What AI Visibility Tracks
AI Visibility checks whether AI assistants mention and cite your site for prompts you choose, across Google AI Overviews, ChatGPT, Gemini, Perplexity, Grok, and Claude. Each check costs 10 credits per prompt per platform.
Last updated 1 day ago
AI Visibility is at Sidebar β Measure β AI Visibility.
The page describes itself precisely: "Track when AI assistants mention and cite your project across Google AI Overviews, ChatGPT, Gemini, Perplexity, Grok, and Claude. Prompts you flag for auto-refresh recheck about once a month."
Why This Exists
People increasingly ask an AI assistant instead of searching and clicking. When they do, the assistant either mentions you or it does not, and traditional rank tracking cannot see any of it.
AI Visibility samples those answers so you can tell.
How it Works
You add prompts, which are questions your customers might ask an AI assistant. NarraSEO runs each prompt against the platforms available on your plan and records whether your brand was mentioned and whether your site was cited.
The Platforms
Which ones you get depends on your plan, as of 2026-08-30:
The Four Metrics
Average position only counts list placements. The tooltip is explicit that prose-only mentions do not count. An assistant mentioning you in a sentence rather than a list contributes to mention rate and not to position.
What Else the Page Shows
- Tracked prompts, with status for each.
- Share of voice, comparing you against competitors.
- Top sources, the sites assistants cite on your topics.
- Most cited pages, which of your pages get cited.
- Visibility opportunities, the gaps.
- AI visibility trend and Mention rate by platform.
- Suggested prompts you can start tracking with Track.
Top sources is the most quietly useful panel. It shows which sites assistants actually draw on for your topics, which tells you what kind of source you need to become.
What it Costs
10 credits per prompt per platform, as of 2026-08-30.
The cost is the number of prompts multiplied by the number of platforms. Before a run you see "Uses N credits" and a breakdown of the prompt and platform-check counts.
Ten prompts on Starter, which has two platforms, is 200 credits. The same ten prompts on Pro, with six platforms, is 600.
This is why prompt limits and platform counts matter more than they first appear.
Prompt Limits
As of 2026-08-30:
At the limit you see "Prompt limit reached. Remove one to add another."
Auto-Refresh
Auto-refresh monthly is a project toggle. Prompts you flag for it are rechecked about once a month.
The hint is worth reading: "Only flagged prompts refresh on their own. Turn this off to work by hand."
So the toggle alone does nothing. You flag the individual prompts you want rechecked, which is a sensible design because it stops your whole prompt set consuming credits monthly when you only care about a handful.
With it on you see the monthly credit cost, for example "Auto-refresh on, 200 credits/mo". That figure is worth checking against your plan's credit allowance before you leave it running.
Next refresh states include "No scheduled refresh", "Flag prompts to schedule a refresh", "Next refresh due now", and a date.
What This is Not
It is not a ranking guarantee. Being mentioned once does not mean you always will be. AI answers vary between runs even for the same prompt.
It is not comprehensive. It samples specific prompts on specific platforms at specific times. It is a measurement, not a complete picture of everything an assistant might ever say.
It is not a lever you pull directly. You cannot submit your site to an AI assistant. Visibility comes from being the kind of source assistants draw on, which is what the GEO signals in your Content Score are about.
Notes
AI Visibility requires a website on the project. Without one you see "Connect your website to start tracking AI visibility".
Results can be partial. When some platforms are unavailable you are told metrics reflect only the completed checks.
The feature is enabled in production. On a local or staging environment without the flag set, the page shows clearly labelled Sample data described as "Illustrative results, not live measurements for this project."
Related
Tracking prompts and reading your results covers the practical steps. Understanding your content score explains the GEO signals that affect whether you get cited.