Key takeaways
- AI visibility is whether AI answer engines know your business well enough to name it when someone asks about what you sell.
- It is measured by asking the engines real questions and counting how often, and how favourably, you come up next to competitors.
- The engines read the same things a careful person would: clear facts about you, pages that answer questions, reviews and mentions elsewhere.
- Most businesses lose for boring reasons: blocked crawlers, inconsistent names and addresses, no page that answers the question.
- The fixes are ranked, not endless. The first month is usually access, entity clean-up and a handful of answer pages.
Ten years ago, a buyer with a question opened Google, typed a few words and scrolled a list of links. Today a growing share of them open ChatGPT, Perplexity or the AI box at the top of Google, ask the question in a full sentence, and take the answer. There is no list. There is one reply, and it names one, two or three businesses.
AI visibility is whether your business is one of the names in that reply. Not whether you rank on page one. Whether the engine knows you exist, knows what you do, trusts you enough to recommend you and can point to your site as the source.
Why this is different from SEO
Search engine optimization was about earning a position in a list. You picked keywords, built pages around them and climbed. The click was the prize, and the reader did the work of reading your page.
Answer engines do the reading for the buyer. They pull from many sources, decide which ones to trust, and write a short summary. The prize is no longer the click. It is the mention inside the answer, ideally with a citation the buyer can tap.
The two overlap. A site that is fast, clear and well linked helps with both. But the engines weigh things a ranking algorithm mostly ignored: whether your facts match everywhere, whether a page answers a question in its first lines, whether other trusted sites talk about you, and whether the engine's own crawler is allowed onto your site at all.
That is what a good outcome looks like. Your name, first, with a reason attached. The reason came from your site and your reviews. The engine just put it into a sentence.
How it is measured
You cannot log into ChatGPT and see a ranking. So we measure it the way a buyer would experience it: we ask.
- Write the questions your buyers actually ask. Not keywords. Sentences. "Who does fixed-price kitchen renovations in Oakville?" "Which packaging supplier in Canada has the lowest minimums?"
- Ask them on the real engines. ChatGPT, Perplexity, Google AI Overviews and Claude, from a clean session, several times, because the answers vary.
- Record who gets named, in what order, and what is said about them. You and your top competitors, side by side.
- Turn that into a share of answer. Out of all the answers to your category's questions, how many name you? How many name you first?
That number is your AI visibility. It moves when the inputs move, which is why the second half of the work is a scorecard of the inputs.
What the engines are actually reading
Nobody outside the labs knows the exact recipe, and it changes. But the engines are consistent about the kinds of evidence they lean on. We group them into ten pillars. The five that matter most for a typical business:
Entity clarity
Can the engine tell exactly who you are? Same name, same address, same phone, same category, on your site, your Google Business Profile, directories and social profiles. A business that is "Smith Plumbing" on the site, "Smith Plumbing and Heating Ltd." on Google and "Smith's" on Facebook is three fuzzy entities instead of one clear one.
Answer-ready content
Do you have a page that answers the question the buyer asked, in the first few lines, in plain words? Engines lift short, direct passages. A page titled "Our services" with three paragraphs of history does not get lifted. A page titled "Fixed-price kitchen renovations in Oakville" that opens with what you do, where, for how much and how long, does.
Citations and authority
Do trusted third parties mention you? Industry associations, local press, review platforms, supplier directories. The engines treat these the way a careful buyer would: as evidence you are real and reputable.
AI crawler access
This is the boring one that costs the most. Many sites block the AI crawlers, usually by accident, through a firewall rule, a plugin default or a robots file copied from somewhere else. If GPTBot or PerplexityBot cannot fetch your pages, nothing else on this list matters.
Reviews and trust
Volume, recency and whether you reply. The engines quote reviews directly, and they notice a business that answered a complaint well.
What actually moves the number
The mistake most people make is assuming this needs a mountain of new content. Usually it does not. In the audits we run, the first month of fixes looks like this:
- Unblock the AI crawlers and confirm they can reach every important page.
- Fix the entity: one name, one address, one category, everywhere, with schema on the site that says the same thing.
- Write or rewrite three to five answer pages for the questions with the most buyer intent.
- Add a real FAQ with the questions people ask on the phone, marked up so engines can read it as questions and answers.
- Ask happy customers for reviews and reply to every one, including the bad ones.
Then you re-run the questions. Some engines pick up changes in days. Google AI Overviews and ChatGPT's browsing tend to take weeks. Over three to six months the gains compound as the engines re-read your site and the mentions pile up.
What it does not do
AI visibility does not replace a good website. If the engine sends someone to a slow page that hides your phone number, you lose the lead anyway. It does not replace reviews or referrals. And it is not a trick. There is no hidden text or prompt injection that works for long. The engines are built to reward the same things a careful buyer rewards: clarity, evidence and answers.
That is the good news. The businesses that win here are the ones that were already good, and just were not legible to a machine. Making them legible is the whole job.