Managing working time: Demand-led labour scheduling
Section five of the Personnel Today Management Resources one stop guide to managing working time. Other sections.
|
Understand the impact of demand on working hours Get advice on how to design job profiles and work patterns to meet demand Prepare for
volatility of demand and other contingencies |
The key to successful demand-led work scheduling lies in the process of understanding, identifying and quantifying business demand, and its corresponding requirement for labour. This requirement then forms the basis upon which a responsive labour supply model may be developed.
The process of business analysis will use historical records of throughput, production and requirements for service, future forecasts of demand and estimates of business improvement.
The traditional view of labour planning assumes the prevailing contractual and working time environment and seeks to understand what levels of production or service can be achieved within these constraints.
The more radical and arguably more rational alternative would seek to understand the model of demand and to then consider the labour structure required to achieve it. It is this latter perspective which drives the principles of demand-led scheduling.
Understanding demand - peaks and troughs, cycles and seasonality
In the example below, demand for labour hours has a distinct seasonal profile, which is relatively predictable - with a winter peak and a summer low period. The graph illustrates the areas of under-utilisation of the labour involved during the summer (below the weekly base hours line) and the usage of overtime to meet the peak during the winter (above the line). The core supply of labour hours is constant at around the 135 level. Clearly the use of overtime at premium rates is expensive and equally clearly the under-utilisation of labour during the low periods is wasteful.
The modular hours profile line on the graph demonstrates a supply model which feeds-in labour at different levels in line with the changing demand across the year. As the sum of the hours worked during the year at the three different levels in the supply model are equal to those in the core contract, there is no overtime element. Similarly, labour under-utilisation in the periods of low demand is for the most part, avoided.
While the example refers to service engineering, this type of problem affects all types of organisation. Business demand may vary through the day, or from day to day, or from week to week or season to season. In some industries, all these variables may occur. For example, consider the demand pattern within a vehicle rescue environment where there are multiple influences on requirement for service. Some of these are predictable and some are not. Instances might include morning and evening peaks, weather-related variations, seasonal fluctuations associated to holiday taking or incidents, and event and venue-related peaks and troughs. As with the service engineering example, traditional models of supply cannot accommodate the array of demand variations without similar inefficiencies.
Similarly, air and other transport services, call centres, warehouses and retail outlets all have demand patterns that, due to proximity, reflect the behaviour of the consumer and require a mode of labour supply to suit.
Industries with long cycle times, or processes that are inherently continuous may manage variations in real demand elsewhere (eg, in the warehouse), maintaining a constant supply of labour and output capability. However, more and more manufacturing industries are now operating on a just-in-time basis and producing to order rather than holding stocks. This creates very different demand profiles and calls for different arrangements in labour supply.
In business administration and banking, cyclical tasks can have a very predictable monthly profile in functions such as payroll, accounts, sales order processing and billing. Again this can create different requirements for labour to achieve the peak cyclical activity and may mean low levels of activity in the troughs.
Shorter cycles, such as those found in call centres, tend to require different numbers of personnel as the day progresses to match what is often an 'M' shaped profile.
Defining demand
The understanding and definition of demand is therefore one of the key elements in demand-led labour scheduling. Without a full appreciation of the drivers, realistic supply models simply cannot be created. The process is orientated towards data analysis and valuable historical data may be drawn from any representative reference period but typically of three, six or 12 months. Many organisations now have facilities to store and access good quality information about business throughput and demand profiles. This may come from purchase ordering and customer relationship management (CRM) systems, manufacturing and warehouse management systems, tills and related technology in supermarkets and fast food outlets, flight schedules and passenger movement records in air services and automated call distribution (ACD) systems in call centres. These, and many other systems and data archives, are valuable repositories of demand data which can be used to specify labour and skills requirements hour-by-hour, day-by-day and period-by-period.
Data related to the local standard measures of activity, may be drawn from a diverse range of measures. For example, tonnes produced per year, room occupation rates and restaurant covers per day, passengers per hour, cases of product per minute, patient procedures per day, incoming/outgoing calls per hour. And standard hour records can all be used to enable organisations to conduct a detailed process of analysis, modelling, projections and validity checks in line with new forecasts.
Patterns and profiles of demand frequently remain the same year-on-year, but absolute volumes can obviously change - the M-shaped weekday curve of the call centre or ambulance control room rarely changes, but the underlying number of calls can change from year to year or season to season for many different reasons.
Skills and functional profiling
Gathered data must then be interpreted into a requirement for labour within different skills categories and functional requirements.
