Demand Forecasting: How to Predict Your Staffing Needs and Build Smarter Rotas

Demand forecasting in workforce management is the practice of using historical attendance data, seasonal patterns, and known future variables to predict staffing requirements before the rota is built. Organisations that do it well spend less on agency cover, run fewer understaffed shifts, and carry lower labour costs than those that schedule reactively.

Most rotas are built backwards. A manager looks at who is available, fits them into the available shifts, and hopes the coverage matches the need. If it does not, the gap is filled with overtime, a phone call to an agency, or a shift that runs short-staffed and hopes for the best.

This is reactive scheduling. It is the industry default in care and hospitality, and it is expensive. Labour costs are typically the single largest controllable expense in both sectors. Scheduling reactively means those costs are determined by availability rather than by need, with no mechanism to anticipate pressure points before they arrive.

Demand forecasting is the alternative. It does not require a data science team or a sophisticated AI system. It requires a clear framework, consistent data, and a scheduling tool that surfaces the patterns already buried in your historical rota records.

This article provides that framework.

What Is Demand Forecasting in the Context of Workforce Scheduling?

Demand forecasting for staffing means predicting how many people, with what skills, are needed at what times, based on evidence rather than intuition.

In manufacturing or retail, demand forecasting is well established. Output targets, footfall data, and transaction volumes drive staffing models with reasonable precision. In care and hospitality, the variables are different but the principle is identical:

• In a care home, demand is driven by resident dependency levels, care plan requirements, and CQC-mandated staff-to-resident ratios. These variables change slowly and are highly predictable.
• In a hospitality venue, demand is driven by reservations, covers, events, and seasonal footfall. These variables change quickly but follow patterns that repeat year on year.
• In home care, demand is driven by client visit schedules, which are contracted in advance and largely fixed, but fluctuate around staff absence, client hospital admissions, and review periods.

 
In all three cases, the data required to forecast demand accurately already exists. The question is whether the organisation is capturing it systematically and using it to build rotas prospectively rather than reactively.

Why Historical Data Is the Foundation of Every Accurate Forecast

The rota as a data asset

Every rota your organisation has ever published is a record of what you believed you needed at that point in time. Every timesheet is a record of what you actually used. The gap between the two, tracked consistently over time, is one of the most valuable operational datasets a manager can have.

Organisations that treat historical rota data as a disposable administrative record discard this asset every time a rota period ends. Organisations that retain, analyse, and act on it gain a compounding advantage: each rota cycle adds more data, and the accuracy of future forecasts improves with every iteration.

The specific historical data points that drive useful staffing forecasts include:

• Shift-level attendance records: Which shifts ran at full capacity? Which ran short? Which had surplus staff that added cost without adding value?
• Absence patterns by day and period: Is Monday morning consistently your highest absence day? Does absence spike in January and August? These patterns repeat.
• Agency usage by time period: If you consistently call agency cover for the same shift type in the same week of the month, the data is telling you something about a structural gap in your permanent staffing model.
• Overtime frequency and distribution: Repeated overtime for the same staff members on the same shift types indicates either a recruitment gap or a rota structure that routinely under-schedules that period.
• Actual versus planned hours: The difference between what was rostered and what was worked, at shift level, across a rolling 12-month period, is the raw material for building a more accurate forward model.

 
The rota data most managers treat as admin is actually a financial record. Every understaffed shift has a cost. Every surplus shift has a cost. When you map those costs back against the periods that generated them, the patterns become obvious and the forecasting becomes straightforward.

Factoring In Seasonality, Holidays, and Special Events

The predictable variables most rotas ignore

Seasonality is the most underutilised forecasting input in shift-based scheduling. The operational pressures of a care home in December are different from those in July. The staffing demands of a hotel in August bear no resemblance to those in February. These differences are not surprises. They are predictable with a high degree of confidence, yet the majority of rotas are built as if each week were identical to the last.

