Deutsche Allgemeinversicherung

Background:

Deutsche Allgemeinversicherung is a large German insurance company focusing in retail insurance. DAV is a major player in the industry but seeks to continuously improve during their fat years to prepare for their lean years. They aim to lower their external and internal costs by increasing its appraisal and prevention process. The back office part of the operation DAV Kundendienstgruppe consisted of over 2000 people in three primary divisions. The DAKG process flow quality appraisal and process flow is analyzed by SPC measures in order to find methods of improvement for the company that will aid in combating the loss of market share. Thus, the PMV project was rolled out in May 1995 and consisted of 15 departments. The trial run was 2 months and used SPC to capture the process defect results.

Problem Statement:

There are two clear issues that persist for DAV and their PMV implementation. The first issue is the measurement challenges that need to be altered for complete accuracy and project success. The measurement challenges include: the sampling size, the lack of weight given to more important/trivial quality control errors, measuring for the legal group, and automatic charting.

The second issue is the how to improve the performance phase not just for the initial implementation but in a way that is sustainable long term. Once the data is collected, analyzing it appropriately is essential. A p-chart is a basic right or wrong statistical measure so just finding the UCL, LCL, and accuracy percentage does not tell the whole story, or necessarily give the sufficient grounds to make changes.

Data Analysis:

In exhibit 4 of the case study a sample size of 300 applications were tested for the Policy Extension Group. The Lower Control Limits were 5.12; The Upper Control Limits were 29.27. The mean or errors for the sample size was 17.2 and the percent accuracy was 93.2667%. Over the course of the 30 week trial there were 2 weeks that surpassed the upper limit. Weeks 23 and 24 had errors of 33 and 46 respectively which are both over the Lower Control Limit. The statistical outliers in this situation seem to have an easy fix, given the fact that they are consecutive to each other. The problem could be