The mandated system rollout
Sam is director of adult social care at a local authority, leading the rollout of a new case-management system to 340 staff across 12 teams. A six-month pilot in two teams cut report-writing time by 35% but also showed a 19% rise in data-entry errors in the first eight weeks and a 6% drop in face-to-face visits. The vendor contract has a fixed go-live in five months, with £1.2m of implementation costs already committed and a further £400,000 payable on full rollout. Staff survey scores on 'I am consulted about change' fell from 62% to 41%. Two team managers have asked for a delay; one has said her team will work to rule if go-live proceeds as planned. The chief executive has publicly committed to the system as part of a digital transformation programme. A neighbouring authority abandoned a similar rollout last year after cost overruns.
1. Assess readiness. Should Sam proceed, pause or change pace, and what evidence should drive that judgement?
Read the worked answer
The pilot results are not one result; they point in two directions.
- A 35% cut in report-writing time is a genuine and material benefit; it is the reason the programme exists.
- The 19% error rise needs interrogation before it is treated as a failure. If it fell back to baseline by week 16, it is a learning curve. If it is still 19% at week 24, it is a design or training problem. Sam should audit a sample of errors and find the root cause.
- The 6% fall in face-to-face visits is the most serious signal. If it is real, the system is displacing the service rather than supporting it, and no efficiency gain justifies that. If it is a recording artefact - visits happening but logged differently - that must be proven, not assumed.
The 62% to 41% fall in 'I am consulted' is a leading indicator of adoption failure. Staff who feel unheard do not sabotage; they comply minimally, and the benefits case quietly evaporates.
Sunk cost is not an argument. £1.2m is spent whether or not the rollout proceeds; the decision must rest on future cost and benefit. The neighbouring authority's failure is relevant not as a warning against change but as evidence that cost overruns cluster where adoption was assumed rather than managed.
Recommendation: change pace, not direction. Proceed with a staged rollout and formal stop/go gates, rather than a single go-live date. Pausing entirely five months out would strand committed cost and damage credibility with the chief executive, but a fixed date with unresolved adoption risk is the more expensive option. Set explicit re-entry tests: error rate at or below baseline, visit volumes protected, and staff confidence recovering.
2. Recommend a plan for the next five months, including how Sam should handle the two objecting team managers and the work-to-rule threat.
Read the worked answer
Fix the evidence first. Commission a rapid review of the 19% error rise and the visit drop, and publish the findings internally, including where the system or the implementation is at fault. Nothing rebuilds trust faster than a leader who reports inconvenient data honestly.
Do not isolate the two managers. They hold real information and informal influence. Give them defined roles - one as a testing lead for the next wave, one as a peer champion with a mandate to raise issues - and a structured route to escalate concerns. Treating their objection as resistance to be overcome confirms their fear and guarantees the work-to-rule escalates.
On the work-to-rule threat, meet staff and union representatives quickly and separately from any disciplinary framing. Distinguish two things: what is negotiable (sequencing, training time, protected visit hours, support levels) and what is not (that the system will be implemented). A threat withdrawn in exchange for a genuine scheduling concession is a better outcome than a grievance.
Protect the service. Ring-fence visit time with an explicit rule that system use does not displace face-to-face work, and measure visits monthly. Reduce caseloads during each team's transition fortnight.
Build capability and capacity. Identify super-users in each team, schedule floor-walking support, and provide protected learning time rather than expecting staff to learn in gaps.
Sequence by readiness. Roll out in waves, starting where managers are committed and support is strongest, so later teams see working proof rather than a mandate. Hold monthly gates with the chief executive, and report adoption and service-quality measures alongside delivery milestones.
3. How should Sam preserve the benefits case and learn from the error rise? What should be measured and reported?
Read the worked answer
A benefits case that is not measured becomes a claim. The original case should be restated as baselines and targets with owners: report-writing time per case, error and rework rate, face-to-face visits per week, cost per case and staff confidence.
Measure leading and lagging indicators together. Lagging: error rate, visits, report-writing time, cost per case. Leading: percentage of staff trained, system logins and task completion, support tickets raised and resolved, super-user cover per team, and the 'I am consulted' survey item. Leading measures give early warning while there is still time to act.
Set thresholds in advance. For example, if a team's error rate has not returned to baseline within 12 weeks of its go-live, that team pauses and the issue is reviewed; if visits fall more than 3% for two consecutive months, the rollout stops pending a service review. Pre-agreed thresholds prevent the debate becoming about optimism versus pessimism.
Use the error rise as a learning asset. Ask what the errors were, where they occurred, and whether the system, the training or the process design caused them. Feed the answers into the remaining waves and the training package, and log them in a lessons register that the programme board reviews.
Commercially, act now. Check whether the £400,000 payment can be linked to quality or adoption milestones. If the contract allows, that converts a vendor risk into a shared one.
Report honestly to the chief executive, including the possibility that the original benefits case needs restating. A director who reports a 35% gain and hides a 6% service drop is not protecting the programme; they are storing up a much larger failure.