Some functions will require a constant level of manning despite volume. Thus in a call centre, there could be a constant requirement for the presence of one shift manager or supervisor at all times while the number of agents may increase or decrease in line with the level of activity.
Sometimes teams of labour may be the fundamental building block. For example, for a warehouse to function at its minimum level, it may require two pickers, a packer and a forklift truck driver, constituting a team of four employees requiring three different skills. Differing numbers of teams may be required at different times of the day, but labour is always fed in by the basic unit - 'a team'.
Multiple supply structures or patterns may need to be designed since demands for labour can vary according to different parts of the business. This flies in the face of traditional rostering where a one-size-fits-all policy is often dominant. It is not unusual to find a single pattern being worked across engineering, warehousing, production, and quality analysis, despite the different demand profiles of these functions.
Similarly, different supply structures may be required for diverse geographical locations. Vehicle rescue demand profiles, for example, vary widely between North East and South West England. The same can be said for patterns of demand for firefighters in commercial areas, where fires tend to break out during the day, as opposed to domestic areas where fires tend to break out during the evening and early night time.
Designing working patterns
The design and development of demand-led patterns can be a time-consuming and complex process albeit rendered easier nowadays with the availability of advanced software systems (see Section 8 ).
Given that labour scheduling and shift patterns are often not well understood, the presence of a computer application to provide a structure and methodology and rapid calculation of solutions is vital to the achievement of effective demand-led systems.
The process is highly mathematical involving multiple variables and 'what if?' calculations to define models which meet:
the complex demands of the business
the varying aspirations of the workforce for acceptability of pattern and leisure time.
Demand data can be analysed in terms of both shift lengths and rota patterns. A manufacturing environment may call for eight- or 12-hour shifts with fixed shift start and end times, but a service-related environment may benefit from staggered starts and longer or shorter shifts to meet daily peaks and troughs. This type of application may call for a range of shifts for full-time people between six and 10 hours. However, a continuous process industry may only require one pattern for the full year, while a seasonal business may require a number of different patterns at different times of the year depending on the demand for each period.
The service engineering example seen in the graph on page 19 finally required three different patterns for low, core and high periods of demand across the year. Each pattern used different shift lengths to cover the respective demand levels of each period, with shorter shifts in the low period and longer shifts with more concentrated individual attendance schedules in the high period.
Different sectors have opted for different mechanisms and devices for labour supply. Retail, for example, has very successfully adopted part-time working to meet peaks at weekends and weekday evenings. Call centres have adopted multiple contract types delivering a range of hours at different times of the day and on different days of the week. Other organisations have followed the annual hours route for a more generically flexible system of modeling labour structure.
In the leisure and hospitality environments part-time working has long been used, but other devices are also widely found. Split shifts are used to match morning and evening demand 'spikes', such as those experienced in receptions, restaurants and bars. An eight-hour day on reception may be split into 0700-1000 and 1600-2100.
As indicated above, in shiftworking environments, patterns can now be designed which feed in labour across the day, week or year in line with need while always balancing to the contractual hours for the prescribed period. Different shift lengths and different rotations may be required to fine tune the model to truly flex to the different patterns of demand identified.
Volatility and contingencies
The concept of demand-led rostering depends heavily upon the predictability of demand patterns and volumes and for the most part there is a good sense in organisations as to when their main demand and quiet times occur. There will, however, always be circumstances which cannot be planned for and unpredicted spikes caused by a whole range of local and industry influences.
For example, the food manufacturing sector is highly subject to the announcement of a 'BOGOF' (buy one get one free) offer by the promotions department of a major supermarket chain. This type of activity can give rise to massive and very short-notice increases in demand for product which can play havoc with the best-planned arrangements, but the manufacturer has to respond. Major incidents have the same effect in the emergency services sectors and sudden and unexpected hot or sunny spells in the UK will affect traffic-based environments in some locations and the hospitality and leisure sectors.
It is important to be able to identify what is true volatility and what is not. Many businesses think they have volatile demand, but when a thorough analysis is carried out it becomes very apparent that what seems to be volatility is in fact highly predictable or sometimes self-inflicted, eg, the closure of a warehouse for holidays at a peak period of demand will cause large spikes of demand for service both before and after the closure. Clearly in this circumstance better planning may have avoided the problem.
There are, however, very real circumstances where volatility is experienced and mechanisms must exist in the demand-led environment to deal with these. The service engineering example above demonstrates well that the precise number of calls for service could not be predicted and at peak times service levels were subject to adjustment. Here, a mechanism was employed to give a buffer of additional hours before overtime was required, with these hours being paid back at quieter times. Similarly, the early onset of the peak period due to bad weather would call for the high season shift pattern to be commenced earlier than planned.
