Feasibility

Feasibility concerns whether a proposed change can be carried out under its actual constraints. For a data or AI use case, technical possibility is only part of the assessment. Required information, permitted use, integration, staff capacity, and maintenance also matter.

Available files may not be usable inputs

A fictional team has years of invoices but cannot reliably link their lines to receipt records. The file count does not establish that an evidence-lookup assistant is feasible. Check identifiers, timing, access, and how unresolved matches will be handled.

Identify the limiting uncertainty

Ask who will act on the output, correct errors, and maintain sources. A technically correct result in a disconnected screen may add re-entry work. A small permitted sample can investigate a key uncertainty before a full build. Feasibility does not establish that the expected benefit justifies the cost or that the change is ready to launch.

Reference: GOV.UK: User Research in Discovery.