Category Archives: 2.2-Workload Quantification

The Wiggle Room

· · Leave a comment

As any planner knows, it is difficult to foresee the events that become obstacles to the future you planned. The workload represents the starting point and is what drives everything else in the scheduling process. Therefore, its accuracy is as important as its definition.

 

This workload changes as it proximity approaches. A planner can review the workload for tomorrow with much more accuracy than the workload in a month from now. The planner will have access to much more accurate weather, trends of the week, latest events, etc that will influence both the number and the shape of the workload.

 

When the schedule is built, it is usually done with a workload that has far proximity and therefore less accuracy. But once the schedule is done and communicated to employees, it is very difficult to make dramatic changes to that schedule (that schedule represents the employee’s life, remember?).

 

Since a planner knows they can’t change the schedule and know that the workload is approximate, it is good practice to leave some wiggle room for the schedule to be adapted later. Of all the formulas that were discussed earlier (rounding the employees, measuring in seconds vs. minutes, etc), one of these steps can be padded or rounded up to give breathing room to both the planner and the employees.

 

Now some of you may say that rounding up is counter-productive and goes against business sense to stay competitive and find ways to cut costs. My argument against that is that if you only focus on the aspect of employee productivity down to the second, you may end up with turnover ratios that are high, absenteeism that is high, overtime costs that are high, etc. The human factor will take over and all the standard HR ratios will start creeping up.

 

Obviously, you can’t round too high since you won’t be competitive at all and not produce any profit as an organization. There is a balance, but with my experience, a balance just a tad in favor of rounding up will prevent hidden issues to creep up later down the road.

 

The important thing in adding wiggle room is to round up ONCE only. If you pad the number of seconds up, then the number of employees up, then the hours up, etc, you will end up with a much higher workload than you normally have.

 

The trick is to be as accurate as you can be (depending on your business) and then identify ONE of your math steps to be rounded up. Ideally, you pick at either end of the spectrum (either on the initial task measure or at the end when you say how many hours or employees are required).

Open and Close Times

· · Leave a comment

No matter your business, someone will be coming in to turn the lights on. That preparation time has to be added to the workload through one of the workload types. Some manufacturing processes, especially in the food industry, need to stagger shift start times so that the first employee to walk in will turn on the lines and let them heat up for production before everyone else walks in to actually work on these lines.

 

These times are important to add to the workload. They represent the preparation of a work area so that it is ready to receive an employee and have that employee productive.

 

The same applies to closing times. The equipment needs to be cleaned, the lines need to be turned off in a certain order, etc. This is also time that is necessary to add to the workload.

 

In some cases, the employee doing the work is the one preparing his environment. In our grocery store, the cashier is the person that will open the register and count the cash at the end once they close it. It is not necessarily related to the open/close times of the store since a register can close in the middle of the day. In these cases, depending on the time required for these tasks, the workload needs to be changed so that every time one less cashier is required for example, 15 minutes area added at the end of that curve. So if 3 registers are required until 8pm and then only 2 are required, the workload needs to reflect that 3 registers are required until 8:15pm.

 

Note: the open/close times for the workload are not the same as setup time for the employees. For example, if the employees need to change before being allowed in the production area and you give them 5 minutes to do so, this 5 minutes is not required for the workload to be started. It is required for the employee to be ready to work and should be added to the employee’s shift as such, but not to the workload. This time is not part of producing work or getting a machine ready. Chapter 4 will discuss setup times more in depth.

