An example of our work in Operations
To show how difficult can be to define the real right sizing of personnel – and the huge amount of value that can be generated for companies – we will leverage on one of our projects on the subject: on processes driven by “call” interventions.
A number of business processes are in fact based on having people handle random events by involving them through calls (“call” interventions).
This has always been the case in the IG world (think for example to ‘Maintenance&Repair teams’ in plants). But today such processes are getting more and more important in all sectors, think for example to the Consumer Goods or the Service industry.
When quantifying the work force needed for such processes, companies face two main issues:
Much more than for other processes, the trade-off between sizing too much and sizing too little is crucial for the success and the economic sustainability of “call” processes (not being at the right place at the right time often means ‘lose everything’ in these cases);
The job descriptions of the people in charge of such processes – even when clearly defined, and most of the times they are not – contain not only “call” interventions, but also other activities of different types (e.g. tasks to do during ‘waiting’ hours, administrative activities,…), which demand different sizing logics.
We have developed a thorough methodology to solve this complex sizing issue. We show here a case applied to the Highway Assistance Service, but it can be transposed to any other process based on the “call” principle.
The methodology is formed by 3 steps:
– Analysis of process activities, definition of job descriptions and identification of the driver to assign to people ‘per shift’ (in the case mentioned, the highway section to ‘cover’ per shift).
– Definition of service level meaning for the “call” process, identification of the parameter to measure it, and choice of the value to use for it (in the case analyzed, ‘time needed to get to the event’ was chosen as measure parameter, and 20 min were set as value).
– Calculation of the work force needed vis-à-vis the job descriptions and the service levels defined.
Clearly, the 3rd step is the most complex, in the sense that it involves a bit of statistic calculations. In particular, two problems must be solved:
1) Random events can occur at the same time, or have time overlaps (for example, 2 accidents in the highways case), with the need of intervention of a second team.
2) “Call” activities must be harmonized with the other-type ones (in the case analyzed, the need for a continuous patrolling of the highway section assigned)
The first problem can be solved by defining what overlap means: 2 events of duration “d” are to be considered overlapping if the beginning of any one of the 2 occurs when the other one has not yet ended (see figure below);
calculating the overlapping probability: given a rate of events “r” of duration “d”, the probability that 2 events overlap is (3/2) r d;
finally, obtaining the value of the driver to assign to people ‘per shift’: the weighted average of the ‘overlapping’ and the ‘non-overlapping’ cases.
In general, a simple way to solve the problem is to consider the combination of the second more important activity (in the case, ‘patrolling’) with the “call” activity.
In the case analyzed, this harmonization phase eventually consisted in considering the ‘team starting position’ at the moment of receiving the call
calculating its impact on ‘time of arrival’.
One last point has to be underlined.
For how the model is conceived, the service level parameter represents an input. Well, used in the opposite versus (i.e., by iterating the calculations using different values for the parameter) – the parameter itself can be used to assess in a precise way the economic sustainability of the work force sizing.
The graph below shows such iteration for the highways case. By looking at the graph, it is evident that the marginal improvement obtained by adding one team to the teams already present on the highway is significant up to a certain point (6-7 teams), but reduces to nothing beyond that point.
In this way it is possible to calculate the true costs of the service level and, finally, obtain the ‘right’ sizing for the process under examination.