Question 20.3: MEASURING THE TIMELINESS OF CHECK PROCESSING Woodstock Compa......
MEASURING THE TIMELINESS OF CHECK PROCESSING
Woodstock Company, a company in the construction industry, had recently experienced considerable expansion of its business volume. Due in part to this expansion, the finance department of the company was having difficulty processing checks to suppliers in a timely manner. Many of its suppliers were being paid beyond the normal 30-day period. This was not only making the suppliers unhappy, but Woodstock was also failing to obtain the discounts many suppliers offered for prompt payments. How could control charts help Woodstock solve its problem?
Objective To see how control charts can help Woodstock find the reasons for untimely check processing and suggest ways of decreasing check processing time.
First, it is important to realize that control charts cannot magically solve a problem such as the one Woodstock faced. However, they can help to show what is happening and point to possible solutions. To produce control charts, we assume that Woodstock measured the processing times for five checks completed each day. Each processing time is defined as the time from when a supplier’s shipment is received until Woodstock sends the check to the supplier. The file Checks.xlsx contains these processing times for 60 consecutive business days. Observations for the first 30 days were used to form control limits. The R chart (not shown here) for these 30 days is well within control, but the \bar{X} chart, shown in Figure 20.18, indicates out-of-control points on days 7 and 10.
Upon closer examination, Woodstock learned that on day 7 the people in finance, trying to improve a process with high variability and large processing times, implemented a change in the check preparation process. However, this change backfired—it actually made things worse—and was eliminated after 5 days. This change is a clear example of an assignable cause. The points we observe in Figure 20.18 are actually the result of two separate processes, those without the change and those (points 7−11) with the change. To understand the original process, Woodstock needed to eliminate points 7−11 and form new charts. This was done, and the plots of days 1–6 and 12–30 (not shown here) showed statistical control.^4
The process was now in statistical control, but this was no place to stop. Woodstock was alarmed at the high average processing times (about 30 days) and the high variability (average R’s of nearly 12 days). Management took a closer look at the check preparation process and discovered several unnecessary steps—duplicate paperwork and excessive “hand-offs” from one person to another. They took steps to streamline the process, and they continued to plot, using the control limits and centerlines from days 1–6 and 12–30. The \bar{X} and R charts through day 60 (again, with days 7–11 eliminated) appear in Figures 20.19 and 20.20.
These control charts both indicate out-of-control behavior, but of the kind Woodstock is happy to see. The R chart indicates a lower level of variability, and the \bar{X} chart indicates a decreased average time to process checks. The R’s are now averaging about 6.5 days, and the average check processing times are about 20 days. These improvements are a direct result of Woodstock’s management interventions, but these interventions were prompted by observing control charts and trying to understand what was causing them.
Even after day 60, Woodstock should not become complacent. First, it should recalculate control limits and centerlines, based on new data, say, from days 51–80. It could use these to check whether the improved process is in control with respect to the new limits. At least as importantly, it should continue to search for potential improvements in the process. If the average check preparation time could be reduced from 30 days to about 20 days, and the variability could be reduced as well, who’s to say that further improvements are not possible?
^4To do this in Excel, we copied the original data sheet to a new data sheet, deleted the rows corresponding to days 7–11, and formed control charts from the first 25 rows of this new data set.
