Skip to content

AI Visibility, and why the page is read-only

What an AI Visibility capture actually measures, why there is no visibility score, and the plain fact that tracking a prompt and running a capture are not built.

Updated

AI Visibility records whether your brand was actually named in an answer to a question, and shows only what a real capture measured. There is no score, no ranking and no estimate anywhere on it.

The page is read-only, and it says so itself. Tracking a new prompt, pausing or resuming one, and running a capture have no control here, not even a disabled one, because there is no route behind any of them in this build. What the page shows is every prompt and capture that already exists. In practice that means a customer who has not had rows created some other way will see an empty page, and that is the honest state rather than a fault.

It is also worth being precise about what a capture is, because the name invites a wrong assumption. A capture sends the tracked question through the same AI text gateway the product uses for writing, and then looks for your brand's terms in the answer as a plain, literal text match. It is not a search engine crawl, not a search results page, and not a measurement of any answer engine other than the one configured here. Nothing in this product measures search rankings, answer-engine optimisation or social search.

BEFORE YOU START

  • Access to the workspace and active membership of the organisation. There is no separate permission on this page.
  • An AI provider configured in this deployment, or no capture can ever have run.
  • Existing tracked prompts. There is no way to create one from either client.

STEPS

  1. Open AI Visibility for the workspace, or go to /app/workspaces/{workspace}/ai-visibility.
  2. Read the note headed "This page is read-only". It names the three controls that do not exist.
  3. Read the summary line. It says either that no provider is configured, or that no capture has run yet, or how many completed captures mentioned the brand out of how many.
  4. Read each tracked prompt, its Active or Paused state, and the terms it is watching for.
  5. Read the captures under a prompt. Each is Measured, Pending, Capture failed or Provider not configured, and each carries Mentioned, Not mentioned or Not measured.

WHAT YOU SHOULD SEE

Most likely "No prompts tracked yet." That is the accurate state of this capability, not a loading failure.

WHAT THIS WILL NOT DO

  • It will not track a new prompt, pause one or run a capture. No control exists and no configuration creates one.
  • It will not produce a score or a percentage. The captures table has no score column by design.
  • It will not treat a failure as an absence. A capture that failed, or that ran with no provider configured, is recorded as "Not measured" and never counted as your brand being absent.
  • It will not compare you with a competitor. There is no competitor list anywhere in this product.
  • It will not measure search engines. A capture is one model's answer, matched literally.

WORTH KNOWING

Tracked prompts belong to the organisation, not the workspace, even though the page is reached through a workspace. Every workspace in the organisation therefore shows the same prompts.

IF IT DOES NOT WORK

  • "No AI provider is configured in this environment, so no capture has ever run." is a deployment state and nothing you can change in the product.
  • "No captures yet. 3 prompts tracked, none captured yet." means prompts exist and nothing has run against them.
  • "No prompts tracked yet." means there is nothing to show and no way from here to add anything.
  • Being sent back to /app means you are not an active member of the organisation. Being sent to the workspaces list means your access does not reach this workspace.

COMMON QUESTIONS

Is this SEO reporting?

No. Nothing here reads a search engine, a results page or an index.

When will the controls arrive?

That is not something this guide can promise. What can be stated is what exists today: the reading surface and the capture engine, with no route to start one.

Why show the page at all?

Because the captures it can show are real measurements, and because the alternative was shipping buttons that fail.

  • Reading your reporting

    What the workspace report totals, why a figure can say "Not reported" instead of zero, and the two ways a figure reaches it.

  • What your recorded results support

    Reading the performance recommendations, the sample size every claim must clear, and the list of questions the recorded data cannot answer.

  • Building your Command Centre

    Arranging the widgets on your Today page, what each panel counts, and the ten widgets that are declared and cannot be added because nothing computes them.