Dashboard Adoption Failure in a Clinical Operations Rollout
A hospital analytics team launched a bed-capacity dashboard for charge nurses across 12 wards. Six weeks after launch, only 23% of nurses had logged in during the past week, and median session length was 40 seconds. The dashboard refreshes every 15 minutes and shows 14 charts, including occupancy trends, discharge projections, and staffing ratios. Nurses say the numbers often disagree with what they see on the ward, and that finding the one figure they need takes too many clicks. The team also learned that the discharge projection model was trained on pre-pandemic data and that ward-level occupancy is computed from admissions timestamps that lag by up to two hours. Leadership wants a usage target of 70% within a quarter and asks whether to add features, run training, or rebuild.
1. Diagnose the adoption problem using the evidence, separating data-quality, design, and organizational causes.
Read the worked answer
Data quality: the two-hour timestamp lag means displayed occupancy is stale, and the pre-pandemic discharge model may be miscalibrated, so nurses see numbers that conflict with the ward. That erodes trust faster than any design flaw. Design: 14 charts with a 40-second median session suggests nurses cannot find the needed figure quickly; the dashboard is organized around analyst interests, not the nurse's task. Organizational: 23% weekly login indicates the tool is not embedded in workflow, and no role or moment is clearly served. Adding features would worsen overload, and generic training will not fix untrustworthy numbers. The likely root cause is that the dashboard answers a question nurses do not ask at the point of care, with data too stale to be credible. Rebuilding around a small number of trusted, timely indicators is the more defensible direction.
2. Recommend a redesign and data-remediation plan, prioritizing actions and stating what you would measure.
Read the worked answer
Prioritize trust first: fix or clearly label the timestamp lag, showing data freshness and last-updated time; retrain or replace the discharge projection on recent data and display uncertainty; reconcile displayed occupancy against ward counts with nurses to find systematic gaps. Then redesign around tasks: identify the two or three decisions nurses make (for example, which patients can be discharged today, where to place an admission), and surface those as the top of the screen with drill-down, cutting the 14 charts to a focused set. Reduce clicks by defaulting to the nurse's own ward. Embed in workflow, for example a morning huddle view or a link from the bed-management system. Measure weekly active use by ward, time-to-answer for the core question, data-freshness indicators, and a trust survey item; track whether nurses override or ignore displayed numbers.
3. Leadership sets a 70% usage target within a quarter. How would you evaluate whether that target is appropriate and whether the rollout is succeeding?
Read the worked answer
A usage target is only meaningful if the tool supports a recurring task. I would first confirm with nurses which decisions recur daily and whether the redesigned view serves them; if it does not, 70% is arbitrary and may drive gaming such as auto-login or mandatory checks. I would define usage as active engagement with the core view during relevant shifts, not raw logins, and pair it with outcome and quality measures: time to place admissions, avoided bed-crisis escalations, data-discrepancy reports, and trust ratings. A staged goal is more realistic: trust and data-accuracy fixes in weeks 1-4, pilot on two wards in weeks 5-8, then scale if time-to-answer and trust improve. If usage rises while nurses still report wrong numbers, the target is being met superficially. Success is better decisions and credible data, with usage as a supporting signal.