GEO setup for an enterprise website: a 30-day plan

Category

Growth

Published Date

ElarisLabs Engineering

Summary

A 30-day GEO setup for enterprise websites: open access to AI crawlers, fix structure, rewrite key pages answer-first, then measure mentions and citations.

Share this article

In this article

  1. What changes when AI engines answer instead of search results?

  2. Week 1: Can AI crawlers reach your site?

  3. Week 2: Is your site structured for AI to read?

  4. Week 3: Do your key pages answer questions directly?

  5. Week 4: How do you measure the results of a GEO setup?

  6. How does a GEO setup change for healthcare, finance and other regulated businesses?

  7. Run your GEO setup with ElarisLabs AI

  8. Frequently asked questions

A GEO setup gets your business named and cited in AI answers from ChatGPT, Google AI Overviews, Gemini, Perplexity and Copilot. For an enterprise website it is less about tricks and more about access, structure and clear answers, and the whole GEO setup fits in 30 days.

This is the GEO setup we follow at ElarisLabs AI when we set up AI visibility for a business, with every step tied to the engines' own documentation, checked on 19 November 2026.

What changes when AI engines answer instead of search results?

An AI engine doesn't show ten links. It runs its own searches (Google calls this "query fan-out"), reads a handful of pages and writes one answer, quoting and linking the pages that answered the question most plainly. Pages that bury the answer, hide text behind scripts or turn crawlers away are skipped.

Diagram of how an AI answer gets built: the question fans out into several searches, the engine reads the pages that come back, writes one answer and cites a few sources

How an AI answer gets built: ask, fan out, read, answer, cite.

The good news is that the fundamentals still work. Google's documentation says plainly:

"There are no additional requirements to appear in AI Overviews or AI Mode."

Its May 2026 guide goes further, telling site owners to "prioritize effective SEO strategies" over what it calls AEO and GEO hacks. So this plan skips the hacks and focuses on four things every engine rewards.

Week 1: Can AI crawlers reach your site?

Every GEO setup starts with access, because nothing else matters if the crawlers can't fetch your pages. Each AI company runs separate crawlers for search and for model training, and you can allow one while blocking the other.

Company

Crawler for search and answers

Crawler for training

Worth knowing

OpenAI

OAI-SearchBot

GPTBot

ChatGPT-User fetches pages when a person asks. Robots.txt changes reach ChatGPT search in about 24 hours.

Anthropic

Claude-SearchBot

ClaudeBot

Claude-User fetches pages on a person's request. Blocking it may reduce visibility in Claude.

Perplexity

PerplexityBot

Not used for training, per Perplexity

Perplexity-User is user-initiated and generally ignores robots.txt.

Google

Googlebot (AI Overviews and AI Mode use Search)

Google-Extended

Google-Extended covers Gemini training and grounding in Gemini Apps and Vertex AI. It does not affect inclusion in Google Search.

Microsoft

Bingbot

See Bing's crawler documentation

Bing Webmaster Tools reports citations in Copilot and Bing's AI summaries.

  1. Open robots.txt. Allow the search crawlers you want to appear in: OAI-SearchBot, Claude-SearchBot, PerplexityBot, Googlebot and Bingbot. Decide on training crawlers separately. In OpenAI's words, "Each setting is independent of the others."

  2. Check your CDN and firewall. Since 1 July 2025, Cloudflare blocks AI crawlers by default on new domains and asks owners to opt in. Check AI Crawl Control, and ask your security team about bot rules on other CDNs.

  3. Put key content in the HTML. "Most AI crawlers fetch but do not execute JavaScript," according to Vercel's engineering team. Prices, specs, FAQs and comparisons should be in the page source, not loaded later by a script.

  4. Send fresh URLs. Keep XML sitemaps current, and use IndexNow to tell Bing and the other participating engines about new and updated pages.

Robots.txt example that allows OAI-SearchBot for ChatGPT search and disallows GPTBot for AI training

Search and training are separate switches. This example shows up in ChatGPT search but stays out of training.

Week 2: Is your site structured for AI to read?

