Staffing studies and reviews

Define the service decision before calculating positions

Consider an illustrative administrative service with a growing backlog. One team proposes an additional position based on last year’s average processing time. Another proposes automation. A third argues that more complete applications would remove much of the workload. All three could be plausible, but none is established by the size of the backlog alone.

A workforce plan implicitly makes choices about service levels, work allocation and risk. Two models can produce different staffing requirements because one assumes a shorter turnaround, a different mix of cases or more automation. If those assumptions are not visible, leadership may debate the number while leaving the consequential service decision unexamined.

The useful output is therefore a decision model with explicit assumptions and alternatives. It should show what staffing can achieve under specified conditions, which activities require scarce skills and what will happen if recruitment or productivity improvements arrive later than planned. Precision in the spreadsheet does not remove uncertainty in its premises.

What workforce audits and technical methods can establish

The National Audit Office’s 22 March 2024 audit of NHS England’s workforce modelling found good-quality central Python code and replicated that component’s outputs. It could not fully replicate the published results because later stages relied on complex, undocumented manual adjustments. It also found that the proposed expansion in training had not been accompanied by an assessment of training-capacity constraints and the costs of overcoming them. This was a review of a particular English model, not a Canadian staffing benchmark. It demonstrates why technically correct calculations and a feasible workforce transition are separate questions.

WHO’s 23 November 2023 Workload Indicators of Staffing Need manual, second edition applies activity-time standards to health-worker workload components. It provides a workload-based planning method, not proof that a historical activity count captures all unmet need or a ready-made staffing ratio for another sector. Local standards, service scope and data quality still need examination.

NHS England and NHS Improvement’s May 2019 core-model technical guidance models weekly demand, available capacity, critical resources, unavailability and waiting-list scenarios. Its stated scope is suitable booked elective-care pathways, not all operations. The useful distinction is between enough total hours and capacity available at the time and stage where the service needs it. Arcus’s diagnostic below applies that distinction as management synthesis, not as a validated cross-sector clinical method.

Why an average staffing calculation can still miss the bottleneck

A transparent starting relationship is workload hours = the sum, across work types, of volume multiplied by representative effort. Compare those hours with the deployable time of the required roles. Keep direct work, essential support activities and time unavailable for service distinct to prevent double counting. This is an accounting model, not a universal staffing formula.

Now compare two operating choices. An additional generalist may increase total processing hours but leave a specialist approval queue unchanged. Cross-training or changing the approval pathway may address that queue, but requires learning time, supervision and authority; it may also remove capacity from another service. If the specialist’s available hours are consistently full while work accumulates immediately before that stage, the second explanation deserves testing. If it is waiting for incomplete upstream work, more specialist capacity may not solve the problem.

Likewise, equal annual demand and capacity do not show whether a service can meet peak-period commitments. Examine arrivals, case mix and deployable skills over the relevant service window. Flexible coverage can address peaks but costs more or depends on scarce availability; permanent capacity may be justified for recurring pressure but creates an ongoing commitment. There is no universal utilization target that resolves this trade-off.

Treat service need, staffing requirement and affordability as separate outputs. Historical throughput can be constrained by the very shortage being studied. Conversely, a stated aspiration is not automatically authorized demand. Leadership must decide the service scope and acceptable waiting or coverage risk before the staffing recommendation can be defensible.

Separate a funded workforce plan from a conditional productivity case

Compare a baseline using current demonstrated practices, a feasible improvement case and a downside case in which recruitment or implementation is delayed. For each, state the service level, skill mix, backlog treatment, funding and dependencies. Do not assume that training expands supply without consuming supervision, or that a tool reduces workload before exception handling has been observed.

Use a cost bridge: recurring people and operating costs, plus one-time recruitment, learning, backfill, process and technology transition costs, less expenditure genuinely avoidable. Report released hours separately from cash savings. A lower modelled staffing requirement does not establish an immediately achievable budget reduction; positions, work allocation and service commitments may not be divisible.

Assign the operating leader responsibility for service and quality assumptions, HR for supply and capability inputs, finance for affordability, and a named analytical owner for the model and change log. Leadership resolves conflicting service priorities. These are proposed engagement responsibilities, to be adapted to the client’s authorities.

At each gate, ask whether an assumption has become observed performance. If recruitment slips, the fallback may be temporary qualified support or phased commitments. If rework falls without harming quality, leadership can decide how to redeploy released capacity. Set the evidence period and trigger with the client; aggregate modelling should not substitute for appropriate individual employment processes.

What the client should receive and be able to maintain

An Arcus engagement can produce a service-and-workload baseline, a role-and-time model with an assumption register, competing-cause tests, option costs and capacity implications, a transition dependency map and decision triggers. The client should be able to reproduce the calculation, identify which inputs are observed or judgement-based and refresh the plan without dependence on the original analyst.

Bring the service standard, recent demand and backlog records, funded and filled positions, relevant deployment information and the disputed assumptions to the scoping discussion. Where data are weak, a targeted evidence-collection phase may be more useful than a falsely precise staffing recommendation. Related organizational-development work can test capability and operating conditions; capacity planning addresses the wider service-resource decision.

What leadership should examine before approving the number

Ask which service promise the plan supports, which evidence establishes the effort assumptions and what remains unknown. Determine whether funding, recruitment or a process dependency is likely to constrain delivery first. Review how the plan changes if a material assumption is wrong, and who must act in that situation.

The review should identify work that can be prioritized, redesigned or stopped, as well as work requiring additional people. It should make an unresourced service commitment visible rather than imply that every gap can be solved through employee efficiency.

Independent support is most useful when departments disagree about the source of staffing pressure, a consequential business case needs a transparent rationale or the organization requires a model it can maintain. Arcus expertise and project responsibilities are addressed through the central consultant directory.