AI visibility is how often, and how accurately, an AI assistant names your business when someone asks it a question you should be the answer to. It is earned through signals engines can verify, not bought, and never guaranteed.
- It’s a discipline, not a channel: AI visibility has no ad account and no bid. It is the by-product of being verifiable everywhere an engine looks.
- The names are mostly the same thing: GEO, AEO, AI search optimization and LLM SEO: different vintages of the same practice, with real but narrow distinctions.
- It is measurable, and most measurement is fake: a percentage with no published method is a number someone made up. The honest unit is X out of 12, per platform, dated.
- Nobody controls the output. Anyone selling a guaranteed ChatGPT recommendation is guessing on your invoice.
The answer moved above the links
For twenty years the deal was simple: rank, get clicked. That deal is being renegotiated in public. Pew Research Center followed 900 U.S. adults through 68,879 real Google searches and found that when an AI summary appeared, people clicked a traditional search result in 8% of visits: against 15% when no summary appeared. They clicked a link inside the summary in 1% of visits.
Ahrefs, looking at 300,000 keywords rather than people, found the same shape from the other end: the presence of an AI Overview correlated with a 34.5% lower average click-through rate for the page ranking first.
Read those together and the conclusion is uncomfortable but useful. Ranking still matters, because it is one of the strongest inputs into what gets summarised, but ranking is no longer the finish line. Being named in the answer is. That gap between “ranks well” and “gets named” is what this category exists to close.
Four things AI visibility is not
Not a score you can buy. Several tools will sell you an “AI visibility score” as a percentage. Ask what the denominator is. If there isn’t a published question set, a stated platform list, and a repeat-run rule, the number is decoration.
Not a separate budget line. There is no auction and no placement to buy. The work that makes an engine willing to name you, accurate identity, real reviews, pages that answer questions plainly and a site crawlers can read: that is the same work that has always earned rankings. The AI layer changed the scoreboard, not the sport.
Not prompt injection or schema tricks. Hidden text aimed at models, fake FAQ markup, invented awards: these are the 2026 equivalent of keyword stuffing, and they carry the same ending. Engines are conservative about who they name because naming the wrong business is their reputational problem, not yours.
Not guaranteeable. Model outputs shift between runs, between accounts, between weeks. Anyone promising you a permanent spot in an AI answer is describing something they cannot control. Here’s why we don’t guarantee any of it, in writing, before you hire anyone.
The six questions an engine settles before it names you
Whatever the acronym on the invoice, every version of this discipline is working on the same six layers. They are ordered: a failure high up makes everything below it irrelevant.
- 01 Eligibility, can you legitimately be the answer?Are you a real, verifiable business serving that area, in that category? No amount of optimisation survives an engine deciding you don’t qualify.
- 02 Crawler access, can they read you at all?A robots.txt line or a bot-blocking rule at the CDN can remove you from the answer entirely. It should be a decision you made, not one your host made for you.
- 03 Extractability, can they lift a clean answer?Answer-first pages, plain sentences, real structure. A page that buries its answer in paragraph nine is a page a machine will skip in favour of one that doesn’t.
- 04 Entity stability, does your story match everywhere?One name, one address, one phone, one service list, on your site, your profile, and every listing. Contradictions are the cheapest reason to skip you. This is the layer entity SEO is named after.
- 05 Citation surfaces, does anyone else confirm it?Reviews with specifics, directories that matter in your trade, local press, supplier and association pages. Engines prefer claims that more than one source supports.
- 06 Measurement, can you prove any of it moved?A baseline before the work, the same questions re-run on a schedule, screenshots kept. Without this layer the other five are opinions.
Every name people use for this, and what each one actually means
The category grew faster than its vocabulary, so the same work carries six names depending on who is selling it. The distinctions are real but small. Each page below defines one term properly, says who uses it, and admits where it overlaps with the others.
A number is only worth what its method is worth
The category has a measurement problem, and it is not a technical one. It is that a percentage looks more precise than a fraction, so tools sell percentages. “Your AI visibility is 62%” means nothing without knowing what was asked, where, how many times, and on what date.
The unit we publish and use is deliberately unglamorous: X out of 12. Twelve real buying questions for your trade and market, asked on Google AI Overviews, ChatGPT and Perplexity, each repeated three times in clean sessions, counted only when you’re named in at least two of three runs. Per platform. Dated. With screenshots. It refuses to be more precise than the underlying reality, which moves week to week.
If that sounds like a low bar, notice how few numbers in this industry clear it. The full protocol is published, question construction, platform list, repeat rule, volatility disclosure, so you can run it yourself, or check whether anyone else’s number was built the same way.
We also run it on ourselves, in public, including the months it goes the wrong way: our own numbers are here.
This page defines the category. It doesn’t sell anything. If you want to see how we run AI visibility as paid work, what’s in it, what it costs, what we measure, that lives on the AI Visibility service page.
Want the one-paragraph version of any term instead of the full treatment? The Visibility Glossary keeps short definitions; the pages above keep the practice.
Fair questions about the category
Is AI visibility just SEO with a new name?
Mostly the same inputs, a different scoreboard. Crawlability, accurate identity, real reviews and clearly answered questions have always earned rankings; they now also decide who gets named in a generated answer. What’s new is the measurement problem: there is no rank tracker for an answer that rewrites itself every few days, and the fact that a first-place ranking no longer guarantees the click.
Should I block AI crawlers to protect my content?
For a publisher whose product is the content itself, that’s a real debate. For a local service business, blocking the crawler removes you from the answer your customer is reading, you’ve protected content nobody was paying you for at the cost of the recommendation you were. Either way it should be a decision someone made deliberately, not a default that arrived with your hosting plan.
Which acronym should I actually care about?
None of them, if you’re the business owner. The label matters to agencies positioning themselves; the work is the same six layers regardless. Care about it only far enough to check that whoever you hire can define their own term without hand-waving, the AEO vs GEO comparison is a decent test to hand them.
How long before any of this shows up?
Eligibility and accuracy problems can resolve in weeks because you’re removing a blocker rather than building an asset. Presence in generated answers builds over months and moves unevenly, and it can go backwards for a month while the underlying signals improve. That is exactly why the baseline gets taken before the work starts, and why a single screenshot proves nothing in either direction.
Can reviews be part of this if I can’t incentivise them?
Yes, reviews are one of the strongest citation surfaces an engine has, and none of that requires bending the rules. Ask every customer, every job, the same way; never offer anything of value in exchange for a review, never filter by expected sentiment, never write or edit a customer’s words. That’s Google’s policy and the FTC’s rule on fake reviews, and both are cheaper to follow than to lose a profile over.
Find out where you actually stand
The free Visibility Check runs 12 real buying questions for your trade and market across three assistants, your X/12, the screenshots, and the first fix worth making. No sales call, and you keep the findings either way.
Get my free Visibility Check