For twenty years, “visibility” in search marketing meant one thing: where do you rank. Position one, position three, page two — the metric was blunt, but it was singular, and everyone agreed on how to measure it. AI search has broken that agreement. When ChatGPT, Google AI Overviews, Perplexity, or Gemini answer a question, there is no ranked list of ten blue links to check your position against. There’s a paragraph of synthesized text, sometimes with a source or two attached, sometimes with none at all.
So what does “visibility” even mean in that world? The answer, according to the third lesson in Ahrefs’ AEO course, is that it isn’t one thing — it’s three. Every time an AI platform generates a response to a query in your space, your brand lands in exactly one of three states. Understanding all three, not just the flashy one, is the difference between an AEO strategy that actually works and one that’s quietly missing two-thirds of the picture.
The Old Model Doesn’t Map Onto AI Search
Before getting into the three types, it’s worth sitting with why the old model breaks down. Traditional SEO visibility is essentially a single continuous variable: your rank. Rank 1 is better than rank 2, which is better than rank 3, and so on down to “not ranking at all.” It’s a ladder, and everyone is standing somewhere on it.
AI search doesn’t generate a ladder. It generates a single answer, assembled from whatever the model decided was worth including — and your brand either made the cut in a visible way, made the cut in a way nobody can act on, or didn’t make the cut at all. Those aren’t three points on a spectrum from “great” to “bad.” They’re three fundamentally different outcomes with different causes and different fixes. Treating them as one blurry “visibility score” hides where the actual problem is.
Type 1: Cited and Linked
This is the visibility everyone wants, and the one most AI-tracking tools are built to measure. It’s the state where the AI platform not only mentions your brand or references your content, but includes an actual clickable URL pointing back to your page. The user reading the answer sees your name and has a direct path to visit your site.
This is the closest AI search analog to ranking #1 in classic organic search. It carries all the obvious benefits — direct traffic, the implicit credibility of being the source an AI model chose to point to, and a measurable, trackable outcome. Perplexity makes this state easy to observe because it exposes its citations explicitly; Google AI Overviews and ChatGPT are less consistent about surfacing links, but when they do, it’s this first type of visibility in action.
Getting here typically requires the same fundamentals that get you ranking well in traditional search — clear structure, authoritative content, technical crawlability — plus content that’s specifically shaped to be extractable: direct answers near the top of the page, well-organized data, and claims backed by evidence a model can point to with confidence.
Type 2: Mentioned But Not Linked
This is the state that’s easy to miss and easy to undervalue, and it’s arguably the more common of the two “you showed up” outcomes. Here, the AI names your brand directly in its answer — “Ahrefs offers a tool for this,” “brands like Toyota and Honda typically include this feature” — but there’s no link. The user gets your name, not a path to your site.
It’s tempting to write this off as a non-event, since it doesn’t drive a click. That’s a mistake. A brand mention without a link still does real work in two ways. First, it shapes the user’s mental map of the category — being named alongside the players they already know signals that you belong in that conversation, even if they don’t click through in the moment. Second, and more importantly for the long run, repeated unlinked mentions across many sources and many AI responses are exactly the kind of signal that gets baked into future model training. Today’s unlinked mention is tomorrow’s stronger association, and eventually tomorrow’s citation.
This is also the state where the connection between AI visibility and traditional brand-building gets clearest. You don’t earn unlinked mentions through on-page optimization alone — you earn them by being talked about consistently and accurately across the wider web: comparison articles, review roundups, forum discussions, industry roundups, and other people’s content that happens to reference you. It’s less “optimize your page” and more “exist credibly across the internet in a way a model would have absorbed.”
Type 3: Not Visible at All
The third state is the one most marketers never check, largely because there’s nothing to see — which is exactly the problem. This is the case where a user asks a question squarely in your category, and your brand simply doesn’t appear anywhere in the response. Not cited, not mentioned, not implied. Absent.
This matters more than it sounds like it should, for a simple reason: you cannot fix a gap you haven’t measured. A brand that only tracks its citation count or its mention count is implicitly ignoring every query where it should have shown up and didn’t. If you’re benchmarking your AI visibility purely against your own historical numbers, you’ll never notice that a competitor is capturing 80% of the “not visible” queries that belong to your category, because those queries never show up in a citation report at all — they only show up when you deliberately go looking for where you’re missing.
Finding this gap requires a different kind of research than tracking mentions: running representative queries a potential customer would actually ask, checking whether your brand shows up at all, and comparing that against where competitors show up instead. Tools like Ahrefs’ Brand Radar are built partly for this reason — not just to count your citations, but to map the queries where the category gets discussed and you’re conspicuously absent.
Why All Three States Matter Together
The real insight in this lesson isn’t any one of the three states — it’s that they form a complete picture only when looked at together. A brand tracking only citations (Type 1) is optimizing for the finish line while ignoring the two much larger buckets of activity that determine whether you ever get there. A brand that’s satisfied with unlinked mentions (Type 2) without pushing toward citations is leaving traffic and conversions on the table. And a brand that isn’t actively hunting for Type 3 gaps has no idea how much of its category conversation is happening entirely without it.
Think of it as a funnel, not a scoreboard. The goal isn’t just to increase your citation count — it’s to systematically move queries from “not visible” to “mentioned,” and from “mentioned” to “cited and linked.” That means different tactics for different stages:
- Closing Type 3 gaps starts with content coverage — are you publishing anything at all that addresses the query in question? If a competitor has a dedicated page on a topic and you don’t, you can’t expect to appear in an AI answer about it.
- Turning Type 3 into Type 2 often comes down to being discussed accurately elsewhere — earning mentions in comparison content, review sites, and community discussions where your category gets talked about, even if you don’t control that content directly.
- Turning Type 2 into Type 1 is where classic AEO content practices matter most — structuring your own pages so they’re easy for a model to extract and cite directly: clear, quotable answers, strong schema markup, and content authoritative enough that a model prefers linking to you over merely naming you.
The Takeaway
AI visibility isn’t a single score, and treating it like one is the fastest way to misread your own performance. The three states — cited and linked, mentioned but unlinked, and invisible — each tell you something different about where your brand stands and what to fix next. A healthy AEO strategy tracks all three, not just the one with a dashboard number attached to it, because the biggest opportunity is usually hiding in the state nobody’s measuring: the queries where you should be part of the conversation and simply aren’t.
Before chasing more citations, it’s worth asking a more uncomfortable question first — how many relevant AI conversations are happening in your category right now where your brand doesn’t come up at all? That number, once you find it, tends to be the real starting point for an AEO strategy.