For years, we’ve been talking about SEO, keywords, Google rankings, and how to attract more visitors to our websites.
But recently, another player has entered the picture.
More and more people are asking ChatGPT, Gemini, and Perplexity questions they used to type into Google.
Questions like:
“Who are the leading manufacturers of print inspection systems?”
“Which battery supplier is suitable for industrial applications?”
“Can you recommend a B2B digital marketing agency that specializes in international markets?”
And as a digital marketer, I’m increasingly interested in a question that goes beyond where our website ranks on Google.
Can AI search engines actually find us, understand what we offer, and consider our company a relevant option when answering these questions?
There’s even a name for this: GEO, or Generative Engine Optimization.
But before rushing to publish dozens of new articles in the hope that AI will notice us, I think we should check something much more basic.
Does our website actually allow these systems to find and understand the information we’re publishing?
Sometimes the problem hides in just a few lines of code.
First, What Are AI Search Engines Looking for on Our Websites?
When someone asks an AI system to recommend suppliers, products, or services, the answer may come from information the model already knows, a live web search, or a combination of both.
Not all AI systems work in the same way, and they don’t necessarily use the same sources.
But they share a common need: clear, accessible, reliable, and relevant information.
If our website blocks important crawlers, loads content in ways that make it difficult to read, or doesn’t clearly explain what our company does, we may be missing opportunities.
That’s why I recommend starting with three relatively simple technical checks.
We’re not talking about rebuilding the website or launching a major development project.
1. Check Who Is Allowed to Crawl Your Website: robots.txt
This may be one of the least impressive-looking files on your website, but it has an important job.
The robots.txt file is usually located at:
https://www.example.com/robots.txt
It lets website owners tell crawlers which parts of the site they may or may not access.
We’re all familiar with Googlebot, Google’s web crawler.
But today, AI systems also have crawlers.
For example, OpenAI operates OAI-SearchBot for search-related functionality in ChatGPT, while GPTBot serves other purposes, including collecting content that may be used for model training.
These are two different functions, and it’s important to understand the distinction.
Anthropic also operates several crawlers, including Claude-SearchBot.
What should we check?
First, we need to see whether the robots.txt file is unintentionally blocking crawlers that are relevant to our business.
For example, if the file contains:
User-agent: OAI-SearchBot
Disallow: /This tells the crawler not to access any pages on the website.
If we want to allow it to crawl our public content, we can remove the restriction or configure an appropriate permission:
User-agent: OAI-SearchBot
Allow: /But here’s an important point.
Allowing crawler access doesn’t guarantee it will visit your website, index your content, or recommend your company.
If your website is blocked at the server level, through Cloudflare, or by another security system, changing robots.txt alone may not solve the problem.
I recommend checking which crawlers are blocked, why they’re blocked, and whether allowing them access makes sense for your business.
I wouldn’t automatically open a website to every crawler on the internet.
2. Add Structured Data to Help Systems Understand Your Business
The second change involves something called Schema Markup, or Structured Data.
It sounds technical, but the idea is quite simple.
We want search engines and other systems to understand what different elements of our website represent, without guessing.
Who is the company?
What is its official website?
What services does it offer?
Is this page an article, a product page, or a service page?
One way to provide this information is through JSON-LD code, using the Schema.org vocabulary.
For example, a company website might include structured information about the organization:
<script type="application/ld+json">
{
"@context": "https://schema.org",
"@type": "Organization",
"name": "Example Company",
"url": "https://www.example.com/",
"description": "Industrial equipment manufacturer"
}
</script>Website visitors can’t see this code.
Instead, it provides structured information that compatible systems can process.
Different types of structured data can be used depending on the page.
For example:
Organization for company information.
Product for product pages.
Service for service descriptions.
Article for professional articles and blog posts.
BreadcrumbList for website navigation and page hierarchy.
But I want to emphasize something important.
Schema markup is not an automatic ticket to AI recommendations.
We cannot promise that adding a particular Schema type will cause ChatGPT or Gemini to recommend a company.
Its value is in making information more structured, consistent, and easier for supporting systems to interpret.
As always, the information in the code must accurately reflect what’s published on the website.
If a company doesn’t offer a particular service, there’s no reason to include it in the Schema just to attract attention.
What About WordPress Websites?
