A few months ago, I noticed the first visitors coming to RankBuddy from ChatGPT.
It wasn't a lot of traffic, but it got my attention.
I spend a lot of time thinking about Google rankings because RankBuddy is an SEO product. But this was different. Someone had asked an AI a question, ChatGPT had decided that a RankBuddy page was useful, and that person ended up on my website.
This article is about getting your website cited by ChatGPT. If you are looking for chat tools that help you do SEO, see my comparison of the best SEO chat tools.
So I started looking more closely at a simple question:
What actually makes a website appear in ChatGPT?
There is already a lot of advice about GEO, AEO, LLM optimization, AI SEO and whatever the next acronym will be.
Some of it is useful. Some of it is just traditional SEO renamed for AI. This guide is my attempt to separate the two.
I’m building RankBuddy, and I’m actively working on getting it discovered not only through Google, but also through ChatGPT and other AI search products.
Here is the process I’m using.
First, You Don't Really "Rank" in ChatGPT
This is probably the most important thing to understand.
There isn't a stable position #1, #2 or #3 like a traditional Google SERP.
Ask ChatGPT the same product recommendation question multiple times and the answer can change. Change the wording slightly and you may get a completely different set of companies.
So when I think about AI search for RankBuddy, I don't ask:
"What position does RankBuddy rank in ChatGPT?"
I ask:
"How often does RankBuddy appear when someone asks questions where RankBuddy should reasonably be one of the answers?"
That's a much more useful way to think about it.
Recent research from Semrush also points in this direction. Their 2026 analysis found that visibility can change significantly between related prompts, so measuring a single prompt can give you a misleading picture of your actual visibility.
The goal isn't one ranking.
The goal is to become a brand that keeps appearing across a group of relevant questions.
Step 1: Make Sure ChatGPT Can Access Your Website
Before doing anything clever, check the boring technical part.
OpenAI says that public websites can appear in ChatGPT search, but to make your content eligible for inclusion in summaries and snippets, you should allow OAI-SearchBot to crawl it.
Check your robots.txt.
For example:
User-agent: OAI-SearchBot
Allow: /
One thing I would not confuse here is OAI-SearchBot and GPTBot.
They have different purposes.
OAI-SearchBot is relevant to ChatGPT search visibility.
GPTBot relates to whether your content may be used for model training.
You can block GPTBot and still allow OAI-SearchBot.
That distinction matters because a surprising amount of advice online still treats every OpenAI crawler as if it does the same thing.
Also check your CDN, firewall and bot protection. Allowing a crawler in robots.txt doesn't help much if Cloudflare or another layer returns a 403 when the crawler actually visits the page.
Step 2: Decide Which Questions You Actually Want to Appear For
This changed how I think about AI visibility.
For Google SEO, I normally start with keywords.
For ChatGPT, I start with questions.
For RankBuddy, examples might be:
"What are the best AI SEO tools?"
"How can I automate SEO for my SaaS?"
"What is the best SEO tool for a small business?"
"Can AI manage SEO automatically?"
"What's a good Semrush alternative for a small business?"
These questions are much closer to how someone actually talks to an AI assistant.
I then group them by topic.
Instead of trying to win one prompt, I want RankBuddy to become associated with topics like AI SEO agents, automated SEO, SEO for small businesses and AI-assisted content strategy.
This is also why I think traditional keyword research alone isn't enough for AI search.
The underlying topic matters more than one exact phrase.
Step 3: Create Content That Deserves to Be Used as a Source
This sounds obvious, but it has changed how I write content for RankBuddy.
Previously, it was easy to think:
"Google wants a 2,500-word article about this keyword, so let's write a comprehensive article."
That isn't enough anymore.
If I publish another article that repeats the same ten points already available on twenty other sites, there is very little reason for ChatGPT to choose mine as a source.
I'm trying to make RankBuddy articles more experience-driven instead.
That means including what I tried, what happened, screenshots where they help, actual workflows, examples from RankBuddy, mistakes I made and opinions I can explain.
This article itself is an example.
I'm not writing about AI visibility because I found the keyword in a keyword tool.
I'm working on this problem for my own SaaS.
That gives me something original to say.
It also makes the article harder to replace with another generic AI-generated SEO guide.
Step 4: Make Your Answers Easy to Extract
I still care about content structure.
But not because there is a secret "ChatGPT formatting formula."
I care because good structure makes the page easier for both people and machines to understand.
When I write an important section now, I try to answer the main question early instead of spending four paragraphs getting there.
A heading like:
How do I get my website into ChatGPT?
should immediately be followed by the answer.
Not by a long introduction.
I also like specific headings, clear comparisons, concise definitions and examples that can make sense without reading the entire article.
The objective is simple.
If an AI system retrieves the page, it should be easy to understand exactly what each section says and when it is useful.
Step 5: Get Other Websites to Talk About You
This is the part I'm spending more time on right now.
And I think it may be one of the biggest differences between optimizing your own pages and building actual AI visibility.
You can say anything you want about your company on your own website.
If ten independent websites describe your product in a similar way, that's a very different signal.
Ahrefs analyzed 75,000 brands and found that branded web mentions had a much stronger correlation with AI visibility than traditional backlink metrics such as the number of backlinks or Domain Rating. Correlation doesn't prove causation, but the difference is large enough that I wouldn't ignore it.
So I've started doing this manually for RankBuddy.
I search for articles where RankBuddy genuinely belongs, especially pages about AI SEO tools, SEO automation and alternatives to traditional SEO software.
Then I find the person who wrote the article.
And instead of sending:
"Hi, please add my backlink."
I read the article, understand how they selected the products and explain why I think RankBuddy could add something useful.
It's slow.