Mixed Workloads

· · Leave a comment

The mixed workloads that have both Time Dependant and Time Independent tasks should be handled as Time-Dependant workload first. A planner needs to establish that demand and then add the time-independent hours to the curve. If we take the Butcher workload as an example, The volume of customers may require two butchers at some points in time but may even dictate that no butcher is required. Since you know you need a butcher at all times during the opening hours, you can measure the ‘free’ hours you have from the periods when no butcher would be required and compare that to the sum of hours required for the time-independent tasks. If you have enough room, then you don’t need to add hours. If you don’t have enough of these free hours, you have two choices as a planner:

 

The first choice is to dictate when these tasks could be done by adding hours to the valleys in the daily curve. Therefore, you keep your workload simple, easy to measure, and you make sure you have enough hours in the day to cover all the required work. The down side of this approach is that the representation of the workload is now twisted since the planner will measure the schedule against what seems to be a time-dependant workload only. So if an employee has to go home early or comes in late, that planner may take decisions based on a workload that does not represent reality.

 

The other choice is to have two separate workloads, one for the time dependant butcher and one for the time independent. This would represent more accurately the real workload when the schedule is measured. In counterpart, this complicates the scheduling process since the planner will now have to keep track of two distinct workloads and also be very precise on what is assigned to which employee and when. It will no longer be a simple one to one relationship between what the employee is scheduled on and what the workload requires.

 

Both approaches have their ups and downs. In my experience, your selection of one method or the other will depend on the size of the business (as explained earlier, the bigger the business, the more detailed it will be), the capacity of the planners, the tools available to the planners, and the capacity for the employees to understand the schedule and its details.

Time Independant Workload

· · Leave a comment

Looking at the list of workloads in our grocery example, you’ll notice that the Clerk has the most time-independent tasks. He only has one time-dependant task which is to answer customer questions. Although it is a valid task that should be part of a job description, it can’t really add significant workload to that person. A question will take much less than one minute and you may have one for each 50 customers that come in. Therefore, the Clerk workload could be considered a time-independent workload altogether.

 

Since this workload is not tied to a particular time of the day, a planner still needs to evaluate and put time on each task. The trick is to represent that work in the form of hours per day. So if you evaluate each task and sum them, you may end up with 5 hours per day for example. This sum may also change per day of the week. There may not be any delivery truck to unload on Sundays. Therefore, the Clerk workload will have a lower sum on that day.

 

This will bring you to different sums on each weekday. Figure 2 shows an example of these sums.

 

Figure 2. Time-independent workload

 

You will also find that even though this work is more regular, it is exposed to seasonal trends just like the cashier workload. This clerk will be stocking shelves more often during Holiday Season.

 

Some industries like healthcare transform the number of hours into Full-Time Equivalent (FTE). So if a full-time employee usually has an 8-hour shift, the sum of hours within the day would be divided by 8 in order to get to a number of employees per day. With this approach, the planner now only needs to compare the number of employees scheduled vs. trying to add the scheduled hours.

 

Although the time-independent tasks are not relative to a time of day, a planner may elect to choose a resolution. That day represents the recurring or resolution time selection, just like the time-dependant workload. Some industries divide the day into pre-determined shifts and indicate the number of hours (or FTEs) required in each part of the day. Manufacturing and Healthcare industries usually have this type of pre-determined cut down of shifts that is equivalent to what employees will receive as shifts. Again, this helps in the process of measuring your schedule. The down side of this approach is the lack of flexibility. As soon as one employee is not working the same shift as what is in the workload, then the planner is back to measuring hours or fractions.

 

For the expression of workload, it is therefore important to be as precise as can be managed. It is also essential to differentiate the expression of the workload and the measurement of a schedule. The measurement, or coverage, can be more complex that the workload expression and they don’t necessarily need to be linked. Chapter 5 will discuss schedule measures in detail.

Time Dependant (Part 2 of 2)

· · Leave a comment

Time Dependant Workload (Part 2 of 2)

Back to our example, if we decide to round the number, the result of the above hour would be two. You do this math for every hour you are open and you end up with figure 1 below. If you lay out your workload through time, you’ll see the number of employees varying during the day. It directly reflects your customer’s flow.