On many WordPress websites, SEO plugins like Yoast or Rank Math can manage structured data without manual coding.
Before adding anything new, check what’s already there.
Sometimes the SEO plugin, theme, or e-commerce platform is already generating structured data.
Adding another layer may create duplicate or conflicting information.
3. Make Sure Your Most Important Content Is Actually Accessible
This may be the most important check.
We invest so much time in content, design, and user experience that we sometimes forget to check how our website looks to a system trying to read it.
A visitor sees a beautiful page with animations, images, interactive cards, and expandable content.
But what can a crawler actually read?
Is the main headline available in the HTML?
Is the product description present as readable text?
Does important content load only after JavaScript runs?
Can crawlers follow the links to internal pages?
And has an important page accidentally been marked as Noindex?
For example:
<meta name="robots" content="noindex">This instruction tells search engines that support it not to include the page in their index.
There are perfectly good reasons to use Noindex.
Thank-you pages, certain internal pages, or content we don’t want appearing in search results may be appropriate examples.
But if that instruction accidentally appears on a key product page, we have a problem.
What should we review?
I recommend starting with the pages that matter most to the business.
The homepage, service pages, product pages, and important professional articles.
Check whether these pages are intended to be indexed, whether their main content is accessible, and whether they can be reached through normal internal links.
Also verify that the XML sitemap is up to date and that canonical URLs point to the correct page versions.
Again, remember this distinction.
A page that can be crawled and processed is not necessarily a page that an AI system will choose to mention.
The first is an important technical foundation.
The second depends on many additional factors.
The Complete AI Search Optimization Checklist
Here’s the checklist I would use before investing more time or money to improve a website’s visibility in AI search.
Website Crawling and Access
- Check the robots.txt file.
- Make sure you aren’t unintentionally blocking relevant AI search crawlers.
- Check whether Cloudflare, server settings, or other security systems are blocking permitted crawlers
- Decide separately which search crawlers and model-training crawlers you want to allow.
Structured Data
- Check whether Organization Schema is present.
- Verify that the company name, website URL, and business details are accurate.
- Review structured data on relevant product and service pages
- Check for duplicate or conflicting Schema markup.
- Make sure the structured data matches the visible website content.
Content Accessibility
- Check that important pages don’t contain unintended Noindex directives.
- Make sure main headings and essential content are available in readable HTML.
- Check that key content doesn’t depend entirely on JavaScript execution.
- Review internal links to important products and services.
- Verify the XML sitemap and canonical URLs.
What Should You Do After These Checks?
- Make sure your website clearly explains who you are, who you serve, and what you offer.
- Publish professional content that answers real customer questions.
- Keep company information consistent across trustworthy external sources.
- Test whether your company appears in relevant AI-generated answers over time.
- Document the situation before and after changes so you can evaluate results without assuming every improvement came from a specific technical fix.
Will These Three Changes Make ChatGPT Recommend Your Company?
The short answer is: not necessarily.
The longer answer is that AI search visibility is much more complex than three technical settings.
AI systems may rely on external sources, brand mentions, professional content, current information, and other signals.
The way a user phrases a question can also influence the answer.
A website that allows crawling and presents clear information has a better technical foundation than one that blocks itself or hides important content.
But that doesn’t mean an AI system will automatically choose to recommend it.
I think this distinction matters, especially as we constantly see promises like “Five Steps to Make ChatGPT Recommend Your Business.”
I wish it were that simple.
Before Spending More on AI SEO, Check the Basics
I work with companies that invest in websites, content, paid campaigns, and organic search, yet don’t always know what’s happening behind the scenes.
Sometimes, a relatively small technical review reveals a problem that may affect the visibility of important pages.
Does that mean every company needs a large, expensive GEO project?
Not at all.
I would start with a structured review of the existing website.
Can it be crawled?
Is the main content accessible?
Is it clear what the company does?
Is the information accurate, professional, and useful?
Only then would I move on to the more advanced questions.
Which questions do we want our company to appear in?
Which sources influence AI-generated answers?
How can we build a stronger professional presence across the web?
Because ultimately, even in the age of artificial intelligence, I wouldn’t want an AI search engine to recommend a company simply because someone added a few lines of code.
I’d want it to find enough reliable, relevant information to justify mentioning that company.
And the code? Its job is simply to make sure it doesn’t get in the way.
