How to read your first AEOTrack result

About five minutes after you add your website, the first check finishes and one email arrives with the numbers and one real answer. This is the moment to judge whether the data is real. Here is how to read that email and the Dashboard behind it, line by line, using the typical first result rather than a flattering one.

The subject line is the headline

The subject says the count, not a score. It reads one of three ways:

• "Acme: named in 8% of 25 AI answers" when you were named at all.
• "Acme: 0 of 25 AI answers named you. Here is who they named" when rivals were named.
• "Acme: 0 of 25 AI answers named you. Here is the first fix", or "Here is why", when nobody was.

Zero is the usual first reading. As of September 2026, 52 of 100 businesses score 0 on their first check on the platform. The email says that, so you do not have to wonder whether something broke.

The sample line

"We asked ChatGPT, Gemini, Claude, Perplexity and Grok 5 questions your buyers ask on 29 September, and read 25 answers." That is the whole sample: one live answer per engine per question, web search on. Twenty-five is not a lot, and the email does not pretend it is. Every number below it is "of 25".

A typical first result

The homepage shows the same card a new account sees, filled with typical numbers rather than one customer's data. Read it top to bottom:

Line	Example value	What it tells you
Named in	0 of 25 answers	How many of the latest answers wrote your name. A link alone does not count.
Where any brand was named	11 of 11 answers named a competitor	Of the answers that named anyone, how many named you. Here, none.
Cited as a source	4 answers, named in none	The engines read your pages and still recommended others. Being read is not the constraint.
Per question	"3 rivals named, not you", "Cited, not named", "Nobody named"	Which questions are lost to someone, which are half-won, which are advice-only.
One thing to do this week	The roundup page all three engines cite lists your two rivals and not you. Get listed there.	The single fix ranked highest by effect.

The score itself appears once there are at least three usable answers, and it comes with a ± range: the range to expect between two checks. A first score in single digits with a range wider than the score is normal. A move inside the range next week is normal variation.

Who they named instead

When no answer names you, the Dashboard leads with the brands that were named and how many answers named each. This is the most useful table in the product on day one. In 35 of 43 companies measured (81%, September 2026) a competitor was named where the company was not. If the names surprise you, add them as competitors; share of voice and the weekly "who is above you" line are built from that list.

Where they got it from

Next to the brands is the list of domains the engines cited, with the count of answers citing each. The Sources page has the full version across every question. This is where the fix comes from. In one account's 34 questions the engines cited 1,685 sources across 380 domains, and four of the ten most-cited pages were "best tools" roundup lists (what those citations showed). Your top ten will look similar. Read those pages as a buyer and note whether you are on them.

One real answer

The email quotes one answer in full: the question, the engine, and the text, with a caption saying which rival it names. Read it once as the buyer who asked. That is what your prospect saw. Every answer is stored and readable in the app, with its sources, so you can check any number against the text it came from.

Advice-only questions

Some questions get advice, not a shortlist. "How do I choose motor trade insurance" returns a how-to and names nobody. Those are marked advice-only, left out of the score, and counted next to it. Each one has a Replace action. Swap it for a question buyers ask by name and it enters the score at the next check. A question where rivals are named and you are not is not advice-only. It is a real zero, and it stays in.

The "How this score works" panel

Under the score there is a panel with the six facts you need: one answer per engine per question; which engines are on your plan; 0 if not named, otherwise 25 plus up to 75; advice-only questions are left out; the ± is the range to expect between two checks; a score older than 30 days is marked stale. The full formula is on the methodology page, and One answer per engine explains why it is built that way.

What happens next

• Day 1. An email with the weekday your first weekly check runs and, if there is one, a question a competitor already wins, with the answer.
• Day 2. The one page to change this week, chosen from the sources the engines read.
• Day 3. Connect Search Console, so the tracked questions come from real searches rather than what we inferred from your site.
• Every week. The weekly results email: score with its range, questions gained and lost, who was named instead or who is above you, and one thing to do before the next check. A Monday digest covers the rest of the team.
• Any day. A manual re-check per question, once a day, to see whether a change was picked up.

If the first result is a zero, Your AI visibility score is 0 has the three moves that change it. For the whole trial, day by day, see The first 14 days.

Before you sign up

The free audit is the same check at a smaller scale: your domain, one question your buyers ask, two engines, and who they named. Free, no account, about a minute, and the question is carried into your account if you start the trial. The trial runs at Pro capacity for its 14 days, no card. The monthly totals of answers read, cited and named go on the Named Report page.