Generative engine optimisation (GEO) can sound abstract. In practice it is a set of concrete jobs across your website, your earned media, your social and video content, the reference sites that describe you, and your measurement.
This checklist is built on what the GEO Compass benchmark shows AI engines actually cite, across 900,000+ answers. If you are new to the topic, start with What is GEO?.
1. Make sure AI engines can read you
AI engines now run a live web search for almost every answer. If they can't fetch your pages, they can't cite them.
- Allow AI search crawlers in robots.txt, including OAI-SearchBot and GPTBot (OpenAI), PerplexityBot, ClaudeBot and Claude-SearchBot (Anthropic), and Bingbot (which feeds Microsoft Copilot). Check your CDN or firewall isn't blocking them either. See OpenAI's crawler documentation.
- Put your key facts in the HTML, not only in JavaScript, images or PDFs. Many AI crawlers don't run JavaScript.
- Publish a sitemap and keep
lastmoddates honest. Engines favour fresh content, and dates help them find it. - Add an
llms.txtfile that lists your most important pages in plain text. - Verify your site in Google Search Console and Bing Webmaster Tools. Bing's index feeds Microsoft Copilot, and Bing Webmaster Tools now reports how often your pages are cited in Copilot and Bing AI answers.
2. Make your own pages easy to quote
Brand and company websites supply 31% of the sources AI engines cite in the benchmark, more than any other type. In branded questions, your site is cited in 73% of answers. Make those pages count.
- Answer the question in the first sentence under a heading phrased the way customers ask it. Keep each answer section to roughly 100–150 words.
- Publish the facts engines need for each question type: specifications, prices, comparisons, eligibility, locations, policies. Product-detail questions are where brands' own sites are cited most (28% of unbranded answers).
- Write comparison and "best for" pages honestly, including when a competitor is the better fit. Comparison questions have the lowest own-site citation rate (8%), because few brands publish good comparisons.
- Use structured data: Organization, Product, FAQPage, LocalBusiness, Article and Dataset, as relevant. See the Google Search Central guidance.
- Keep entity names consistent. Use the same official name, description, address and leadership details everywhere. AI engines can treat "Brand", "Brand Group" and "Brand Asia" as different entities.
- Date and refresh key pages. AI engines show a strong recency bias. Update statistics and examples at least quarterly.
3. Win the sources that describe you
In category questions your own site is cited only 12% of the time. The rest of the answer comes from third parties.
- Find out which sites AI engines cite for your category, in each market. Filter the explorer by industry and market, or run your own prompt tracking.
- Prioritise those publications for PR, even if they aren't on your traditional tier-one list. The journalists and creators AI engines cite are often specialist, local or not media at all.
- Pitch the formats AI engines like to cite: question-led headlines, expert reviews, "best of" lists, Q&As and long-form explainers.
- Get onto the comparison and review sites that dominate your category, from software review platforms to automotive aggregators to travel sites.
- Correct inaccuracies at the source. If an answer is wrong, find the cited page and fix it there.
4. Treat video and community as sources
- Publish on YouTube if Google's AI surfaces matter to you. YouTube appears in 38% of Google AI Overviews answers with sources, against 4% of ChatGPT's. See how each engine cites.
- Structure videos for extraction: chapters, timestamps, full transcripts, detailed descriptions and spoken phrasing that matches real questions.
- Brief creators the same way. Well-structured creator content is cited long after a typical social post has faded.
- Show up accurately where communities talk about you, such as Reddit, Quora and enthusiast forums. Reddit is cited in about 12% of ChatGPT answers with sources, and community platforms matter more in markets such as India.
- Use LinkedIn for executive and company content. It appears in 16% of Perplexity answers and 14% of Google AI Mode answers.
5. Keep reference sources accurate
- Maintain accurate Wikipedia and Wikidata entries where your organisation qualifies, following the platforms' conflict-of-interest rules. In some markets, such as Malaysia, Wikipedia is cited in a quarter of answers.
- Check business directories, maps listings and industry associations. Institutions and associations supply 8% of citations overall, and more in sectors such as energy and education.
6. Measure what AI says
- Track a fixed set of real questions over time, across branded and unbranded intents, the engines your audience uses, and each market and language. Measure share of model (how often you are named), answer position, sentiment, accuracy and the sources cited. See how to measure AI visibility.
- Re-check after model updates. Within the same studies, a single model change in August 2026 cut earned media's share of one engine's citations from 42% to 30%. Visibility is tracked by engine and model version, not once a year.
Where to start
If you only do three things this quarter: unblock and verify (steps 1 and 5), fix the facts on your top 20 pages (steps 6–10), and find the ten third-party sites AI engines cite most for your category in each market (step 12). Then measure.
Frequently asked questions
- How do I get my brand mentioned by ChatGPT?
- Make sure ChatGPT's search crawler can read your site, publish clear factual pages that answer the questions customers ask, and earn coverage in the publications, review sites and communities ChatGPT cites for your category. In the GEO Compass benchmark, earned media supplies about a quarter of ChatGPT's citations.
- How long does GEO take to work?
- Website fixes can be picked up within weeks, because AI engines now run a live web search for almost every answer. Earned media, reference and community work compounds over months. Track a fixed set of questions monthly to see progress.
- Do I need an llms.txt file?
- It is a low-cost addition. An llms.txt file gives AI agents a plain-text map of your key pages. It complements, but does not replace, a crawlable site, a sitemap and structured data.
Sources
- Overview of OpenAI crawlers, OpenAI
- Does Anthropic crawl data from the web?, Anthropic
- Perplexity crawlers, Perplexity
- Introducing AI Performance in Bing Webmaster Tools, Microsoft Bing
- AI features and your website, Google Search Central
- The /llms.txt file, llmstxt.org
- Schema.org vocabulary, schema.org
Figures marked as GEO Compass data come from the anonymised GEO Compass benchmark of AI answers collected by We AI & Data Advisors. Related: AI search in Australia · AI search in Singapore · AI search in Malaysia · AI search in Indonesia · AI search in India · AI search in Europe