Discovery and Scoping
Explore how to identify bottlenecks and define precise objectives in a public sector agency's document processing workflow. Understand how to trace application stages, use real data for discovery, and set measurable success criteria to improve efficiency and compliance while maintaining security and fairness.
Why start with faster extraction when applicants wait eleven weeks? The agency’s typing-heavy workload identifies a substantial use of caseworker effort, but discovery must establish what keeps applications open. The resulting decision record will connect the observed delay to a bounded pilot and measurements that protect applicants while testing service improvement.
Where does time go?
The fictional Housing Assistance Agency has an average processing time of 11 weeks, a backlog of 140,000 applications, and 600 caseworkers spending most of their time typing and checking document data. The minister’s target is processing within four weeks within a year.
Start discovery by tracing an application through the process:
Receipt: The agency receives the submitted packet.
Completeness checking: Someone establishes whether the required evidence is present and usable.
Evidence request: The agency requests missing or invalid supporting material.
Applicant response: The applicant obtains and supplies that material.
Resumed review: The case returns to review, including any further checking or correction.
Decision: A caseworker reaches and records the outcome.
For each stage, distinguish active work, time waiting in an agency queue, and time waiting for documents. A caseworker walkthrough explains what people do; process records help establish when it happens. Neither source alone necessarily gives a complete account.
This distinction changes the engineering question. Saving typing minutes measures an effort improvement. To explain a reduction in receipt-to-decision time, identify how the intervention changes work, queues, or document waiting along that application’s path. Keep stage durations unresolved until the agency supplies evidence.
Inspect submitted applications
Blank forms show intended structure. Submitted packets reveal handwriting, supporting-document variation, unreadable material, omissions, and the corrections workers actually make.
The agency’s discovery asks six questions: Can you inspect real applications rather than blank forms? Where do the eleven weeks go? What is the formal security classification? Which field errors cause harm? What must the audit trail contain? How will fairness be demonstrated publicly?
Turn those questions into evidence requests:
Submitted packets: Inspect what arrived, including attachments and page quality, rather than assuming every packet follows the template.
Process records: Locate receipt, checking, evidence requests, responses, resumed work, and decision events.
Caseworker walkthroughs: Explain ambiguous events, repeated checks, correction work, ...