← Back to Blog

Methodology September 6, 2026 Updated 21 Oct 2026 9 min read By Swapnil Patle, Product Head

Not Named by AI Yet? What Your First Result Means and What to Do

A score of 0 means that no answer in your latest check named you. It does not mean the engines cannot read your site. It is a common first result for a small or new brand, because the engines name the businesses they find on roundups, comparison posts and review sites, and a brand that is not on those pages yet is left out. Where any business was named, the result shows who. If you just opened your dashboard to a zero, the number is telling you something useful.

Since scoring version 6 (27 September 2026) advice-only questions are left out of the headline score and a bare link no longer counts as a mention, so a zero is a zero on the questions where a brand could have been recommended.

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 with a badge, and since scoring version 9 they are left out of the headline score and scored apart as the branded score, so the headline is the number that measures discovery.
• 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.

The short version: five steps

If you only read one part of this post, read this. In plain words, a zero means the engines could not find enough clear, consistent evidence that you fit the question. These five steps are the usual way out, in order:

1. Let the AI crawlers in. Check that robots.txt allows GPTBot, ClaudeBot, PerplexityBot and Google-Extended, and that who you serve, what you do, where and what it costs are written in plain text on your site, not only in images or scripts.
2. Answer the buyer’s question on your own site. One clear page per main question, with the answer in the first lines.
3. Get onto the sources the engines read. Directories, review sites, comparison and “best of” lists, industry associations: the sites that come up as sources for your questions.
4. Say the same thing everywhere. The same business name, category and one-line description on your site, Google Business Profile, LinkedIn and every directory.
5. Re-check weekly and give it four to six weeks. One check moving is not a trend; a few weeks in a row is.

The free audit shows that these steps exist for your site. The detailed tasks for each of your questions, the list of sources the engines read and tracking the movement week by week come with the 7-day free trial or a paid plan.

Where the app shows this for your own site

When your first result is a zero, the Home page opens with a card titled Why no answer names you yet. It is written from your own stored answers, with no guesswork: how many answers it read, how many pages on other sites the engines read for your questions, whether they read your own site, and who they named instead. Under it is a short “What to do” list of three to five steps, most specific first:

1. Get onto the page the engines read most for your questions, named, with how many of your questions it was read for and which businesses were named from it. Then the next one.
2. Answer your most-asked question on your own site, in the first lines of one page.
3. Say who you are the same way everywhere: name, one-line description and location.
4. Publish a fair comparison with the business named most, when there is one.

On the trial and on paid plans those steps carry the real sites, pages, questions and names. After a trial ends, the same steps show without the specifics. Every version ends the same way: answers usually change in the checks after the pages they read change, not on the same day, and the checks show it when they do.

On a paid plan, the Weekly plan page then lays the whole action plan out over four to six weeks, with an estimated score band and your real weekly score next to it; how that plan is built explains the weeks and the estimate.

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 ± range, the range to expect between two checks — 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 ("Acme Corp" against "AcmeCorp"), add the alias, and the stored answers will be re-read; when the answers link a name to your site that does not count yet, the Dashboard asks "Is this you?". Third, the competitor list: if the businesses 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; a place or a directory is not a competitor.