How many times have you heard the saying, “Work will expand to fit the time available”? Even in industries where capacity expands and contracts depending on how much work there is to do, teams can still appear to be working hard all of the time.
In value stream mapping exercises, Lean Six Sigma practitioners often identify “wait time” in a process, which invariably leads to a longer process cycle time. While wait times can easily be identified as waste, it is even more beneficial to probe a little deeper to determine what is driving the wait. The payoffs can be not only instant, but also painless.
Case in point: A financial services company conducted a Lean Six Sigma DMAIC project to improve its quotation process, which it uses to estimate the cost of its various services. While conducting a value stream mapping exercise, the Black Belt in charge of the project examined the wait times in the process and came up with a surprising result, yielding a quick win that was easy to implement and provided big returns.
One of the goals of the project was to reduce the standard processing time of the quotation process, which at that point was measured against an internally derived target of delivering a quote within five days after receipt of a work item.
The initial Lean perspective was to identify and remove non-value-added steps, thereby reducing the process cycle time. This was in line with senior management’s desire to increase customer satisfaction by delivering quotes to customers faster, which in turn was putting pressure on the operational processing teams.
The process was mapped and process cycle times were calculated for the quotes. These quotes were split into the following three types, based on complexity and size, which did not include wait times:
Type A = 120 minutes processing time
Type B = 195 minutes processing time
Type C = 320 minutes processing time
When mapping the wait times in the process, it became apparent that one factor was based wholly on the manager’s work-allocation strategy. This can be identified as queue time, where a work item waits for a resource to become available. The queue strategy used by the manager in the process was based on “latest start date,” which meant that a work item was not picked up until it reached the point of “completion date minus processing time.”
Though the team was measured on the internal target of five days, the process also captured the customer target date, which customers supplied on the quotation request form. Customer target dates ranged from “same-day service” to “two weeks.” This information, however, was not used as it was thought that any customer demand rate less than the current five-day standard was unachievable.
The project team began exploring the concept that had initiated the project, namely that customer satisfaction could be increased by a quicker turnaround of quotes. Gathering voice-of-the-customer feedback through telephone surveys, however, revealed that this assumption was incorrect. The most important factor to the customers (aside from quality and price) was receiving quotes when they needed them. Sometimes they needed a quick turnaround, while in other instances they were happy to wait for a longer period.
Analysis of the incoming work over the previous three-month period was made based on the comparison of two different target-date measures that could drive the work allocation: 1) the daily profile of work completed based a five-day processing measure and 2) the daily profile of work requiring completion based on customer target date.
The null hypothesis was that these two profiles, representing the work the teams were required to process each day, were statistically the same. This would support the premise that the processing teams had the capability to meet the customer target date, even where this was less than five days.
Figure 1 below depicts the results of a multiple Y box plot test that was used to measure the two profiles of target dates and determine if they were the same.
Figure 1: Customer Target Date vs. Actual Work Completed
The box plot indicates that the two profiles of daily work requiring completion – split into three types (A, B and C) for each – were statistically similar.
Figure 2 shows the results of a two-sample t test performed on the same data.
Figure 2: Two-Sample T-Test and CI:
Customer Target Total, Actual Completed Total
Difference = mu (Customer Target Total) – mu (Completed Total)
Estimate for difference: 0.05
95% CI for difference: (-7.74, 7.83)
T-Test of difference = 0 (vs not =):
T-value = 0.01
P-value = 0.991
DF = 173
The important figure from the t test was the p-value, which showed the probability of being wrong if we rejected the null hypothesis or, conversely, the probability of the two sample means being the same. In this instance, the p-value was 99.1 percent. This supported the visual box plot, which indicated that the daily profiles of work requiring completion were statistically the same.
At the start of the project, the prevailing perception among the project sponsor’s team managers had been that managing workloads based on customer demand would be unpredictable and impossible The statistical analysis using Lean Six Sigma tools, however, proved to the project sponsor that this assumption was false.
In practice, this discovery was a simple quick win, of which the operators were not even aware. They continued to pick items following the same scheduling rule as before, based on completion date minus processing time, but provided a much more customer-focused service. There remained some volatility in daily work volumes, but this instability was now driven by customer demand rather than internally derived targets.
Not all financial services processes will be open to such a quick win, but it is worthwhile looking at the work scheduling strategies behind a queue time very early on, as these often dictate the process cycle time and the amount of work in process (WIP). There are a number of options available that can instantly reduce cycle time and WIP with little or no change to the process.
The following work scheduling strategies can also be mixed and matched where required:
1. Operator decides – The manager or operator decides which work item is processed depending on some unknown factor. Sometimes this is about “cherry picking” the most critical items, but often it is based on which customer shouts loudest.
2. First come, first served – Work items are completed in the order in which they arrive.
3. Last come, first served – Often this happens when a department is swamped with work and items are simply picked from the top of the pile.
4. Shortest processing time – The work items that take the shortest time to process are completed first.
5. Due date – Work items are completed in the order of their due dates.
6. Latest Start Date – Work items are allocated based on their due dates minus the processing time.
Often in a process, practitioners will see an “operator decides” strategy, particularly when employees have been empowered to choose items from a pool of work. Items also jump the queue when the customer complains, which adds additional variability to the wait time and rewards “difficult” customers.
A scheduling strategy that is often overlooked in the above list is shortest process time. During times where there is more work than can be completed by the team, using the shortest processing time strategy will ensure that the maximum number of work items are completed, reducing WIP. The number of customer queries chasing work items and customer complaints also will be reduced, freeing up valuable resource time to concentrate on core activities.
The shortest processing time strategy can be coupled with other strategies by assigning priority to work items that have reached the latest start date. This will maximize the amount of work being completed by resources and minimize the impact of overdue work completed. A similar strategy based on customer importance or value can also be implemented.
In the case of the financial services company example, the opportunity to make a positive impact on process cycle time and WIP early in the project allowed the Black Belt to provide the project sponsor with a measurable and fairly painless improvement.
As an initial quick win, this project provided the Black Belt with some credibility with the process owner and the process operators, who had been under pressure to continuously improve performance that had not been aligned to what the customer needed.