Public-Service Demand and Capacity Planning
Public agencies must provide enough capacity where and when people need a service. That task becomes difficult when eligibility rules, population movement, seasonal demand, policy changes and channel adoption alter both the volume and location of requests. A demand forecast is therefore useful only when management can translate it into a decision about staffing, service locations, appointments or digital capacity.
Historical demand describes yesterday's service design
Transaction and application records show the demand that reached the service. They do not show every person who needed the service but could not find a suitable location, appointment or channel. Historical volumes may also reflect earlier eligibility rules, opening hours and digital access rather than the population's underlying need.
Management must separate observed demand from unmet or displaced demand before using the history as a capacity target. Waiting times, abandoned applications, repeated contacts and movement between channels help reveal where the existing service design suppressed or redirected requests.
Policy and channel changes can break a stable forecast
A forecast based only on recent volumes may perform well until a policy changes the eligible population or a digital channel alters how people apply. The total number of requests may remain similar while demand moves between locations, times or service teams. That shift can create queues in one part of the system and unused capacity in another.
The analysis must therefore represent the events that can change demand. These events may include a new entitlement, a renewal cycle, a population increase, a school term, a public campaign or the introduction of online appointments. Analysts should estimate each effect at the level where management can act, such as location, channel, service type and time period.
Forecasts must remain coherent across the service network
National, regional, location and channel forecasts describe the same demand at different levels. If analysts estimate each forecast independently, the parts may not add to the total. Management could then approve staffing and location plans that cannot all be correct.
Hierarchical forecasting reconciles these levels. It produces a total forecast that remains consistent with the forecasts for regions, locations and channels. Prediction intervals also show the range that capacity may need to absorb rather than presenting one precise number as certain.
Capacity turns the forecast into an operating decision
Demand alone does not determine the plan. Each location and channel has operating hours, staff skills, appointment lengths, technology limits and service standards. Management must decide how much demand the system should absorb, how long people may wait and which capacity can move when actual demand differs from the forecast.
Marketways connects the demand model to these constraints. We compare staffing, opening-hour, appointment, location and channel options under the same demand scenarios. This comparison shows which plan protects access and where additional capacity merely moves a queue to another stage.
A policy change requires a new capacity test
Consider an agency that expands eligibility for a service while encouraging applicants to use a digital channel. Management cannot infer the new staffing requirement by increasing last year's volume by one percentage. The policy may attract new applicants, move existing applicants online and generate additional support contacts from people who cannot complete the digital process.
Marketways would estimate the eligible population, model take-up and channel movement, and test how each scenario changes applications, support requests and completion. We would then compare capacity plans for digital processing, assisted service and specialist review. The chosen plan would identify the demand assumptions that matter most and the trigger for revising capacity.
AI can keep the plan current, but it cannot set the service promise
An AI workflow can monitor actual demand, compare it with the forecast and prepare a capacity exception for review. It can also retrieve the relevant drivers and show whether the variance arose from location, channel, service type or timing. These steps reduce the time required to understand a developing queue.
However, the workflow should not decide who receives priority or reduce a service standard merely because capacity is constrained. Management must define the public-service objective, acceptable waiting time, protected groups and escalation authority. The AI workflow can then operate within that policy and preserve a record of each recommendation.
Success requires access, completion and sustainable capacity
Management should measure whether people can reach and complete the service within the intended time. Waiting time, abandonment, repeat contact, channel transfer and cost per completed service reveal different parts of that result. Utilisation matters as well, but maximum utilisation can make the service fragile when demand changes.
The final plan should state the demand range it can absorb, the assumptions behind that range and the conditions that require a new decision. This makes the forecast an accountable capacity policy rather than an annual number.
How Marketways develops the plan
Marketways defines the service promise, reconstructs observed and unmet demand, and estimates how policy, population, seasonality and channel use affect future requests. We then connect the forecast to staffing, location and digital constraints and test the plan under credible demand scenarios.
The engagement may combine Sales Forecasting and Demand Planning, Workforce Planning and Capacity and Forecasting and Time-Series Analysis. The related public-service delivery case study shows how several decisions can form one transformation programme.