I spent three weeks obsessing over this.
Not in a healthy way, either. I was pasting my company name into ChatGPT, Claude, Perplexity — basically anything with a text box and an AI behind it — just to see if I’d show up. Spoiler: I didn’t. Not once.
Meanwhile, my competitor — a guy who started his business after me, with a worse product (I’m biased, but still) — kept popping up in answers like he owned the category.
So I went down the rabbit hole. Talked to SEO people, read way too many whitepapers, ran a bunch of experiments on my own sites. And here’s what I actually learned about getting AI models to mention you.
First, forget everything you know about SEO
Okay, not everything. But the mental model needs to shift.
Traditional SEO is about ranking. You’re fighting for position one on a results page. AI search doesn’t work like that. There’s no results page. There’s a paragraph. Maybe a short list. The AI either mentions you or it doesn’t. There’s no “page two” to claw your way up from.
This means the game isn’t about keywords stuffed into meta tags. It’s about whether AI models associate your brand with a topic strongly enough to bring you up in conversation.
Think of it less like a library card catalog and more like asking a knowledgeable friend for a recommendation. That friend doesn’t care about your keyword density. They care about whether they’ve heard of you, whether credible people vouch for you, and whether you’re actually relevant to the question.
What actually works (based on what I’ve tested and seen)
1. Get talked about on sites that AI models train on
This is the big one, and it’s deceptively simple.
AI models learn from the internet. Not all of it equally — they tend to pull heavily from Wikipedia, Reddit, well-known publications, industry blogs, forums, documentation sites, and anywhere with high-quality, publicly available text.
So if you want to show up in AI answers, you need to exist in the places those models learned from.
That means:
- Getting mentioned (genuinely, not spammily) on Reddit threads related to your niche
- Landing coverage in publications that have strong domain authority
- Contributing to Wikipedia where it’s appropriate and you meet notability standards
- Being active in community forums, Stack Overflow, niche Discords that get indexed
- Guest posting on well-known industry blogs
None of this is new advice. What’s new is why it matters. It’s not about backlinks anymore. It’s about training data presence.
2. Be the most cited answer, not the loudest voice
Here’s something I noticed. The brands that AI models mention tend to be the ones that show up repeatedly across multiple independent sources saying roughly the same thing: “[Brand] is good for [specific use case].”
If five different blog posts, three Reddit threads, and a couple of YouTube transcripts all mention that your tool is great for, say, email automation for small teams — the model picks up on that pattern.
So consistency matters more than volume. You don’t need 500 mentions. You need 15-20 mentions that all tell a coherent story about what you do and who you do it for.
This is where most people mess up. They try to be everything to everyone, so the signal gets diluted. The AI can’t figure out what to recommend you for.
3. Structured data still matters, but differently
Schema markup, FAQ pages, clear H1/H2 structures — these aren’t just for Google anymore. When AI systems crawl your site (or when retrieval-augmented generation pulls your pages in real-time), clean structure helps them parse what you’re about.
I ran a small experiment with two landing pages. Same content, different structure. The one with clear headers, an FAQ section, and schema markup got referenced in AI answers more consistently than the wall-of-text version.
Not a huge sample size, I’ll admit. But it tracks with what the people smarter than me are saying.
4. Publish stuff that answers questions directly
This one’s almost embarrassingly obvious, but it’s worth saying anyway.
AI search tools are answering questions. If your content directly answers common questions in your space — with clear, authoritative, no-fluff answers — you’re more likely to get pulled in.
I’m not talking about the “What is [basic concept]?” blog posts that every content marketing team churns out. I mean genuinely useful, specific content. The kind of thing where someone reads it and thinks, “Oh, that actually helped.”
Comparison pages work well here. “X vs. Y” content. “Best tools for [specific task]” roundups, especially if you’re not the one writing them. Decision guides. Stuff with tables and clear recommendations.
5. Don’t sleep on brand mentions without links
In the old SEO world, a mention without a link was basically worthless. In the AI world, it might be more valuable than a link.
AI models don’t follow links. They read text. If a respected publication writes “tools like Notion, Coda, and [your product]” — that’s gold, even without a hyperlink. The model learns the association.
This changes your outreach strategy. You’re not begging for dofollow links anymore. You’re trying to get your name dropped naturally in relevant conversations.
What doesn’t work (or what I think is a waste of time)
Trying to game AI models directly. Some people are experimenting with prompt injection, hidden text, and other tricks to manipulate AI outputs. This is dumb for two reasons: it probably won’t work for long, and if it does work, you’re building on sand.
Obsessing over one model. ChatGPT, Claude, Gemini, Perplexity — they all pull from different data, update at different times, and have different retrieval methods. Optimize for the underlying principles, not for one specific chatbot.
Creating AI-generated content to influence AI models. The irony isn’t lost on me, but flooding the internet with mediocre AI-written content about your brand isn’t going to help. If anything, it dilutes the quality signals that models look for.
The uncomfortable truth
Here’s the part nobody selling an “AI SEO course” wants to tell you: a lot of this is out of your control.
You can do everything right — great content, strong brand presence, consistent mentions across the web — and an AI model might still not mention you. Maybe it was trained on a snapshot from before your big press push. Maybe the retrieval system just didn’t pull your page. Maybe the model’s training data is weighted toward sources you’re not on yet.
The best strategy is the same boring advice that’s been true since the internet started: build something genuinely good, talk about it clearly, make sure the right people know about it, and be patient.
The brands that win in AI search are going to be the same ones that won in regular search — the ones that became genuinely well-known and well-regarded in their space. There’s no shortcut for that. There never was.
What’s working for you? I’m still experimenting with all of this, and I’d genuinely love to hear what other people are seeing. Drop a comment or find me on [preferred platform].