Your AI Visibility Score Is 0. Here Is What That Means and What to Do

Two in three brands that run their first AI visibility check score under 5. On our own data, 23 of the first 36 domains that ran a check landed between 0 and 5, and only one of those had a score above 30. If you just opened your dashboard to a zero, you are in the majority, and the number is telling you something useful.

What a zero actually says

An AI visibility score is built from what the engines answered. For each question you track, each engine is asked the question with its live web search on, and the answer is scored on four things: was the brand named at all, where in the answer it appeared, whether the engine cited your own site as a source, and how it described you. A zero means the first gate never opened: no engine named you in any answer.

That is a statement about the answers, not about your website. The engines did not fail to read your site; they answered the question from the sources they trust for that topic, and those sources do not mention you. In the same check, the engines almost certainly did name someone — and that list is the most useful thing on the page.

What a zero does not say

• It does not mean your pages are badly built. Schema markup, meta descriptions and page speed decide whether an engine can read you. They do not decide whether it recommends you. A perfectly structured page that nobody else references still scores zero.
• It does not mean the questions are wrong — unless they are branded. A question that contains your name will always surface you, and it measures nothing. Check that your tracked questions are the ones a buyer types before they know any brand names; AEOTrack marks branded and near-brand questions so they are not counted in the headline score.
• It does not mean the engines are broken. If an engine shows an error or a grey chip, that engine did not answer and is excluded. A zero from engines that did answer is a real zero.

Read the two lists under the score

Every check stores the full answer. Two lists come out of that, and they are the plan:

Who they named instead. The brands that appeared in the answers to your questions, with how many answers named each. These are not necessarily your biggest competitors; they are the brands the engines associate with your category today. If a name you have never heard of appears in eight of ten answers, it is because eight of the sources the engines read mention it.

Where they got it from. The sites the engines cited while answering. A handful of roundup posts, review sites, directories and community threads typically account for most of the citations in a category. Those sites are the actual battleground: appear on them and the answers change; publish more on your own site and they mostly do not.

The three moves that change a zero

1. Get onto the pages the engines already read

Sort the cited sources by how often they were pulled in. Work down the list. For each, ask: can I get a factual entry here? Review sites and directories accept listings. Roundup authors accept a well-written pitch with real specifics. Community threads accept a genuine answer from someone at the company. One mention on a source that is retrieved for six of your questions is worth more than six new pages on your own site.

2. Publish the comparison you want quoted

If every retrieved source frames the category around three incumbents, there is no version of the comparison that includes you for an engine to find. Write it. Name the rivals, be specific about where they are stronger, and put the facts about your product in short, self-contained sentences that survive being lifted out of context. Engines extract sentences; a benefit spread across a paragraph does not extract.

3. Make sure the engines can read what you have

This is third, not first, because it is the smallest lever — but it is free. Allow the AI crawlers in robots.txt, keep an FAQ block that answers the questions you track, and add Organization schema so the engines can attach your name to your domain. The free audit on this site checks all of it in a minute.

What to expect, and when

A third-party page has to be written, published, crawled and then retrieved for the right question before an answer changes. That is weeks, not days. The first thing that moves is usually a single engine on a single question, and the tracker will show it as one green chip in a row of grey ones. Then citations appear before mentions: engines start reading your page as a source before they name you in the text. Both are progress. Watch the per-engine breakdown, not just the headline number, and expect the number to arrive with a confidence range — a jump from 0 to 8 on a single check is real only when the range says so.

If the score stays at zero after two months

Check three things. First, the questions: are they actually about the problem you solve, or are they about the category label? "Best CRM for dentists" and "best CRM" are different markets. Second, the brand terms: if the engines write your name differently ("AEO Track" against "AEOTrack"), add the alias, and the stored answers will be re-read. Third, the competitor list: if the brands being named are not on it, add them, so the share-of-voice view tells you how far behind you are rather than nothing at all.