 

Figure 1. Number of employees required per time of day

 

Some of you may say that since employees have a different productivity (some cashiers are faster than others for example), you could schedule according to the customer output directly. Why translate to number of employees, right? This would be the optimal way to schedule since you could have your fastest cashier on staff and therefore need one less cash register open, which would save you money. Then why isn’t it done this way? Simply to keep things manageable. Otherwise, a planner would need to measure against all sorts of different drivers to establish a schedule and it would become quite a challenge to manage. Imagine your fastest cashier calling in sick: does this mean you replace that person with two other cashiers? The daily management of the schedule would quickly become overwhelming. So be practical and express your workload in number of employees required.

 

The cashier workload is now ready. You’ll notice that in our example, there are some big peaks and valleys. The number of customers would require up to ten registers for one hour. In our store, not only we do not even have ten registers, but we can’t ask folks to come in for only one hour of work. There will be constraints (see chapter 4) that impose a minimum duration when someone is scheduled to work.

 

Again, to help yourself and for simplicity, you can smooth out the employee requirements by reducing the peaks that represent noise. The more your resolution gets to be precise, the more of these peaks you will see. The same goes for valleys where all of the sudden no cash register is required because you don’t expect customers between 9am and 10am let’s say. You’ll still need a minimum of one cashier open no matter what the customer count is. So in your translation exercise, put a minimum and a maximum. Also, look at differentials between the previous hour and the next hour. For example, if you need 2 cashier for one hour, then 10 cashiers for the next hour, then again 3 cashiers for the next two hours, you can see that this 10 should be 3 or 4 but will not be 10. Rule of thumb: when you have peaks like this, you may have the ability to schedule shifts that overlap (the end of the shift overlaps by one hour the beginning of the next one). Therefore, you can make this workload at the previous hour (2) added to the next hour (3) and make that number a 5 instead of a 10.

 

The same logic applies to valleys, but in this case, you may elect to keep the valleys intact and not readjust according to the previous and next hours. These valleys will allow you to schedule breaks later in the process or dispatch the time independent workload.

 

This daily workload not only will be different on each weekday, but will also change from week to week as seasonal patterns will influence customer behavior. Holiday season will have higher traffic than the beginning of the year when no special events are happening. Total volume will therefore change from week to week.

Time Dependant Workload (Part 1 of 2)

· · Leave a comment

Let’s take the Cashier workload and analyze how we will quantify that one. This workload has only time dependant tasks and therefore is directly related to the number of customers showing up at the cash register. These customers will have different number of items to buy and will pay through different methods (cash, credit, check, etc). As explained earlier, each customer will take some time to check out and each cashier will have a limited capacity in number of customers.

The first thing to do is to pick a resolution and a unit of measure. The unit of measure is the time it takes to serve a customer. The resolution is the period of time where you will measure the count of customers. You can count the number of customers per 15 minutes or per hour or anywhere below, above, or in between. Usually, you choose a resolution that is manageable depending on the tools you have. If you have a pen and paper, I would suggest never going lower than one hour. If you use Excel, you could manage all the way down to 30 minutes. High end software allow for as low as 5 minutes in some cases.

 

You may wonder why would anyone manage a schedule down to the 5-minute increment? In our simplified example of one store, the amount of time to service a customer will be counted in minutes. No point in counting seconds because I have four cash registers which of only two are open most of the time. So if I say it takes 55 seconds to serve a customer and I round it up to one minute to keep my math simple, the difference on the week is insignificant. Take 5 seconds times the number of customers visits, say two thousand, makes 10,000 seconds a week, which is roughly 3 hours over the whole schedule. Not much. By padding the time required to serve a customer, you are adding hours to your workload quantity that is not necessarily required but that saves you aggravation and funny math.

