An insurance BPO serving 50+ clients ran its entire operation manually. Tasks arrived through email, client management systems, and PDFs — no single intake point, no standard process. PAs and SPAs processed work, QC reviewed it, AMs ran teams across multiple clients, and every layer above that built its own reports by hand for the layer above.
Knowing how many tasks had come in, how many were closed, pending, or stuck in QC was effectively guesswork. Assigning tasks ate up time on its own — and without a system to prevent it, the same task could be handed to two people by mistake.
Every management layer prepared its own report for the layer above, manually. Delays were built into the process, and numbers were sometimes simply wrong by the time they reached management.
Tasks reach the platform the way they always did — over email, through a client's management system, or as generated files — and land on an FTP location. The application reads each file based on where it lands: which client's folder, which sub-folder. From that, it identifies the client, the task type, and the relevant benchmark, then creates an entry automatically. No manual data entry.
Users only see tasks for the clients they're associated with. Picking up a task locks it — no one else can touch it, which closes off double-assignment at the source. Once marked complete, a task moves to QC if the workflow requires it. Every stage is tracked.
AMs, Operation Managers, and Senior Managers can see live task status — received, completed, pending, in QC — at whatever level of the hierarchy they sit. No chasing updates, no waiting for a report to come up from the layer below.
Reports can be pulled on demand — filtered by date range, client, or other criteria. Scheduled reports go out automatically by email. What used to take hours of manual compilation at every management layer is now minutes.
A 3-person full-stack team designed, built, and maintained the entire platform across a 5+ year engagement covering both the original build and ongoing maintenance — running on two separate Ubuntu servers, one for the application and one for the database.