Gathering and Interpreting AutoRec Data

It took the full month allotted in their preliminary plan for the team to gather data on their three targeted measures. They were fortunate, because the data collection period covered the end of the first Quarter of the year, so they could see how the process performed during both calm and busy cycles. (They knew it was important for data to be representative of how work levels and other factors vary over time.)

Here are the conclusions they drew from each of the measures:

• "Delivery Defects." The data gathered on this key output measure (actually, several measures) was compiled in a spreadsheet. As Elena from Procurement noted, "There are lots of things we could look at in this data!" For the time being, though, they developed two views of the data:

  1. The performance of the process was determined to be a DPMO of 122, 800 or 2.7 sigma.

  2. Defect data was broken down by type and displayed on a Pareto Chart (explained below). This revealed that most of the problems related to incompatibility, with hardware problems showing the most incidents.

• "Process Cycle Time." Average cycle time, from order entry to delivery, was found to be 17.3 days. A breakdown of the time involved in the major process steps (from the SIPOC diagram) showed that the largest amount of that time was devoted to Order Assembly—11.6 days.

• Order/Shipment Discrepancy. For this measure, the team was able to make use of existing data on defective deliveries. They were checking to see whether the orders themselves had been done incorrectly or if the problems were arising somewhere later in the process. The data were conclusive: For about 93 percent of the defective orders they examined from the previous four months, the Order Specification Sheets (OPS forms) were different from what was actually shipped to the customer. They also checked a significant proportion of those to find that the OPS forms were accurate—that is, the information did reflect the proper customer configuration.

Altogether, the data gave the AutoRec team a much clearer picture of the problem, and helped them to narrow their focus as they begin the search for root causes of the defective deliveries. They were able to update their Problem Statement based on the findings in Measure:

Forty percent of orders delivered to AutoRec corporate clients are not meeting customer requirements, including 30 percent for Hardware and Software incompatibility problems. These defects are hurting our image, creating customer dissatisfaction, and costing us roughly $350, 000 per month to rework rejected orders. Continued high levels of delivery errors threaten our position as a leader in this growing industry.