Physician study splits failed digital rollouts into a data problem and a design problem

Design was expected to move cognitive load in the opposite direction. The researchers hypothesized that usable systems cut extraneous load, the effort lost to excess navigation steps, misaligned workflows, alert fatigue and visual clutter ^[11]. Better data adds effort, but of the useful kind. "Better data makes physicians think harder about what matters," the article's author writes ^[4]. One score for user effort or satisfaction can therefore hide a working data fix behind a rise in effort. It can also credit a design fix that left the data as poor as before ^[14].

The resistance diagnosis persists because the two faults are rarely pulled apart. When a digital investment underdelivers, organisations almost universally blame implementation problems, insufficient training or change management breakdowns, according to the article ^[8]. "These explanations are structurally inevitable when leaders don't distinguish between a data quality failure and a system design failure," the author writes ^[9]. I'd expect training to stay the favoured remedy for a plainer reason too. It can be bought without reopening the data pipeline or the vendor's interface.

The cost shows up in the next budget cycle. The article describes a corrective investment aimed at the wrong problem, followed by another underperforming system and another misdiagnosis ^[15]. The article does not estimate what that loop costs. For a leader holding a stalled rollout this quarter, the trade-off is speed against accuracy. Splitting the postmortem into a data question and a design question takes longer and may need two owners, because, in the article's words, "improving one does not improve the other" ^[7]. Skipping the split is faster. It also sends next quarter's money toward training, the explanation the article says follows from never drawing the distinction ^[8][9].

A skeptic would say a survey of doctors about health records is thin evidence for a sales floor or a warehouse. The researchers chose clinical medicine because it is the highest-stakes and most extensively documented setting for knowledge work under time pressure, and the article argues the cognitive mechanisms are not unique to it ^[12]. That argument belongs to the authors. The evidence comes from one profession and one class of software ^[10][13]. The article sums up the result in a single line: the research "confirmed both effects with statistical precision" ^[3]. I would treat the separation of data and design as a sound diagnostic habit for any rollout, and the size of each effect outside medicine as unknown.