• Applied DMAIC technique to support staffing team of Tesla motors, by reducing time to hire contingent workers
• Root cause analysis was done to identify the worker shortage in production team and later improvements methods were applied, which led to efficiency in hiring from 44.85-71.77%
• Tools used: Design: Charter, Pareto, CTQ; Measure: SIPOC, Fish bone/Ishikawa, Statistics, XY-diagram; Analyse: Root cause analysis, FEMA, Hypothesis; Improve: 5 Why’s, VSM; Control: SPC or control chart
1. Improving the Hiring Process for
High Volume Production
ISE 250 Project
Professor Steele
May 2016
Compiled by:
Melanie Gameng
Navathej Gobi
Gautam Singh
Solomon Arun Kumar Thangaraj
2. 2
Contents
Abstract........................................................................................................................................... 3
Background..................................................................................................................................... 3
Scope............................................................................................................................................... 4
DMAIC ........................................................................................................................................... 5
Improvement Opportunity: Define Phase ................................................................................... 5
Hiring Process Flow Chart ...................................................................................................... 5
SIPOC Diagram....................................................................................................................... 6
Current State of the Process: Measure Phase.............................................................................. 7
Pareto Charts: Major causes of failure .................................................................................... 9
Histogram .............................................................................................................................. 10
Project Metric: Z-Score (Sigma score).................................................................................. 13
Value Stream Map................................................................................................................. 15
Analysis and Findings: Analyze Phase ..................................................................................... 16
Ishikawa/Fishbone Diagram.................................................................................................. 16
5 Whys Diagram.................................................................................................................... 17
Recommendations: Improve Phase........................................................................................... 18
Z-test...................................................................................................................................... 18
Monitoring and Control: Control Phase.................................................................................... 19
Improvement Methods........................................................................................................... 19
Summary/Conclusion.................................................................................................................... 20
Appendix....................................................................................................................................... 21
3. 3
Abstract
An energy/car manufacturing company growing rapidly with an increasing demand for its vehicles.
Several departments across the company play an integral role in delivering the cars such as
engineering, production, etc. This project is focused on production – direct/indirect labor. The
purpose of this project is to improve the time and quantity in the hiring process for contingent
workers to satisfy the production shops requirements. Upon applying the 6-Sigma DMAIC
problem solving process, we suggest that the hiring process could be improved in terms of time
taken and quantity, by implementing weekend open houses and on the spot testing/interviews.
Background
The energy/automobile company’s manufacturing site is located in Fremont, California since May
2011. The CEO’s plan was to introduce a car model that was well-designed from an aesthetic
stand-point, technology driven and most importantly all-electric to prove to the world, an electric
car does not have to be ugly, as well as the first step to advancing the sustainable transportation.
Part 2 was to release a luxury sedan $65K and a family CUV $120K and to have the profits be
invested into the research and development into the part 3. Part 3, is to release a more economical
vehicle for the masses with a base price of $35k. In order to achieve the company’s goals, cars
need to be delivered to the customers as fast as possible.
Our project is supporting the Staffing team in reducing the time it takes to get contingent workers
out to the production floor to build the vehicles. The assembly line is less than 5 years old and
fairly complex in its nature and having many capabilities. The largest volume of employees at the
company consists of the direct and indirect labor employees. These are the people who are
physically putting the vehicle together and warehouse personnel. As the different departments
4. 4
from the line need temporary work force to achieve vehicle output goals, they often issue short
time notice (usually days) and the requests can be 100+ quality people a week.
The Staffing department is using 3 agencies to provide the temporary staff. All 3 agencies use the
same hiring process from pre-screening the candidates to testing to interviews to orientation.
Currently the agencies are requested to provide the workforce of 120 qualified persons/week.
However, these agencies were not able to achieve the target and leading to inadequate labor.
Any direct labor workforce gaps can be costly to the company. This project identifies several areas
of improvement to the contingent worker hiring process to increase the number of temporary
employees in the pipeline (ready-to-go) as requested by the business and reduce the time it takes
to identify these new skilled temporary employees. Ultimately, the process can be scaled to any
future direct labor needs.
