1. Why is DAV using SPC? What are the primary challenges in applying Statistical Process Control to a service industry compared to manufacturing?
SPC is a scientific approach to measure and improve the efficiency of existing processes. The way it works is: a) Fixing the control limits or the guardrails for the processes that one wants to measure. (In terms of how accurate one wants the outcome of a process to be) b) Measuring random samples (random being the key) of existing processes (measure the accuracy or the time taken to service insurance customers, essentially pick the process area to measure) to find the sample mean, and variance within the process. c) Post the measurement phase, come the execution phase where one finds out the changes that need to be made in the existing process to ensure that the accuracy stays within the specified control limits.
Unlike in manufacturing, where it’s easy to distinguish what is the “right” way of doing things, and the focus is more on improving the manufacturing process, in the services, the “right” way of doing things is defined by the people and it’s often subjective. More specifically: 1) The same SPC cannot be applied on a blanket scale to every team irrespective of how efficient they already are or how difficult objectifying their work as “good” work is. 2) The definition of right and wrong can quickly become very subjective based on different perspectives of individuals. 3) All teams cannot be measured in an objective manner since some of the work they do cannot be evaluated from this perspective. 4) People might feel it’s a burden 5) Senior managers can easily point fingers at the employees without deeply understanding the issue. 2. How large should each sample be for the experiment Schoss and Kluck describe on page 7?
The sample size depends on the level of accuracy that each process needs to measure itself. If the accuracy need is more, the sample size should be more. If the accuracy size is less, the sample size should be less. Sample size was set at 300/week (from Exhibit 4). 3. The first 12 weeks of the data on exhibit 4 represent the diagnostic period for the policy extension group. What are the 3-sigma control limits for the process? In which of the subsequent weeks is the process out-of-control (if any)?
The control limits that have been defined based on the sample size (n), variance of the sample from the mean, and P( the ratio of the number of errors/total sample size). There are two control limits; the Upper Control Limit and the Lower Control Limit.
UCL = P+3* Sqrt (P(1+P)/300)
LCL = P- 3* Sqrt (P(1+P)/300) Computing P from excel (Sum of errors / # of weeks / sample size), we get P = 17.2 UCL = 20.26 LCL = 16.18
Assuming UCL is 21, and LCL = 16
Weeks 4, 11, 12, 4, 15, 16, 22, 23, 24, 25, 26, 27, 30 are the ones where the processes seems to be out of control 4. What is your recommendation for Kluck going forward in their deployment?
Recommendations going ahead:
SPC should definitely be deployed, but with a few considerations: a) Identify objectively, what are the drivers to distinguish between right and wrong based on employee feedback and management consensus b) Identify which processes need what levels of accuracy. Certain processes would need a high accuracy rate for parameters like customer response level, others may not need c) Identify what the right sample size is for different processes. Teams which are already efficient might need to be samples less frequently than the teams. d) Objectify the levers of a particular process, and try to objectify these parameters. e) Focus on the improving the process, and not point fingers at the people.