The Ward That Wasn't in the Room
St Alban's Hospital, 900 beds, is automating linen and sterile-supply movement with 14 automated guided vehicles across three service lifts and two shifts. The project held workshops with facilities, IT and nursing directors. The porters' union was not invited; infection control sent written comments only. A junior analyst spent two days in contextual inquiry and observed porters using an unlisted service corridor and manually holding lift doors open, defeating the AGV dispatch logic. The vendor's throughput model assumes a mean lift wait of four minutes; the analyst measured nine minutes at the 07:00 peak. Go-live is eight weeks away and the business case depends on releasing six porter full-time equivalents. The union has heard rumours and has asked for a meeting. Infection control has not signed off the AGV cleaning regime.
1. Diagnose the stakeholder engagement failures and propose an engagement plan that fits the eight-week window.
Read the worked answer
The failure is not a missing invite list; it is that the stakeholder set was defined by organisational convenience rather than by who can affect or be affected by the system. Porters are primary users and their union has high power and high interest, so they belong in design, not communication. Infection control is a veto holder whose sign-off gates go-live but who was treated as a commenter. Nursing directors were consulted but ward nurses, who will load and unload the AGVs at peak, were not.
Within eight weeks: map stakeholders by influence and impact, then meet the union in week one, before any announcement about roles. Run a joint walk-through with porters and infection control to validate the observed workarounds. Add a porter representative to the weekly design review as an observer with a route to raise issues. Give infection control a dated review of the cleaning regime with evidence. Publish a short, plain communication plan covering what changes, what does not, and how concerns are handled. Feed each engagement output into the ConOps and requirements log so the activity leaves evidence, not just meeting notes.
2. How should the contextual inquiry findings be used, particularly the nine-minute lift wait?
Read the worked answer
Treat the observation as requirements data, not as an anomaly to be averaged away. First test its quality: two days at peak is a small sample, so instrument the lifts for two weeks to establish the distribution, not just a mean. The workaround behaviours are equally important: the unlisted corridor and held doors reveal that the current process depends on human judgement the AGV design did not model.
Update the throughput model with the measured distribution and re-run the business case. If throughput falls below the target, the options are operational and architectural: lift priority for AGVs, a dedicated service lift, retimed linen deliveries, or retaining porters for exception handling. Do not simply add AGVs, because the constraint is lift capacity. Present the revised model with its assumptions to the business case owner. If six FTE releases no longer hold, that is a finding to surface now, not after go-live.
3. What should the SE lead advise about the union conversation and the workforce implications?
Read the worked answer
Advise transparency and sequencing. The union should hear from the project before it hears a rumour, and the message should distinguish role redesign from headcount. The business case may assume six FTE releases, but presenting that as a settled outcome before consultation turns a design conversation into an industrial dispute and encourages workarounds to be hidden.
Propose a joint session covering what the AGVs will and will not do, which tasks remain manual, and how porters will be involved in exception handling and validation. Avoid individual monitoring data being used for performance management; if AGV data can track people, say explicitly how it will not be used. Record workforce assumptions as project risks with owners. The SE lead should also tell the project board that industrial risk can delay go-live more than a technical defect, so consultation is schedule-critical, not a courtesy.