Assignment: Case Study 3 – Deutsche Allgemeinversicherung

Course: OPM 7120

Section: G01

Professor: Richard Pound

Group Number: 6

Names: Ariful Haq Chowdhury, Matt Odger, Mushfiqur Rahman, Xiaofan Wang, Xin Huang
Question 1
Considering First 12-week data, we are calculating P Bar and deriving Upper Control Limit
(UCL) and Lower Control Limit (LCL).
Total Sample Size for First 12 Weeks = 3600
Total Error made for First 12 weeks = 188
Average Error Percentage in First 12 week,
𝑇𝑜𝑡𝑎𝑙 𝐸𝑟𝑟𝑜𝑟 𝑓𝑜𝑟 𝐹𝑖𝑟𝑠𝑡 12 𝑤𝑒𝑒𝑘 188
P Bar = 𝑇𝑜𝑡𝑎𝑙 𝑆𝑎𝑚𝑝𝑙𝑒 𝑆𝑖𝑧𝑒 𝑓𝑜𝑟 𝐹𝑖𝑟𝑠𝑡 12 𝑤𝑒𝑒𝑘 = 3600 = 0.052222

So, (1-P Bar) = (1-0.52222) = 0.947778

𝑃 𝐵𝑎𝑟 (1−𝑃 𝐵𝑎𝑟)

So, UCL = 𝑃 𝐵𝑎𝑟 + 3 𝑋 (√𝑊𝑒𝑒𝑘𝑙𝑦 𝑆𝑎𝑚𝑝𝑙𝑒 𝑆𝑖𝑧𝑒)

0.052222 𝑋 0.947778
= 0.052222 + 3 𝑋 (√ ) = 0.090756
300

𝑃 𝐵𝑎𝑟 (1−𝑃 𝐵𝑎𝑟)

And, LCL = 𝑃 𝐵𝑎𝑟 − 3 𝑋 (√𝑊𝑒𝑒𝑘𝑙𝑦 𝑆𝑎𝑚𝑝𝑙𝑒 𝑆𝑖𝑧𝑒)

0.052222 𝑋 0.947778
= 0.052222 − 3 𝑋 (√ ) = 0.013688
300

Individual Error Percentages for each week has been calculated as well. (See Exhibit 3)

We used this UCL and LCL as ceiling and bottom of the chart and plotted all the error
percentage for each week to see if the error percentage breaches any of the limit control. The Plot
is given below

Control Chart
0.2000
Error Percentage

0.1500

0.1000

0.0500

0.0000
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30
Week Number

P P Bar UCL LCL

Here we can see from the plot the Week 23 and 24 breach the UCL (Upper Control Limit)
meaning that error percentages increase and breaks through Upper Control Limit.

Page 1 of 3
 23 weeks from the week of December 1st would put the outlier errors in May Next year. It
is safe to assume they would not have as many part time workers working in May because their
busy season is from September to December. With the average staff amount and many people
likely taking vacation at this time, they would be working with a skeleton staff and may be issuing
overtime. Overtime may lead to tired employees and increased errors.

Though the fall in errors in week 25 means that they may have hired temporary associates
to reduce the errors noticed in week 23 and 24, but the consecutive slightly increase in error from
week 25 to 29 means that the temporary associates will also be having trouble with processing
policies correctly probably because of lack of experience, insufficient training compared to the
regular associates and large amount of processing.

Question 2:
Implementation plans:

1. Employee involvement is an important factor affect the performance. Training employees

to utilize statistical techniques to measure and analyze data, which make employees felt
that they are empowered and contributed a lot to the company.
2. Providing customers surveys to make sure customers are satisfied with their services and
the new policy procedure is under control.
3. Setting dates to review goals and objectives are met. Ensuring make continuous
improvement in quality.
4. Providing a reward system to facilitates growth of new policy as well as keeps employees
productive and be more efficient. Also, it helps organizations becomes more competitive
and retain key employees.
5. Using the gathered data to improve current plans or to modify execution to better suit
goals.

Challenges:

1.) Better teams do more sampling

If the tasks all take the same amount of time to complete, they should take the same sample
size to ensure that even the low accuracy numbers have a big enough statistical significance to
make judgment on. More specific areas like the legal division will have a smaller sample size due
to time and judgment needed per assessment. It would need to be explained that a certain time
percentage of every day is required for sampling, not necessarily a certain sample size for all
divisions.

