AI search and AI visibility, explained

By Tanya Van Gastel, CEO and founder of Rankingonai.com

Need this done for your SaaS? Work with Tanya Or invite her to speak at your event.

I run an agency that helps SaaS companies get recommended by ChatGPT, Perplexity and Google. People ask me what that actually means, so here is the short version, in my own words.

What AI search is

AI search is when someone asks a chatbot instead of typing a few words into a search box. The big ones are ChatGPT, Perplexity, Claude and Gemini, plus Google's own AI Overviews, the generated answer that now sits at the top of many Google results. The person types a full question, like which scheduling tool is best for a small B2B team, and gets a written answer with a handful of named products in it.

That answer is not magic. The model reads pages from the web, mostly the same pages that rank well on Google, and summarizes them. In ChatGPT you can open the sources panel and see exactly which pages it used. I show this in my conversation with Cal.com's head of go-to-market (watch the video).

Why it matters for SaaS

Buyers now ask AI which product to choose. In my experience the traffic you see from ChatGPT in analytics is a small part of the picture, because the chatbot often names a product without linking to it. The person reads the name, then Googles it, and the visit shows up as branded or direct traffic. So AI visibility is worth much more than the referral number suggests.

It is also winner-take-all in a way social media is not. Someone asks the question once, gets three names, and picks one. If you are not in those three, you were never in the running. For Cal.com, one in three meetings on the sales side is now attributed to AI search, which is why I think this is the most important marketing channel a SaaS company can work on right now.

How brands get recommended

Three things, and none of them are complicated.

Be talked about on the sources AI reads. Ask ChatGPT the question your buyers ask, open the sources, and write down the pages it cites. Usually they are best-of lists, comparison articles and review sites. Get into those pages, and publish your own honest comparisons and best-of posts so the model does not have to guess how you differ. I go through this on the Zapier breakdown (watch the video) and on the MWC panel (watch the video).

Be extractable. AI answers are built from clear statements it can lift out of a page. Say what the product is, who it is for, what it costs and which use cases it covers, in plain sentences, with headings and short lists. Answer the specific questions people ask a chatbot, such as the best tool under a certain budget or for a certain team size. Vague marketing copy gives the model nothing to quote.

Be refreshed. The models lean towards recently updated pages, so refreshing a good article beats publishing another new one. Add the things that signal authority: real customers, funding, partners, an author bio. And do not mass-publish AI-generated articles. It reads as spam to Google and to the models, and it can undo the work you have done.

Where to start

Run a small audit. Ask ChatGPT three things: which product it recommends for your main use case, which is the best in your category, and what people say about your brand. Note who shows up and which sources it used. Then fix the basics: get analytics tracking signups or demos properly, pick a few bottom-of-funnel keywords, write the comparison pages, and reach out to the listicles that already rank. Give it three months. The full beginner walkthrough is in my generative engine optimization guide (watch the video), and if you want help, that is what Rankingonai.com does.

Want help with this for your SaaS? Work with Tanya Or invite her to speak about AI search at your event.

Watch the beginner guide

My 12-minute walkthrough of generative engine optimization, from which LLMs matter to a three-prompt audit. The full transcript is on this site.

Back to the main page about Tanya Van Gastel