But it has already taught me much more about how websites get mentioned than exporting 10,000 backlinks from an SEO tool ever did.
A tool I found while doing this manually
Funny enough, this exact process is how I found Sightivo.
I was searching for relevant AI SEO articles, found one of their posts, looked up the author and reached out.
Then I realized Sightivo is building software around much of the process I was doing manually.
It finds publications, comparison pages and other places where your company could reasonably be mentioned. It can also surface the people behind those pages and help turn those opportunities into outreach.
What I especially like about the idea is that it isn't just looking for random backlinks.
Sightivo also looks at where brands are being mentioned and cited in AI answers, then uses those gaps to surface potential opportunities.
For a founder doing outreach manually, that's a much more interesting use case than another database with millions of backlinks.
This is one of the tools I'd look at if the manual process starts becoming the bottleneck.
Step 6: Don't Focus Only on Links
This is another thing I'm changing in my own approach.
For traditional SEO outreach, the obvious goal is usually a backlink.
For AI visibility, I'm interested in the mention itself too.
If an article about AI SEO tools describes RankBuddy as an AI SEO employee that monitors a website, finds opportunities and helps execute SEO work, that context matters to me even if the link itself isn't the strongest backlink in the world.
Again, the research here is interesting.
Ahrefs found branded web mentions correlated strongly with AI visibility, while backlink metrics showed much weaker correlations.
That doesn't mean backlinks suddenly don't matter.
They still matter for SEO, discovery and authority.
It means I wouldn't reject a good brand mention just because it isn't a perfect dofollow backlink.
The words around the brand matter too.
Step 7: Make It Obvious What Your Brand Actually Does
This is something small companies can easily get wrong.
On one page you're an "AI SEO platform."
On LinkedIn you're an "SEO automation tool."
In a directory you're an "AI content generator."
Then somewhere else you're an "SEO copilot."
Eventually nobody, including an AI system, has a clean picture of what the company actually is.
I'm trying to make this much more consistent for RankBuddy.
The positioning I'm using is:
RankBuddy is an AI SEO employee that helps businesses grow their traffic across Google and AI search.
The exact wording doesn't need to appear robotically everywhere.
But the entity, product category and core problem should stay consistent.
I want Google, ChatGPT and a human reader to arrive at roughly the same understanding of what RankBuddy does.
Step 8: Create the Pages AI Assistants Already Need
Another mistake would be publishing only educational blog posts.
People ask AI assistants a lot of commercial questions.
They ask for alternatives.
They compare products.
They ask which software is better for a specific situation.
They ask which tools they should buy.
That means comparison and alternative content matters.
Semrush's 2026 research found that comparative content produced significantly more brand mentions than purely informational content in the dataset they studied.
So for RankBuddy I'm building content around questions such as Semrush alternatives, Moz alternatives, automated SEO tools and specific workflows where someone would realistically be choosing software.
But there is an important catch.
I don't want these articles to become fake comparison pages where RankBuddy magically wins every category.
If I test a tool and it's better for a certain use case, I should say that.
If RankBuddy is better suited to a different kind of user, I should explain why.
That makes the article more useful, and it gives both search engines and AI systems something more credible than marketing copy.
Step 9: Measure Mentions, Citations and Actual Traffic Separately
There are three different things I care about.
First, does ChatGPT mention RankBuddy?
Second, does it cite rankbuddy.io?
Third, does anyone actually click through to the website?
Those are not the same thing.
Semrush found that citations and brand mentions frequently don't happen together. In its study, 62% of AI citations did not result in the brand itself being named in the generated answer.
So I wouldn't use only citations as the KPI.
And I definitely wouldn't check one ChatGPT conversation once per month and call that tracking.
For traffic, OpenAI makes things relatively easy. ChatGPT adds utm_source=chatgpt.com to referral URLs from ChatGPT search, so you can identify that traffic in your analytics.
That's actually how I became more interested in this topic in the first place.
I started seeing real people reaching RankBuddy through AI search.
Even small numbers are interesting when you know the person asked an AI for help before arriving.
What I Would Not Spend Too Much Time On
There are a few things I think are getting more attention than they deserve.
llms.txt is interesting, and I've experimented with thinking about how AI systems consume websites, but it is still a proposed standard rather than a guaranteed ranking mechanism. The current llms.txt specification itself describes it as a proposal.
I would add it if it's easy.
I just wouldn't expect it to make ChatGPT suddenly recommend your company.
The same goes for adding endless schema types because someone says "AI loves schema."
Use structured data where it accurately describes the page and helps search engines understand it.
But I wouldn't treat schema as a shortcut around weak content, zero brand mentions and no authority.
And I definitely wouldn't publish 100 generic AI-written articles hoping volume alone will solve AI visibility.
If every page says the same thing everyone else already says, you haven't created a reason to cite you.
The Process I'm Using for RankBuddy Right Now
My current approach is much simpler than most GEO frameworks.
Make sure AI search crawlers can access the site. Decide which topics RankBuddy should be associated with. Publish useful content based on actual experience. Create comparison and alternative pages where they genuinely help buyers. Build mentions on relevant third-party sites. Keep RankBuddy's positioning consistent. Track whether the brand appears across groups of relevant prompts. Watch actual ChatGPT referral traffic.
Then repeat.
I'm still early in this process.
I don't think anyone has AI search completely figured out yet, especially because the systems themselves keep changing.
But I do think one principle is becoming increasingly clear:
You can't control what ChatGPT says about your company only from your own website.
You need a good website.
But you also need the rest of the web to know you exist.
That's why I'm spending less time looking for a secret "ChatGPT ranking factor" and more time creating things worth citing and getting RankBuddy mentioned in the places where potential customers already look for recommendations.