2.) Measurement challenges

To decide which is a mistake or not, Annette can look to industry standards or professionals
that she knows. There are others mentioned that are out there measuring their performance and
would be using similar metrics. Secondly, she can allow each of the departments to meet as a team
and decide what is an inaccuracy and which is a pass. Letting the employees and immediate
managers decide this will give them autonomy, pride over the results, and will ensure a realistic
measurement outlook instead of measuring something that doesn’t matter. All metrics will come
from the bottom and be confirmed with the executives before commencement. In addition,

Page 2 of 3
managers can provide a weighting error system that specify critical and less critical errors. Using
a clear weighting system help associates pay attention to the details and reducing errors.
3.) Measuring lawyers
For the legal division, it is possible for them to have two measurable data points. The first
can be an honest review of your own legal work. After letting a week pass in-between the work
done, you would have an honest measurement of how you thought the legal advice was presented.
Managers could randomly verify each employee’s work and rating scale once per week.
The second measurable data point would be consumer reviews. Upon completion of a legal
assignment, the client can mark if “their legal issue was resolved” (Yes/No) and then have a sliding
scale to rate from. It may be hard to judge legal opinions as a member of the general public, but a
lot of clients would factor in the lawyer's customer service and perceived legal representation
throughout. Providing customer surveys to see the client loyalty, satisfaction and referrals is an
efficient way to measure the performance. Additionally, building a system to measure how long
the case need to be finished with different levels of complexity. It is hard to
measure the numerically results but at least managers have a general idea of who is better at what
kinds of cases as well as working efficiency.

4.) Automatic charting

Assuming in the case that they are referring to the IT department not marking their progress,
but only setting up a system that has a TRUE/FALSE function for any mistakes, this is an honest
and appropriate idea to be implemented throughout the divisions. Once the metrics are met for
each of the divisions, it would be a matter of setting up the system specific to each department
(that uses online systems for their metrics).
As this is costly, it will help the employees understand that this system is here to stay and
will allow management to easily track results in real-time online, as well as tweak the metrics and
continue temporary improvements in the future.

6.) Employee perception

To implement, we will have to ensure that the employees understand why there is
measurement being done. If they don’t understand where the potential challenges are, they won’t
buy into the system and may cut corners on getting accurate data. As a positive for the employees,
understanding that less errors mean less time spent doing work twice, and the easier their job
becomes.
Like the last point, it’s about getting buy-in from the employees and letting them know that
there is no repercussion for having negative results. Management’s role in this is to give people
that are making more mistakes than others the tools to get back up to speed. If there is no buy in
from the employees, they will be tempted to falsify data and make the accuracy levels seem higher
than they truly are.

Page 3 of 3
Appendix

Exhibit 1 Typical Process Flow Chart

Exhibit 2 Process Flow with an extra associate for cross check
Calculation of the 3 sigma control limits.
P = Error Percentage
P Bar = Average Error Percentage for First 12 weeks
(1 – P Bar) = Average Correct Percentage for First 12 Weeks
UCL = Upper Control Limit & LCL = Lower Control Limit.
The Table for Calculating the P (Error Percentages, UCL and LCL is given below.
Exhibit 3 Control Limits
Week No Sample Size Error P P Bar UCL LCL
1 300 18 0.0600 0.0522 0.090756 0.013688
2 300 15 0.0500 0.0522 0.090756 0.013688
3 300 18 0.0600 0.0522 0.090756 0.013688
4 300 6 0.0200 0.0522 0.090756 0.013688
5 300 20 0.0667 0.0522 0.090756 0.013688
6 300 16 0.0533 0.0522 0.090756 0.013688
7 300 16 0.0533 0.0522 0.090756 0.013688
8 300 19 0.0633 0.0522 0.090756 0.013688
9 300 20 0.0667 0.0522 0.090756 0.013688
10 300 16 0.0533 0.0522 0.090756 0.013688
11 300 10 0.0333 0.0522 0.090756 0.013688
12 300 14 0.0467 0.0522 0.090756 0.013688
13 300 21 0.0700 0.0522 0.090756 0.013688
14 300 13 0.0433 0.0522 0.090756 0.013688
15 300 13 0.0433 0.0522 0.090756 0.013688
16 300 13 0.0433 0.0522 0.090756 0.013688
17 300 17 0.0567 0.0522 0.090756 0.013688
18 300 17 0.0567 0.0522 0.090756 0.013688
19 300 21 0.0700 0.0522 0.090756 0.013688
20 300 18 0.0600 0.0522 0.090756 0.013688
21 300 16 0.0533 0.0522 0.090756 0.013688
22 300 14 0.0467 0.0522 0.090756 0.013688
23 300 33 0.1100 0.0522 0.090756 0.013688
24 300 46 0.1533 0.0522 0.090756 0.013688
25 300 10 0.0333 0.0522 0.090756 0.013688
26 300 12 0.0400 0.0522 0.090756 0.013688
27 300 13 0.0433 0.0522 0.090756 0.013688
28 300 18 0.0600 0.0522 0.090756 0.013688
29 300 19 0.0633 0.0522 0.090756 0.013688
30 300 14 0.0467 0.0522 0.090756 0.013688