Structure is how an engine works out what a page is about, who wrote it and which facts it can trust. These are the 13 checks the ElarisLabs AI page audit runs on every page AI might cite, with the pass mark for each:

Check

Pass when

Structured data

Valid JSON-LD that matches the page (Article and FAQPage on posts; Organization or Product elsewhere)

H1

One H1, 10 to 90 characters, that names the topic

Page title

30 to 65 characters, saying what the page offers

URL slug

Short, lowercase and hyphenated, with no IDs or query strings

Answer-first opening

The first sentence says what the thing is, with a number in the first 50 words

Brand visibility

Your business named near the top and at least 3 times in the text

Depth and links

600 or more words on articles, 3 or more internal links, and at least one list or table

Tables in HTML

Comparisons and specs in real HTML tables, not images

FAQ

4 to 8 real buyer questions, answered in two or three sentences, with FAQPage markup

Table of contents

Long articles link to each section, so answers can cite the exact part

Named author and dates

A real person in the byline and in the markup, plus published and updated dates

Sources and quotes

Facts linked to where they come from, and at least one named person quoted

Branded next step

A heading that names your business, with a link to act on

One honest caveat: Google's guide says structured data "isn't required for generative AI search", and that no special markup is needed. We keep it on the list because Google still recommends it for SEO, and because markup that matches the visible page helps every engine read authors, dates and prices the same way.

Week 3: Do your key pages answer questions directly?

Pick the 10 to 20 pages that matter most to revenue: products, pricing, comparisons and your best guides. Then rewrite them so an engine can lift the answer as it is.

  • Open with the answer. Say what the thing is in the first sentence and give a number (a price, a time, a size) in the first 50 words.

  • Use buyers' questions as headings. Take them from sales calls and from the questions you track, and answer each one in its first two sentences.

  • Add statistics, quotations and sources. In the Princeton-led GEO study, these three changes lifted visibility in AI answers by up to 40%, while keyword stuffing barely helped.

  • Compare honestly. Buyers ask "X vs Y" and "best X for Y". A comparison page that says where a rival wins is more likely to be quoted than one that never names them.

  • Earn real mentions. Answers lean on the pages they already trust: reviews, lists, forums and trade press. Get included there on merit; Google warns that pursuing inauthentic mentions "isn't as helpful as it might seem".

Before and after: a brochure-style opening that AI skips, and an answer-first opening with service, area, time and price that AI quotes

Write the answer, not the brochure: AI quotes the plain version on the right.

Week 4: How do you measure the results of a GEO setup?

Measure the answers themselves, not just traffic. Set a fixed list of 50 to 250 questions per region and language, ask them daily on the three engines your buyers use most, and record mentions, citations, share of voice and accuracy.

  • Google Search Console now has generative AI performance reports for impressions in AI Overviews, AI Mode and Discover.

  • Bing Webmaster Tools shows how often Copilot and Bing's AI summaries cite your pages, with the grounding queries behind them.

  • Your analytics show visits referred by chatgpt.com, perplexity.ai, gemini.google.com and copilot.microsoft.com. Count signups and sales from them, not just sessions.

  • An AI visibility tool asks hundreds of questions every day and shows which pages and competitors get named instead of you. We compare six of them in Best AI visibility tools in 2026.

In our tracking, citations usually move before mentions: an engine starts reading and linking a page before it starts naming the business behind it.

How does a GEO setup change for healthcare, finance and other regulated businesses?

The GEO setup is the same, but the bar for trust is higher. Google's Search Quality Rater Guidelines treat health, safety and financial topics as "Your Money or Your Life" pages, where accuracy and expertise matter most.

  1. Name qualified authors. Put a real, credentialed person in the byline and in the markup, with a profile that shows their expertise.

  2. Review every claim. Send rates, fees, eligibility rules and medical claims through compliance before publishing, and date every update.

  3. Link primary sources. Regulators, standards bodies and peer-reviewed studies, not other blogs.

  4. Correct answers at the source. If an answer gets your fees or approvals wrong, fix the page it cites and keep your facts consistent everywhere they appear.

  5. Vet your tracking vendor. Ask for SOC 2 Type II, single sign-on, audit logs and data retention terms before you connect analytics or Search Console.

Run your GEO setup with ElarisLabs AI

ElarisLabs AI Visibility checks your site the way AI crawlers see it, runs the 13 checks above on every page AI cites, and asks your buyers' questions every day on the engines you choose. Each gap comes with a suggested fix. Start a 7-day free trial or see how AI Visibility works.