Generative engine optimization (GEO) is the practice of shaping your pages so AI search tools such as Google AI Overviews, ChatGPT search and Perplexity cite them in their answers. It has nothing to do with geography. The term comes from Aggarwal et al., "GEO: Generative Engine Optimization" (opens in a new tab) (arXiv 2311.09735, 2023). This article answers what is generative engine optimization and ends with a checklist you can run today.
Start with access. Check robots.txt, your security plugin and your CDN against the crawler documentation before you rewrite a single post. Then you know whether AI crawlers can reach the page at all.
Key takeaways
- Run
curlagainst your robots.txt and your top post today, using the exact crawler names from the OpenAI, Perplexity and Google docs. - The paper’s "up to 40%" visibility gain was measured on its GEO-bench benchmark. It is a ceiling, not a forecast for your site.
- Rewrite key posts so each passage answers one question and still makes sense when quoted alone.
- Validate your structured data in Google’s Rich Results Test after every schema change.
What is generative engine optimization? Shaping pages so AI answers cite them
GEO means writing and configuring a page so that a generative engine picks it as one of its sources. A generative engine synthesizes an answer from several sources. The Aggarwal et al. abstract (opens in a new tab) describes these engines as tools that gather information from multiple sources and summarize it with large language models. It names website and content creators as the third stakeholder, with little control over when and how their content is shown.
The paper was submitted on 16 Nov 2023, last revised on 28 Jun 2024 (v3) and accepted to KDD 2024. A 2026 piece in The Successful Registrar, Talarico’s "What is generative engine optimization? The rise of zero-click culture" (opens in a new tab) [1], also treats GEO as a new set of three letters next to SEO. Neither source tells you which setting to change, so the sections below do.
Check how Google AI Overviews, ChatGPT search and Perplexity reach your pages
Read each vendor’s own documentation to see how it builds answers and displays source links. Each vendor decides how it selects and displays those links, and that changes over time. Read the vendor page before you act on any blog post, including this one.
| Platform | Official page to read | What to confirm there |
|---|---|---|
| Google AI Overviews | Google Search Central, "AI features and your website" | What a page needs to appear as a supporting link, and which controls exist |
| ChatGPT search | OpenAI crawler documentation | The exact user-agent names and how robots.txt rules apply to each |
| Perplexity | Perplexity crawler documentation | The exact user-agent name and how to allow it |
Crawler names and settings are the part that goes stale. Copy them from those pages, not from a tutorial, the way any developer would copy an API parameter from the docs. Then test your own server with the exact names.
curl -s https://yoursite.com/robots.txt
curl -I -A "USER_AGENT_FROM_THE_VENDOR_DOCS" https://yoursite.com/your-post/A 200 status means the server delivers the page to that user agent. A 403 usually means a firewall, security plugin or CDN rule is blocking it. Any of these layers can block a crawler, so check your security plugin and CDN rules as well as robots.txt. If you blocked AI crawlers on purpose, Block AI Crawlers in robots.txt and Stay in Google Search shows how to do it without losing Google results.
Read the "up to 40%" figure as a ceiling
The paper (opens in a new tab) reports that GEO can boost visibility by up to 40% in generative engine responses. That number comes from GEO-bench, the benchmark the authors built from diverse queries across multiple domains. The abstract also says the efficacy of these strategies varies across domains, which the authors take as a reason for domain-specific optimization.
Warning: Do not plan around 40% for your blog. It is the best result on one benchmark, and the same abstract says results differ by domain.
A 2026 study by Kazaryan on commercial real estate (opens in a new tab) [2] (in Russian) applies GEO to one industry. It treats brand mention, source attribution and the risk of errors as three separate metrics. Track them separately too. Being named in an answer and being linked are different outcomes.
What GEO changes in practice: passages, sources, entities, schema and access
Classic SEO work still applies. GEO adds five checks on top of it, and access is the one to settle first.
- Crawler access. Allow the crawlers named in the vendor docs in robots.txt, the security plugin and the CDN. Test with the
curlcall above after every plugin or firewall change. - Citable passages. Open each section with a short answer that works without the text around it. Cut references like "as mentioned above".
- Sources inside the text. Link the primary source next to the claim it supports, as this article does with the paper.
The last two checks help an engine match your text to a specific thing.
- Named entities. Use the full name of each tool, law or standard and spell out acronyms once.
- Structured data. Output Article and FAQPage schema from your SEO plugin or theme, then run each URL through Google’s Rich Results Test.
Example: Weak: "As mentioned above, there are a few ways to fix this." Citable: "A sitemap can return a 404 after a plugin or settings change. Check your permalink settings and the plugin’s documentation." The second version carries its own subject and its own fix.
TechnicalTip’s WordPress GEO Checklist covers these steps for WordPress sites. Page speed is a separate job, covered in How to Improve INP: A Field-to-Fix Checklist for Developers. If you’re considering an llms.txt file, read How to Add llms.txt to WordPress: 3 Methods and Who Reads It first.
Next step: open the three vendor pages in the table, copy the exact crawler names, run the two curl commands against your robots.txt and your best post, and fix any 403 before you edit content.
Frequently Asked Questions
How does generative engine optimization (GEO) work?
A generative engine synthesizes an answer from several sources and summarizes them with a language model. GEO aims to raise the visibility of your page in those responses. The paper measured that effect on its GEO-bench benchmark and found it varies by domain.
How is generative engine optimization done?
You allow the AI crawlers, rewrite key posts into self-contained passages, cite primary sources, name entities in full and add valid structured data. Access comes first, because content that cannot be fetched cannot be cited.
How do I start generative engine optimization on a WordPress blog today?
Check robots.txt, your security plugin and your CDN for blocks on the crawlers named in the vendor docs. Then pick your three best posts and rewrite each section opening as a standalone answer. Validate the schema in Google’s Rich Results Test.
How do I judge the best generative engine optimization agencies or companies?
Treat any "best" list of providers with caution. Ask whether a provider checks crawler access first, shows domain-specific evidence and cites primary sources such as the Aggarwal et al. paper. Be wary of anyone who promises the 40% figure.
References
- "What is generative engine optimization? The rise of zero‐click culture", Talarico, 2026, The Successful Registrar. doi:10.1002/tsr.70095 (opens in a new tab)
- "Применение generative engine optimization в маркетинге коммерческой недвижимости", Казарян, 2026, The Applied Economic Researches Journal. doi:10.47576/2949-1908.2026.2.2.021 (opens in a new tab)
