File Management System Overhaul
Leading a team of fourteen to clear a backlog of over a million physical refugee files and restructure the archive — an operations and team-leadership problem more than a technical one.
Screenshots of live humanitarian systems are not published. Where visuals appear, they are illustrative rebuilds created with synthetic data that reproduce the structure and design of the original work without exposing any operational or personal information — on data-protection grounds.
Context
The physical archive holds a file for every registered family, indexed by date and case number — effectively a library serving a population of 1.2 million people. A large backlog had built up, and retrieval delays were holding up registration appointments and protection casework downstream.
A file that cannot be found on the day is, operationally, a file that does not exist. That is the real cost.
Approach
I led a team of 5 filing staff and 9 filing enumerators and ran backlog clearance as a managed workstream alongside live intake — the archive could not pause while we fixed it.
- Redesigned the index. Restructured indexing and inventory around date and case number so that location is derivable from the identifier, not from memory.
- Real-time tracking. Introduced structured tracking through the File Management System, so a file’s status is recorded at the moment it moves rather than reconciled later.
- Two streams, one team. Split capacity between backlog clearance and same-day intake, with daily targets for each, so the backlog shrank without new arrears forming behind it.
How it went
Backlog clearance, 2019–2023
Illustrative trajectory — the start and end points are real, the path between them is synthetic
Outcome
1M+ files cleared
The backlog of over one million files was fully cleared between 2019 and 2023, and structured tracking has kept retrieval delays at zero since. Registration and protection colleagues stopped having to plan around whether a file would turn up.
What I’d do differently
This was solved as a process problem because that was what the situation allowed. The durable answer is digitisation with indexed scanning, which removes the retrieval question altogether — but it carries data-protection obligations that a physical archive does not, and those would need designing first, not bolting on afterwards.