Documentation/Features

Group mining and recommendations

ServiceChanger scans your tenant and suggests which groups belong to which attributes, cleans up, and flags drift.

What it does

Group mining analyzes your existing groups, memberships, and requests and suggests how to base them better on attributes. Instead of spending months figuring out which group belongs to which job function, you get concrete suggestions with the reasoning included.

Every recommendation comes with the numbers: how many users it affects, what share of the attribute already has the group, and the expected impact if you apply it.

The four categories

ServiceChanger groups recommendations into four categories.

Suggestions

Proposals to link a group to an attribute so that membership runs automatically instead of by hand. Examples:

  • A group with no attribute link whose members mostly share one attribute.
  • A group that is often requested through the Self-Service Portal by people with the same attribute.
  • A group with several attribute links where one broader attribute already covers nearly all members, so you can consolidate.
  • A group that is repeatedly granted as temporary access to people with the same attribute, a candidate to link permanently.

Cleanup

Proposals to tidy up: empty or unused groups, duplicate links, and memberships that no longer fit anything.

Drift

Signals that reality has diverged from your rules. For example users who are in a group but no longer have the matching attribute, or the other way around.

Quality

Quality signals about your attribute and group model, so your structure stays clean and understandable.

How a recommendation is formed

ServiceChanger looks at overlap between group membership and attributes and applies thresholds before anything shows up as a recommendation. A link suggestion only surfaces with enough users and enough coverage, so you do not get noise from accidental overlap. Consolidation suggestions explicitly show who would gain access and who would lose it, so you never accidentally link too broadly.

Applying a recommendation

Every recommendation has three possible actions:

  • Apply: ServiceChanger creates the suggested link or change.
  • Dismiss: you hide the recommendation. It does not return unless the situation changes.
  • Leave: you keep it for later.
An applied recommendation can be tested on a small test group first, just like any other rule, before you roll it out more broadly.

Related