done in one visit; common fixes include igniters, flame sensors and blower motors"), then cover warning signs, what drives the price, response times, the suburbs you serve and your license details. Your Business Profile, Yelp, BBB and Bing listings would carry the same name, phone and service list. A local news mention or a supplier's dealer page would confirm you independently. None of that is a trick. It simply makes you the easiest business for an assistant to describe accurately.

What about llms.txt?

llms.txt is a proposed plain-text file, placed at your site root, that lists your most important pages in a format meant for language models. It is easy to add and does no harm.

But keep expectations low. It is a proposal, not an adopted standard, and major AI search providers have not said they use it to choose sources. Google has said it does not use it for Search. Add it if you like; do not count it as GEO work done.

How do you measure AI visibility?

There is no Search Console for ChatGPT. A simple, repeatable method works better than any single score:

1. Pick 20 to 30 real questions customers ask, with and without your city. Example: "Who is a good emergency plumber in Denver?" and "How much does it cost to replace a water heater?"
2. Run them in each assistant you care about, ideally from a logged-out or clean session.
3. Record three things: whether you are named, whether your site is linked, and which competitors or directories are cited instead.
4. Repeat monthly with the same questions. Answers vary run to run, so look at the trend, not one result.
5. Check your analytics for referral visits from AI tools, and track calls and forms by source.

The "who gets cited instead" column is the most useful. It tells you which directories, review sites and publications the assistants trust in your market, and where you should be listed.

Common GEO mistakes

• Chasing tricks. Hidden text, prompt-like instructions in pages and keyword-stuffed FAQs are not a strategy and can backfire.
• Thin "AI pages". Dozens of near-identical question pages add little. One complete page usually beats ten shallow ones.
• Inventing facts. Made-up stats or reviews can be repeated by AI tools and damage trust when customers check.
• Ignoring local signals. For a service business, reviews and consistent listings often matter as much as the website. See our guide to AI search for local businesses.

If you want this done for you, our AI search optimization service covers audits, page rewrites, entity cleanup and monthly visibility tracking.

Questions

Terms in this guide: AI Overviews · AI Mode · robots.txt · llms.txt

Get a free proposal