Websites, workspaces and accounts

RankX AI has three levels of structure. An account is the thing that is billed, and it holds one credit wallet. A Client Workspace is one client business, and only agency accounts have more than one. A Website is one site you track, and it is the unit almost everything else hangs off.

Most of the product is Website-scoped. Almost everything else is account-scoped. Very little is scoped to a workspace, and knowing which is which explains most of the navigation.

The model

graph TD
  A[Account: billed, one credit wallet] --> W1[Client Workspace]
  A --> W2[Client Workspace]
  A --> WP[Your own workspace: never capped by a client budget]
  W1 --> S1[Website]
  W1 --> S2[Website]
  W2 --> S3[Website]
  S1 --> D[Prompts, keywords, audits, content, history]

A Direct account has exactly one workspace, its own, and the product hides the client dimension entirely rather than showing a level with one item in it.

A Website

One site you track. It carries a domain, a market, a brand profile and a competitor set, and it owns:

• Tracked AI visibility prompts and their check history.
• Tracked keywords, their positions and their AI Overview captures.
• AI Shopping prompts and the product catalogue.
• Website Audits and their findings.
• Content: briefs, articles, campaigns.
• Tasks.
• Its integrations: WordPress, Search Console, Analytics.

If a screen asks which Website you mean, that is why. Almost every measurement in RankX AI is a measurement of one site.

A Client Workspace

One client business. It exists on the agency track and holds that client's Websites, its visibility permissions, its credit ceiling, its report schedules and its Agency Client users.

It is the unit you report on and share. See Client Workspaces.

Every agency also has its own workspace, which is not a client and is never capped by a client budget.

An account

The thing that is billed. It holds:

• One credit wallet. There is never more than one, on either track.
• The plan, and every structural limit that comes with it.
• Staff members and seats.
• MCP credentials.
• Email sending settings and, on the agency track, white label.

What is scoped where

Settings in RankX AI are organised by exactly this, in three bands by blast radius: website settings require you to pick a Website, organisation settings never do, and account settings are about you. A settings section never mixes two bands, which is why the picker appears on some screens and not others.

The vocabulary, used consistently

These docs use one word per thing, and the product does too:

The last two are retired terms rather than style preferences. The abbreviation leaked into a page title, a chart heading and four stat labels before it was caught, and "Citation Sources" sat two navigation groups from "AI Overview Citations" telling a reader nothing about which engine either came from.

Limits follow the model

Which is why they read the way they do:

• Websites are counted per account. On the agency track they are a pool distributed across workspaces however you like.
• Client Workspaces are counted per account.
• Tracked keywords and prompts are capped per Website, so a plan allowing 300 keywords allows 300 on each Website rather than 300 in total.
• Seats count staff on the account. Agency Clients do not consume one.
• Credits are one balance per account.

See plans and limits.

Where to go next

• Choosing your track, if you are deciding.
• Client Workspaces, for the agency layer.
• Plans and limits, for the numbers.