AI Visibility
How RankX AI measures whether the six tracked AI assistants name your brand, and what each screen in the AI Visibility section is for.
AI Visibility is where RankX AI answers "do the assistants know we exist". It asks each tracked prompt on the assistants you pick for it, out of ChatGPT, Gemini, Claude, Perplexity, Grok and Microsoft Copilot, on the schedule you set for it, and records whether your brand was named, which competitors were named instead, and which sources the answers leaned on.
Five screens, and each answers a different question.
The five screens
| Screen | The question it answers |
|---|---|
| Visibility Overview | Where do we stand overall, and against whom |
| AI Prompts | Which questions are we asking, and how does each one perform |
| AI Chat Feed | What did the assistants actually say |
| AI Answer Citations | Which sources are the answers built on |
| AI Readiness | Can the assistants read our site at all |
The last one is different in kind from the other four. The first four measure answers; AI Readiness measures your site, and it needs no runs at all because it reads what is published.
Visibility Overview
The section's front door, and the only screen that leads with a comparative figure.
Share of voice is your mentions over checks with a known verdict, alongside the same figure for every rival named in those same answers. It leads because a new business is genuinely absent from AI answers, and a bare "0%" is both demoralising and uninformative. "This rival appears in 21 of 25 answers and you appear in none" is the same fact, and it is something you can act on.
Beside it: your rank among the brands compared, how many brands were in the comparison, how many distinct assistants named each rival, and the single most-named rival.
Two zeros that mean opposite things. A blank share of voice means nothing has been measured yet. A share of voice of zero means RankX AI measured and your brand was genuinely absent from every analysed answer. The first is a gap in the sample; the second is the finding worth working on.
What a check is
One tracked prompt, asked on one assistant, on one run. That triple is the atom of everything in this section, and two consequences follow:
Your check count is prompts multiplied by the assistants each one targets. Twenty-five prompts asked on five assistants is 125 checks per run, and on all six it is 150. A prompt set to two assistants is two checks, not six. That is also how the spend works, which is why credits and metering opens with the same multiplication.
A single check is not a measurement. Assistants are non-deterministic: two runs of the same prompt frequently return different brand lists. The reliable unit is a rate across a panel of prompts over several runs.
When it runs
RankX AI evaluates scheduled runs nightly at 02:00 UTC and runs the prompts that are due. Nothing runs every night unless you set it to. Every prompt carries its own schedule, daily, every three days, weekly, fortnightly or monthly, or the Website default, raised to a floor: your plan's floor, or 7 days while you are on trial, whichever is higher. See check cadence.
You can also run a check manually at any time, as often as you like, for one prompt or for all of them. The floor governs the schedule, not your ability to ask.
The very first data in an account is the onboarding baseline, dispatched the moment you complete setup.
Reading the numbers honestly
Three rules, and they are enforced in the product rather than being advice:
A rate divides by checks with a known verdict. A check that ran but could not be analysed is excluded from both halves and counted separately. If nothing has a known verdict, the rate is blank, and blank means unknown rather than 0%.
Per-assistant figures are the real figures. The six assistants disagree with each other by design, and an average hides the disagreement, which is usually the most useful thing in the data.
A failed assistant is refunded and reported by name. If a check fails, RankX AI names which assistant failed and does not charge you for it. It never surfaces the raw upstream error.
Where to go next
- AI Prompts, which is where the panel is managed and where most of your spend is decided.
- AI Chat Feed, the raw answers, and the first screen to read when a number surprises you.
- AI Answer Citations, the sources the answers were built on.
- Tracked platforms, what each assistant is good for.
- AI Readiness, whether assistants can read your site at all.
- How RankX AI measures visibility, for the method in full.
Last updated
The Dashboard
One Website's whole state on one screen. The Needs Attention feed, the twelve cards and which of them you can hide, and why a new account is mostly empty.
AI Prompts
The panel of questions RankX AI asks the assistants. Writing good ones, the three statuses, the per-plan cap, and why the prompt list is your credit bill.