 

But if our grocery store is successful and if we buy more stores and now have a chain of hundreds of store, you can multiply that 3 hours by 200 stores which makes 600 hours of extra time scheduled. Use any hourly amount, it starts to become significant. And if you perform the same exercise for all tasks, your over-scheduled number becomes very significant. This is why large enterprises do buy software that measure to the second the tasks that are performed in order to reduce the waste that accumulates due to rounding. Your workload definitions will therefore get longer and more detailed and will be measured more precisely as you grow your business. It will be directly proportional. If you don’t, you are exposed to over scheduling once you get to the schedule.

 

So for the sake of simplicity in our example, we will pick a one minute per customer average serving time and a one-hour resolution. Therefore if it takes one minute to serve one customer, our cashier has a capacity of 60 customers per hour. We take the expected business of that hour (say we expect 100 customers), this would create 100 minutes of work to do in one hour. You divide the amount of work by the resolution (100/60) and you get 1.66 employees. Since you can’t schedule two thirds of a person, you have to decide what to do with the fraction (do you schedule someone or not). You may elect to cut the fraction all the time, make the fraction whole all the time (if the result is 1.1, make it 2) or simply round the result. This will have a direct impact on service and cost. If you decide to cut the fraction all the time, it’s good for cost, but bad for service. If you decide to make the fraction whole all the time, it’s good for service but bad for profit. You could also decide to have a different rule depending on the relative part of the fraction. In the math above 0.66 out of 1.66 is a big proportion. You could have a requirement for 10.66 employees. In this case, deciding between 10 or 11 employees will have a much reduced impact on customer service because the capacity is not greatly affected (versus choosing between 1 or 2 employees which represents 50% reduction in capacity). You can already feel that even though you haven’t even started talking about employees, a simple rule of thumb decided by a planner may have a huge impact on the business’s success.

 

Workload Quantification

· · Leave a comment

There are two distinct perspectives of workload quantification that greatly influence a schedule. The first is the time-dependant service. For example, cashiers in our grocery store will be at their register at certain times. You would expect the number of cashiers available to grow as the customers come in. This work is time-dependant since the volume of customers walking in fluctuates with a direct dependency on the time of the day and the day of the week.

 

Best way to illustrate this example is to plan for only one cashier. This cashier will take a certain amount of time for each customer. If the customer volume is low, that cashier can handle the flow of let’s say 2 minutes per customer. So as long as no more than one customer walks in every 2 minutes, there is enough of one cashier. But if there is one customer every minute, then one cashier will not sustain the flow. Each minute, that cashier’s workload increases by 2 minutes because a customer was just added to the line. After 4 minutes, that cashier served 2 customers, and has 2 customers left in line. By the time the cashier takes care of these 2 other customers, another 4 customers are now waiting in line. So the scheduler should be planning for a second cashier during this increase of customer flow. That is a time-dependant work.

 

The second type of work relates to time-independent tasks. For example, our grocery store needs to stock shelves, clean the floor, etc. This type of task can be accomplished sometime during the day but does not have to happen at a specific time. It is independent of the business volume as to WHEN it will happen. It is not necessarily independent in duration since for example stocking shelves will take longer if you had a particularly busy day with lots of sales. But that task will not matter when you do it in the day (except of course if the shelves are empty…)

 

Let’s go back to our example and tag each task as to what it is: TD (Time Dependant) or TI (Time Independent)

 

Cashier:

1.e.i. TD: Service customer at register

2.a.i. TD: Open and close register

Bagger:

1.a.i. TI: Clean entrance

1.b.i. TI: Gather carts from parking lot

1.e.ii. TD: Bag the items

1.f.i. TI: Pick up carts around the exit

Clerk:

1.c.i. TI: Stock shelves

1.c.ii. TD: Service customer on item locations

2.b.ii. TI: Go through shelves inventory

2.c.i. TI: Empty truck

2.c.ii. TI: Store goods in storage room

2.d.i. TI: Stock shelves

Butcher:

1.d.i. TD: Service customer at butcher

1.d.ii. TI: Clean slicer and other bench tools

2.a.ii. TD: Prepare/store meat

2.b.i. TI: Go through meat inventory