Loading

WebsiteBy Bait · · 11 min read

SEO and GEO for large companies: how to show up in search and AI answers

GEO (generative engine optimization) is the work of becoming a source in AI answers. See how SEO and GEO add up on a large company's website.

Violet lighthouse-shaped structure emitting beams through transparent panels.

SEO and GEO (generative engine optimization) for large companies are two sides of the same job: making sure the organization's information is found, understood and used as a source. SEO (search engine optimization) is the practice of making pages crawlable, indexable and relevant to search results. GEO is the name given to the effort of appearing as a source in answers generated by artificial intelligence, such as Google's AI summaries and conversational assistants that search the web. People also say AI SEO or GEO SEO for the same concern.

In large companies, the issue carries its own weight. The website brings together thousands of pages, several brands, countries and languages, investor relations, careers and press sections, with content produced by different teams. When someone asks a search engine or an assistant what the company does or how it compares with another, the answer is assembled from what is published and accessible. Confusing or outdated information becomes a poor answer, or an answer without the company.

What SEO is and what GEO is

SEO deals with three practical questions. Can the search engine access the page? Can it understand what the page is about and index it? Does the page answer the searcher's question better than the alternatives? The work includes information architecture, technical performance, content, internal links and trust signals.

GEO is a newer term. It appears in the title of a 2023 academic paper, "GEO: Generative Engine Optimization", which proposed measuring and improving the visibility of content in the answers of generative engines. Since then, the market has used the acronym for any initiative aimed at AI answers. No search engine has published a GEO specification, and that matters: much of what is sold as a GEO technique is hypothesis, not documented rule.

Does GEO replace SEO?

No. In Google's case, the official guidance on AI features in Search states that fundamental SEO practices still apply to AI Overviews and AI Mode, and that there are no additional requirements or special optimizations needed to appear in these experiences. To be shown as a supporting link, a page must be indexed and eligible to appear in Search with a snippet (the text excerpt shown in the result). Meeting these requirements does not guarantee being shown.

In practice, GEO broadens the SEO question. Beyond a page's position for a keyword, the team starts to watch whether the company is described correctly and cited when the topic is its own. The technical and editorial foundation is the same; what changes is how the result is observed.

How AI answers choose and cite sources

No system publishes its selection formula, and there is no secret factor to uncover. What is documented is enough to guide the work.

  • AI Overviews and AI Mode, according to Google, may use query fan-out: the system runs several related searches, across subtopics and sources, to build the answer and show a wider set of links. The two features may use different models, which is why the links vary.
  • Google states that there is no need to create text files for AI, new markup or specific structured data to appear in these features.
  • Control remains the same as in Search: robots.txt for crawler access, and the nosnippet, data-nosnippet, max-snippet and noindex directives to limit what is shown.
  • Assistants with search have their own crawlers. OpenAI documents OAI-SearchBot, used to show websites in ChatGPT's search features, separate from GPTBot, used to collect training content. Blocking one does not block the other.

The consequence is direct. If robots.txt, bot protection or a firewall blocks search crawlers without anyone having decided so, the website is left out of answers for a technical reason. Review these rules with the security and technology teams, deciding what to allow for search and what to restrict for training.

Crawling and indexing at scale

Large websites face a problem that small ones hardly have: the volume of URLs. Google maintains a crawl budget guide (the set of URLs Google can and wants to crawl) aimed at sites with more than one million pages that change weekly or more than ten thousand that change daily, figures the guide itself treats as estimates.

  • Consolidate duplicate content: filters, sort orders and campaign parameters generate variations of the same page.
  • Return 404 or 410 for removed pages and eliminate soft 404s, empty pages that respond as if they existed.
  • Keep sitemaps up to date and avoid long redirect chains, which are common after successive migrations.
  • Publish the main content as text, not locked inside images, scanned PDFs or components that depend on interaction.

Performance also comes in here. Fast, stable pages help both people and crawlers, as detailed in the article on Core Web Vitals, WCAG and conversion.

Multiple languages and countries: hreflang

Groups with international operations often have the same page in several languages, sometimes with versions by country. The hreflang attribute tells Google which pages are equivalent versions in another language or region, and it can be declared in the HTML, in the HTTP header or in the sitemap.

According to Google's documentation, each version should list itself and all the others; if two pages do not point to each other, the annotation may be ignored. The x-default value indicates the destination page when no language matches the user's. Google also states that it does not use hreflang or the lang attribute to detect language: the text must actually be translated. In groups with many brands and domains, this discipline depends on multi-brand website architecture decisions made before the first page goes live.

