Why does operational software stay bad for so long?
Because nobody with authority uses it. The people who suffer a poor interface are usually several levels below whoever approves the budget, and their complaints arrive as vague dissatisfaction rather than a business case. Meanwhile the system works, in the sense that transactions complete, so the cost stays invisible. It shows up instead as training time, error rates, support tickets, and staff turnover, none of which get attributed to design. Building that case is often the first useful thing an agency does.
How do we measure whether operational UX work paid off?
Not with satisfaction scores. Agree on task completion time, error and rework rates, support ticket volume, and time to competence for new staff before the project starts, and take a baseline. These are all measurable with instrumentation you probably already have, and unlike consumer metrics they translate directly into money. An agency that cannot propose a measurement plan for operational work has not done much of it.
Our system is twenty years old. Do we redesign or replace?
Usually neither in one move. Full replacement is where operational modernisation projects fail, because a decade of process logic sits in the old system and no one has documented it. The workable path is a UX audit to find where the time actually goes, then targeted redesign of the highest-frequency workflows while the underlying system stays put. That produces measurable returns within months and generates the documentation a replacement would eventually need anyway.
Should the agency also build it?
More often than in other categories. Operational interfaces are constrained by data models and integrations that are expensive to change, so a design produced without engineering input frequently turns out to be unbuildable as drawn. Either hire a partner with engineering in the room, or ensure your own engineers are involved from the first workshop rather than at handoff. The specification-over-the-wall model fails here more reliably than anywhere else.
How do we get access to the people who actually use the system?
Plan it as a named project cost. Operational users are on shifts, on floors, and in vehicles, not at desks, and shadowing them requires scheduling, supervisor cooperation, and sometimes safety clearance. This is the most common cause of delay in operational projects and the most common corner cut. Any agency that proposes designing from stakeholder interviews alone is designing from a description of the work rather than the work.
What does it cost and how long does it take?
A UX audit on a defined system commonly starts in the tens of thousands and delivers in four to eight weeks. Research and redesign of a core workflow runs twelve to twenty weeks. Full platform modernisation is a multi-quarter programme delivered in phases, and should be. European studios sit materially below US consultancy rates for comparable work, which matters on engagements measured in quarters rather than weeks.