A structured approach to seasonal forecasting identifies the following variables in advance:

Annual recurrence patterns:

• Winter pressures in care: higher resident health acuity, increased hospital discharge rates, and staff illness peaks in December through February create predictable demand spikes that require additional staffing cover
• Bank holidays: public holidays create simultaneous demand for premium pay and higher staff absence rates. A care home that has historically run short-staffed on bank holiday weekends has the data to pre-plan cover rather than scramble for it
• Summer holiday overlap: July and August generate simultaneous peaks in annual leave requests, particularly among staff with school-age children. Organisations that do not model this in advance routinely find themselves approving leave that creates gaps they then pay agency rates to fill

Hospitality-specific demand drivers:

• Seasonal footfall: coastal and rural venues experience demand distributions that are highly concentrated. A venue operating at 35% capacity in February and 95% in August cannot staff both periods efficiently from a fixed headcount model
• Events and sporting calendar: research by Experian found that major national sporting events can translate into a £233 million uplift in domestic hospitality spending. The Euros final of summer 2024, for example, drove drink sales up by 124% on the day in managed venues. A manager with access to the events calendar and historical transaction data can translate that into a specific staffing uplift requirement, by shift and role
• Conference and function bookings: unlike walk-in trade, event bookings are known weeks in advance. Translating confirmed bookings into staffing requirements at least four weeks ahead is a straightforward calculation that most operations perform reactively on the day

Care-specific demand drivers:

• Resident review cycles: dependency assessments, care plan reviews, and GP visits create predictable peaks in the administrative and care workload that require appropriate staffing
• Inspection readiness periods: CQC inspection activity has its own patterns. Providers who monitor their inspection history alongside their staffing data can identify whether compliance pressure correlates with specific staffing configurations

Building a seasonal adjustment into the rota model

The practical mechanism for incorporating seasonality into a rota forecast is a rolling 52-week comparison. For any given week in the upcoming rota period, the question is: what did actual hours look like in the equivalent week last year, and the year before? Where those data points show a consistent pattern, the forecast for the upcoming period adjusts accordingly.

This is not sophisticated modelling. It is pattern recognition applied systematically to data you already have.

Using Rota Data to Identify Your Busiest Times

How to read your own operational patterns

The most actionable output of a demand forecasting exercise is a heat map of staffing pressure across the week, the month, and the year. Once you can see where your operation consistently runs hot and where it consistently runs cold, rota construction changes fundamentally.

A weekly pressure analysis typically surfaces the following types of pattern:

Once these pressure points are mapped, the rota stops being a weekly administrative task and becomes a forward deployment model. Instead of asking “who is available?”, the manager asks “what do I need, and do I have it covered?”

The agency cost signal

Agency spend is one of the clearest indicators of demand forecasting failure. Every agency shift booked at short notice represents a period where the staffing model did not anticipate a demand it should have seen coming.

Tracking agency usage by shift type, day of week, and time of year creates an immediate forecast improvement. If the data shows that agency cover is consistently required on Sunday nights in February and March, that is not a random event. It is a signal that the permanent staffing model has a structural gap in that slot, and that the solution is a rota adjustment, not a monthly call to an agency at premium rates.

Agency dependency is almost never bad luck. It is the price of not having looked at the data. When you map your agency spend against your rota history, you stop treating it as an operational cost and start treating it as a procurement problem with a known solution.

From Forecasting Framework to Rota in Practice

What smarter rota-building looks like

A demand-led rota process follows a consistent sequence:

1. Review the historical baseline. Pull the last 12 months of attendance data for the equivalent period. Identify whether the upcoming period is typically high, medium, or low demand relative to the rest of the year.
2. Map the known variables. Note bank holidays, local events, confirmed bookings, resident reviews, or any other factors that will affect demand in the specific rota period being planned.
3. Identify structural gaps. Cross-reference the historical data with current headcount. If the data shows that Monday morning cover has been consistently short for six months, a rota adjustment will not fix a recruitment problem.
4. Build the rota against need, not availability. The required shifts are determined first. Staff are then assigned to those shifts. Gaps become visible before the rota is published rather than after the shift starts.
5. Set a review trigger. Define the conditions that will trigger a mid-period rota review: an absence threshold, a demand spike from a confirmed event, or a compliance requirement that requires additional cover.

How Orta Supports a Data-Driven Scheduling Model

Orta’s workforce management platform provides the scheduling infrastructure that makes demand forecasting operational rather than theoretical.

