What clients can see
The eight per-workspace switches deciding which sections an Agency Client sees, why absent means allowed, and the three surfaces that are ungoverned.
Each Client Workspace carries eight switches, one per section, deciding what its Agency Clients see. They govern the client experience only: your own team is never filtered by them.
The rule that shapes everything else on this page: an absent setting means allowed. A workspace you have not configured shows the client every section the product shows client logins at all, and an agency switches a section off rather than switching the rest on. (Chat is the one section client logins do not see in this release, whatever its switch says: see the table below.)
The eight
| Permission | What a client loses when it is off |
|---|---|
| Rank tracking | Google positions, AI Overviews and AI Overview Citations, and the whole Traffic section: Search Console, Analytics and Indexing |
| AI visibility | How their brand appears in AI assistant answers, the chat feed, AI Answer Citations, and AI Readiness |
| AI Shopping | Whether their products appear in AI shopping answers, the merchant list and the ads view |
| Content Studio | Briefs, articles, the content plan and calendar |
| Website audits | Technical audit results and issues |
| Reports | Shared performance reports |
| Tasks | The board of recommended actions |
| Chat | Nothing: Chat is for your own team and is not shown to client logins, so this switch denies nobody today. It lets you set the policy in advance: a workspace where you have switched it off stays off whatever changes later, rather than every client gaining a credit-spending assistant at once |
Two of those cover more than their name suggests, and both are worth reading before you switch one off.
Rank tracking also governs Traffic. Search Console reports average position per query, which is exactly what this permission promises to withhold. Leaving Traffic visible would have made switching rank tracking off a false promise.
AI visibility also governs AI Readiness. It is the same subject, what assistants can see of this brand, and giving it a separate switch would have grown the set without adding a decision.
Why absent means allowed
Because the safe direction is one-way.
Every Client Workspace that existed before a permission did behaves exactly as it did, and every workspace you have not configured shows everything. So a permission added to RankX AI in future cannot silently take a section away from a client already using it.
The cost of that choice sits with you: a workspace you have not configured shows every section a client can reach. Set the permissions before the first client user logs in.
Three surfaces are deliberately ungoverned
Keyword Research, Search Trends and Topic Clusters have no switch. The set is closed at eight by decision rather than by oversight, and adding a switch is a product change rather than a setting.
AI Prompt Research is NOT one of them. It sits beside the other two research surfaces in the rail but it is governed, by the AI visibility permission, because it feeds the prompt surface that permission already covers. Turning AI visibility off takes prompt research with it.
If any of the ungoverned three is genuinely sensitive for a client, keep that work in a workspace the client is not invited to.
A fourth surface, the overview landing page, is ungoverned for a mechanical reason instead: a client user has to be able to land somewhere, and every redirect in the product targets it.
Switching one off removes a whole group
The permissions largely follow the product's own navigation grouping, so switching one off usually removes a whole section from the client's rail. RankX AI drops a group whose items all disappear, so a client never sees a heading over an empty list.
Two groups are deliberately mixed, in opposite directions, and both keep the client's rail coherent.
Research holds AI Prompt Research, which the AI visibility permission governs because it feeds the prompt surface that permission already covers, beside Keyword Research and Search Trends, which are ungoverned. So switching AI visibility off leaves a two-row Research group rather than a stranded header.
Content holds Content Studio, which the content permission governs, beside Topic Clusters, which is ungoverned. So switching content off leaves the group heading with Topic Clusters under it, which is again a coherent group rather than an empty one.
The switch is a boundary, not a decoration
Hiding a section from the navigation is a convenience. RankX AI also guards the page, so a client typing the URL directly is refused rather than served.
That distinction is the difference between a permission and a preference, and it is why hiding a nav item is never the whole implementation.
A practical order
- Create the workspace and add the client's Websites.
- Set the eight permissions to what that client is paying for. Off is the deliberate act; leaving one alone shows it.
- Check the portal as the client will see it, before inviting anyone.
- Then invite the Agency Client users.
Where to go next
- The client portal, for what a client sees.
- Roles and permissions, for the role boundary these sit on top of.
- Client reports, which reach clients who never log in.
Last updated
Roles and permissions
The four RankX AI roles, what each can do, why only one can approve an MCP connection, and why Agency Clients do not consume seats.
Per-client credit budgets
A ceiling on the shared agency wallet, not a sub-wallet. Why the wallet can be full while a client is stopped, and what uncapped and zero each mean.