The Gleneagles Hotel (see Section 10 ) uses the concept of banking hours to deal with unanticipated highs and lows of demand as it constantly strives to achieve the highest standards of service. Control and fairness issues can arise with this type of arrangement, but they can be minimised with careful management.
In annual environments (see Section 6 ) reserve hours and committed hours are used to cover for contingencies and unplanned spikes in demand or shortfalls in labour supply.
Labour supply issues
Demand analysis and profiling, therefore, takes a key role in this process. However, other influences exist which may affect the ability of the organisation to deliver the labour required at the appropriate time. The most detailed plans to meet demand can be confounded by the now many different factors that can affect attendance at work by the employee. These can range from:
long- and short-term sickness
jury service
parental leave
compassionate leave
annual and service-related holidays
increases in public holidays
flexible working
training.
And there will be others.
In traditional environments, these supply-side influences are met by the use of the usual mechanisms. In the demand-led environment, absence from work in whatever form needs some account to be taken and appropriate plans made.
By far the largest disturbance to labour supply in traditional models derives from holiday taking, which can account for between 10% and 15% of total working time available depending on the environment. In an annual hours system (see Section 6 ), holidays are very often rostered into the patterns of work which removes the problem of cover and administration. Traditional models, however, will still require means of providing the cover for holidays. This may derive from peripheral staffing or, alternatively, cover shifts can be built into the pattern in line with estimated holiday demand profiles. In the latter scenario, control becomes paramount to ensure the appropriate levels of holidays are taken such that full utilisation is maintained and a backlog of untaken holidays does not build up.
Training has to occur for all businesses and means time away from productive work. However, this can be planned for at times of low demand and scheduled into the working patterns with appropriate cover arrangements.
Sickness occurs at different levels within all organisations and it is difficult to know when, for example, a flu epidemic may occur or when an individual may suffer an accident. If the normal level of sickness within a business is 5% per annum, the hours required for cover can be calculated and included within the supply model and the mechanisms described above for use with demand volatility may be invoked to cover the absence.
Other forms of absence are less easy to anticipate or plan for. However, the principles remain the same, as they will occur to some degree and an estimate of additional hours to cater for cover can be set into the supply model.
While the demand-led approach seeks to minimise the use of temporary or overtime working these mechanisms are still available and can be used to cover the less predictable elements of volatility and labour supply.
|
COVER SUMMARY |
|||||||
|
Time period |
Mon |
Tues |
Wed |
Thurs |
Fri |
Sat |
Sun |
|
1200-1400 |
1 |
1 |
1 |
1 |
1 |
1 |
0 |
|
1200-1400 |
1 |
1 |
1 |
1 |
1 |
0 |
0 |
|
1400-1600 |
2 |
2 |
2 |
2 |
2 |
0 |
0 |
|
1600-2000 |
2 |
2 |
2 |
2 |
1 |
0 |
0 |
|
2000-2200 |
3 |
3 |
3 |
3 |
1 |
0 |
0 |
|
2200-0000 |
3 |
3 |
3 |
3 |
0 |
0 |
0 |
|
0000-0600 |
2 |
2 |
2 |
2 |
0 |
0 |
0 |
The above cover summary and graph represent the labour being delivered by the shift pattern below.
|
DEMAND-LED WAREHOUSING EXAMPLE FOR PEAK BUSINESS PERIOD |
||||||||
|
Week/ team |
Mon |
Tues |
Wed |
Thurs |
Fri |
Sat |
Sun |
Totals |
|
1 |
1400-2000 |
1400-2000 |
1400-2000 |
1400-2000 |
14000-2000 |
|
|
40 |
|
2 |
2200-0600 |
2200-0600 |
2200-0600 |
2200-0600 |
|
|
|
32 |
|
3 |
1400-0000 |
1400-0000 |
1400-0000 |
1400-0000 |
|
|
|
40 |
|
4 |
0600-1400 |
0600-1400 |
0600-1400 |
0600-1400 |
0600-1600 |
0600-1200 |
|
48 |
|
5 |
2000-0600 |
2000-0600 |
2000-0600 |
2000-0600 |
|
|
|
40 |
|
Totals |
44 |
44 |
44 |
44 |
18 |
6 |
0 |
168 |
NB - different levels of supply are scheduled for different days of the week and times of the day; the 2000-2400 overlap Monday through Thursday provides peak 'within the day' cover for loading of overnight HGV deliveries.
|
Section one: Why
employers must tackle working time |