The platform’s shift scheduling tools are built specifically for the complexity of 24/7 shift-based organisations, including care homes, nursing homes, and hospitality venues running variable rotas with multiple shift types. The features that directly support a demand forecasting approach include:

• Shift templates that encode the staffing model for recurring patterns, so high-demand periods are always built on a consistent structural baseline rather than reconstructed from scratch each cycle
• Skills matching that ensures only qualified staff are assigned to roles requiring specific competencies, preventing coverage solutions that create compliance risks
• Real-time attendance tracking that captures actual versus planned hours at shift level, generating the historical data set that makes future forecasting progressively more accurate
• Open shift broadcasting that allows managers to fill identified gaps quickly through staff who have indicated availability, reducing the time-to-fill on short-notice coverage
• Agency integration controls that surface the cost of agency usage against permanent staffing, making the true cost of reactive scheduling visible in a way that drives proactive change

 
For providers managing the specific challenge of automating shift cover through employee self-service, Orta’s mobile-first approach gives staff visibility of their rota and the ability to flag availability directly, reducing the manager coordination overhead that delays cover decisions.

The how digital scheduling transforms care services article sets out in detail what the operational transition from reactive to demand-led scheduling looks like for regulated care providers specifically.

The Competitive Advantage of Scheduling in Advance

PwC’s 2025 UK Hotels Forecast identified data-driven performance management as one of the primary strategies for protecting margins in a sector where operating costs are rising faster than revenues. The same principle applies in social care, where Skills for Care data consistently highlights agency dependency and unplanned overtime as among the largest contributors to unsustainable cost structures.

The organisations that will build sustainable staffing models in both sectors are those that treat the rota as a strategic document rather than an administrative one, and the data that underpins it as an asset rather than an afterthought.

Demand forecasting is not a technological leap. It is a discipline. The technology makes that discipline easier to maintain consistently, but the core practice is available to any manager willing to look at the patterns already in their own data.

Frequently Asked Questions

What data do I need to start demand forecasting for staffing?

At minimum: 12 months of shift-level attendance records showing planned versus actual hours, a record of agency usage by shift type and time period, and an absence log with dates and roles. Most organisations already hold this data in some form. The challenge is consolidating it into a format that surfaces patterns rather than burying them in spreadsheets.

How far in advance should I be building my rota?

For most shift-based organisations, a four-week forward rota is the minimum for effective demand management. Six weeks is better for periods that involve bank holidays, seasonal peaks, or known high-demand events, as it gives enough lead time to recruit bank staff, adjust contracted hours, or pre-plan agency cover at agreed rates rather than emergency rates.

How does demand forecasting reduce agency spend?

Agency spend is predominantly driven by short-notice gaps that the permanent staffing model did not anticipate. Demand forecasting identifies those gaps in advance, when they can be filled through overtime agreements, bank staff activation, or rota restructuring at lower cost than same-day agency booking. Providers who model this consistently typically find that a significant proportion of their agency spend is addressable through rota planning rather than recruitment.

Can demand forecasting work for a small care provider?

Yes. The framework scales down as well as up. A single care home with 30 staff has enough rota history within six months to identify meaningful patterns in absence, overtime, and agency usage. The data set does not need to be large to be useful. It needs to be consistent.

What is the link between demand forecasting and CQC compliance?

CQC inspectors assess whether providers can demonstrate safe staffing levels at all times, including during pressure periods. A provider that can show a demand-led rota model, with documented evidence of how staffing decisions are made against operational need rather than availability, presents a significantly stronger compliance narrative than one that rebuilds the rota from scratch each week. See the what is workforce management guide for a broader explanation of how data-driven workforce planning intersects with regulatory requirements.

How does Orta help with demand forecasting specifically?

Orta provides the scheduling infrastructure and historical attendance data that demand forecasting requires. The platform captures actual versus planned hours at shift level, tracks agency usage, and provides the rota management tools to translate a forward staffing model into published schedules efficiently. Book a demo to see how the scheduling and workforce tools work together for your type of operation.

Reactive scheduling is a choice, not a constraint. Every organisation running shift-based operations already holds the data needed to forecast demand and build rotas that match it. The question is whether that data is being used or discarded at the end of each rota cycle.

Explore Orta’s workforce management tools | Book a demo