We use activity chaining when:

The same user will perform multiple tasks in immediate succession (often multiple UI forms).
You want a seamless flow (a wizard-style experience) so the user doesn’t go back out to their Task Inbox between steps.
The tasks are lightweight and fast (minimal delay between them).

Example: A user begins a form to enter customer details, then right away continues to the next step of uploading supporting documents and then reviewing the entry. By enabling activity chaining, after they submit the first step they automatically go to the next form without needing to find a new task.

When Not to Use Activity Chaining

You avoid activity chaining when:

The next task is for a different user (so chaining would force the user to continue when they should not).
There is expected time delay or background processing between steps (e.g., generating a report, waiting for system-side work).
The tasks are not meant to be a continuous UI sequence (for example: employee → manager approval → external vendor).
The chain would exceed performance/maintenance limits (for instance, too many nodes chained, or long waits).

Example: An employee submits a request. After that, the request goes to a manager for approval which may happen hours later. Chaining is not used here because the manager is a different user and there is a time gap.