How can you tell whether a digital change actually worked?

Write down the baseline before changing it

Choose a small number of observations tied to the reason for investing: time to resolve a case, avoidable rework, the share of enquiries answered or the number of disputed records. Record how each is measured and what it looked like before launch.

Usage is useful context but not a benefit by itself. A tool may be heavily used because people have no alternative, while the original delay remains.

Look for work that moved elsewhere

Speak with customers and staff, examine awkward cases and check whether a new spreadsheet or inbox has appeared. Improvements in one team's measure can conceal extra work for another.

Decide in advance when to review the result and who can change the design. If the evidence is mixed, refine the process or stop an unhelpful feature rather than defending the launch.