Structured data that describes the organization

Structured data is markup in the code, usually in the schema.org vocabulary, that describes content in a standardized way. For a large company, the starting point is Organization markup on the home page or the corporate page. Google states that it helps distinguish the organization from others and that there are no required properties: it recommends including the relevant ones, such as name, legal name, URL, logo, contact information and official profiles (sameAs).

The markup must represent the visible content of the page. Correct markup does not guarantee prominence in results, and misleading markup can lead to a manual action.

The brand as a clear entity

Search and AI systems try to understand who each piece of information refers to. In corporate groups, the parent company, trade brands, subsidiaries, former names and similar products coexist in the same ecosystem.

  • Keep a corporate page that explains the group's structure and the relationships between the parent company and its brands.
  • Use the same name, description and contact details on the website, in official profiles and in public documents.
  • Record name changes and discontinued brands, so that old information does not look current.
  • Give each brand, product or unit a reference page.

Verifiable content, with authorship and date

Google's guidance on helpful content asks for content made to help people and proposes three questions: who created it, how and why. Google recommends accurate authorship information where readers expect to find it and states that E-E-A-T (experience, expertise, authoritativeness and trustworthiness) is not a ranking factor on its own but a set of qualities its systems try to identify through various signals. Of these, trust is the most important.

In practice: answer the main question at the start, publish your own data with source and date, identify who is responsible for the content and keep reference pages up to date.

What not to do: mass-produced content

The most common temptation is to publish hundreds of AI-generated pages to cover every search variation. Google's spam policy calls this scaled content abuse: generating many pages with the main purpose of manipulating rankings rather than helping users, no matter how they were created. Examples include using generative AI to produce many pages without adding value and rewriting third-party content, including through automatic translation.

The policy does not prohibit AI in production. It penalizes scale without value. One page per city with the same text and only the name swapped exposes the domain to risk.

How to measure presence in search and AI answers

  • Search Console: according to Google, traffic from AI Overviews and AI Mode is included in the performance report, under the Web search type, with no separate report.
  • Brand and non-brand: Search Console's branded queries filter separates searches containing the name of the company or its products from all others. The classification is done by an AI-assisted system and can be wrong.
  • Landing pages and actions: track where visits arrive and what they do next.
  • Citations in AI answers: keep a fixed set of questions and record the system, date, mention, citation as a source and accuracy of the description.

This last measure is emerging and imperfect. Answers change depending on the person, the moment and the wording, and a single snapshot does not represent what others see. Use it as a trend, never as a scoreboard.

Hypothetical example: measuring citations without fooling yourself

Consider a hypothetical manufacturer that tests 40 questions about its category every month in 3 AI systems: 120 records per round. In the first round, the company is cited as a source in 18 of them, or 15%. In 9 of those 18, the description uses the name of a division sold years earlier. The situation is illustrative.

The priority is not to publish more pages. It is to fix the corporate page, update the Organization markup, review official profiles and take down old documents. In the following rounds, the team tracks whether the correct description starts to prevail, knowing that no correction guarantees future citations.

SEO and GEO checklist for the corporate website

  • Robots.txt and bot protection reviewed, with an explicit decision on search crawlers and training crawlers.
  • Important pages indexed, eligible for snippets and with their main content in text.
  • Duplication, soft 404s and redirect chains under control.
  • Reciprocal hreflang across languages, with x-default defined.
  • Organization markup consistent with the corporate page.
  • Authorship, date and sources visible in reference content.
  • No mass production of pages without value of their own.
  • Search Console segmented into brand and non-brand, plus a periodic panel of questions.

Frequently asked questions about SEO and GEO

What is GEO?

GEO, or generative engine optimization, is the work of getting a brand and its content to appear as a source in AI-generated answers. The term appears in a 2023 academic paper and has spread through the market. On Google, the fundamentals are the same as for SEO.

What is the difference between SEO and GEO?

SEO looks at the presence of pages in search results: position, clicks and queries. GEO looks at the company's presence inside AI answers: whether it is mentioned, cited as a source and described correctly. The technical and editorial foundation is practically the same.

Do I need an llms.txt file to appear in AI answers?

For Google, no. The official guidance says there is no need to create text files for AI or new markup for the AI features in Search. Other systems have their own rules, which should be checked in each one's documentation.

How do I know if my company appears in AI answers?

On Google, traffic from these features is included in Search Console, under the Web type. For the answers themselves, the feasible approach today is a fixed panel of questions tested periodically. It is an imperfect measure, because answers vary by person and moment.

Can I use AI to produce website content?

Yes, as long as the result is useful, reviewed and has value of its own. Google's spam policy treats generating many pages to manipulate rankings as abuse, not the use of AI itself.

To bring this into your planning, learn about the Website practice and our SEO and GEO work and the article on multi-brand website architecture.