Conflict-of-Interest Auto-Detection: Why Manual Routing Isn't Enough
A report that ends up routed to the accused person's own manager - or, in a smaller org, to someone close enough to the accused that impartiality is genuinely in question - isn't a hypothetical edge case. It's what manual routing eventually produces at scale, especially in organizations where the same small group of people wears multiple hats: department head, case handler, occasional company admin covering for someone on leave.
Nobody sets out to route a case to a conflicted person. It happens because routing rules are usually built around category and department, and nobody's cross-checking every assignment against every accused person's actual working relationships in real time.
What the check actually does
When a case is created, the routing logic doesn't just match category and department to find a handler - it checks whether the accused person, based on the department and role information captured in the questionnaire, matches the department and role of any Case Handler or Company Admin who would otherwise receive the case. If there's a match, the system doesn't route it to that person and hope someone catches the conflict later. It sets the case status to needs manual assignment and notifies the Super Admin directly - metadata only, no case content included in that notification, so the conflict gets surfaced without exposing sensitive details to someone who doesn't need them yet.
Why this has to be automatic, not a step in a checklist
A step in a checklist relies on someone remembering to do it, every time, under time pressure, for cases that might not obviously look conflicted at first glance - the connection between "accused person" and "person about to receive this case" isn't always visible unless someone specifically checks. Automatic detection doesn't rely on anyone remembering. It runs the same check on every case, every time, without fatigue or a bad week making it skip a step.
What this deliberately doesn't do yet
Worth being direct about the current scope: this is a v1 mechanism built around department and role matching, not a full conflict-of-interest management system. It catches the case where an accused person's department and role overlaps with a handler or admin who'd otherwise get the case - it doesn't yet catch more subtle conflicts, like a handler who has an undisclosed personal relationship with the accused outside the org chart. A dedicated External Reviewer workflow for more complex conflict-of-interest scenarios is a known gap, planned for a later phase, not something the current system claims to solve.
Why the manual fallback matters as much as the detection
Auto-detection without a functioning fallback is just automatic rejection with nowhere for the case to go. The Super Admin notification exists specifically so a flagged case doesn't sit invisibly in a "needs manual assignment" state indefinitely - someone with appropriate authority and no case-content visibility gets alerted that a routing decision needs a human judgment call, and can make that assignment deliberately instead of by default.
The broader point
Conflict-of-interest risk in workplace investigations is rarely dramatic - it's usually structural and mundane, the natural result of small organizations where the same people fill multiple roles. A system that only catches conflicts a careful human happens to notice will miss the boring, structural ones most often. Automatic detection exists specifically to catch the boring case, every time, not just the obvious one.