Scope
Having learned that the problem faced by an industry is inadequate workforce in production, the
main scope of this project will be developing an improved method and implementing it efficiently
to achieve the target of hiring the required number of persons in a timely manner. We analyzed the
existing process and found some root causes. Then, we designed few improvements like
conducting interviews on the same day, creating a database pool (pipeline), etc. (will be explained
in the later sections) to the existing process and measured the efficiency of improvement using x-
y graph, histograms, box plot, and pareto charts. These processes will be discussed in detail in the
following sections.
5. 5
DMAIC
Improvement Opportunity: Define Phase
Hiring Process Flow Chart (Swim Lane Diagram)
Figure 1. Above is the hiring process flow. A larger chart can be found in the appendix.
Currently, the temporary workers are hired through the three primary staffing agencies. The
production shops in factory site raise a notice to HR department that the ‘x’ number of persons are
required to accomplish the job on time. Upon reviewing, HR requests these staffing agencies to
hire them a specified number of persons. Then, the staffing agencies advertise the job opening
through different mediums such as posting on their job board, flyers, word of mouth, email, job
fair and events. Then, the candidates are pre-screened to evaluate their eligibility. After clearing,
the candidates are asked to take skills assessment test, where they will be tested on their abilities
specific to job. After, the candidates who pass the test undergo a face-to-face interview and
evaluated further. If the candidates were selected after interviewing, background checks get
6. 6
initiated and reviewed. Once the candidates meet hiring the hiring criteria, the next step for the
candidate is to participate in a pre-employment physical to ensure safety at the workplace, such as
proper lifting, able to stand for long hours, etc. The successful candidates are then given a start
date and assigned to specific department. If the candidates fail in any of the above steps, they are
able to reapply after 6 months. The overall process flow is depicted in Figure 1.
SIPOC Diagram
The SIPOC diagram maps the scope of the project and helps to identify the potential gaps between
the supplier, input specifications, between the output specifications, and the customer’s
expectations that helps to identify the scope of process improvement. It analyses the feedback and
feedforward process flow between the Supplier, Inputs, Process, Output, and Customers.
In the current hiring process before the implementation of process improvements, it included
significant delays in the pre-screen process to identify potential employees taking approximately
4 to 5 days before a candidate is invited for taking the assessment test and interview.
Figure 2. SIPOC before improvement process
7. 7
After the SIPOC mapping and analysis, the following improvement methods were implemented in
the respective categories.
Supplier- Lay more emphasis on employee referrals, conducting open house events frequently
and thus encouraging more on-the-spot interviews and targeting associate level students at
community colleges who are flexible in work schedule and have passion for Tesla.
Process- Improving more number of testing stations and recruiters to accommodate more
candidates in less time.
Figure 3. SIPOC after improvement process
Current State of the Process: Measure Phase
Control chart helps us to monitor the process changes over time. Control charts are of two types:
(A) Control chart for continuous data and (B) Control chart for discrete data.
8. 8
In our study, the process change over time was the number of employees hired per week. The
employees which were qualified but failed to show up (i.e. join manufacturing team) were
considered as defects. Also, the number of employees hired every week could be considered to be
variable set of data as different number of employees were hired every week. Therefore, the data
set contains discrete and variable data and as a result, P-chart was plotted (refer to figure 4 below).
Using P-chart, we can count the total number of defects.
From the Figure 4 (A), we can see that the control limits (UCL and LCL) are not the straight line
which is due to the fact that the sample size (total people who appeared for interview) that we
chose was varying over time. It can be noted that the sample 2 and sample 5 are outside control
limits and other samples show a lot of variability. Therefore, the data set is not a good fit.
Figure 4 (A): Control Chart (P-Chart) for hiring process (before improvement process)
By applying the improvement phase, we can see that the values lie within control limits as shown
in Figure 4 (B). In addition, there are less overall variations among the data points
comparatively.
9. 9
Figure 4 (B): Control Chart (P-Chart) for hiring process (after improvement process)
Pareto Charts: Major causes of failure
As shown in figure 5, the Pareto chart shows time required by different categories of hiring process.
As we can see from the graph that the first two regions i.e. Third party and Agency together
constitute more than 90% of the total hiring time.
Figure 5: Pareto Chart of causes of low hiring process
10. 10
One of our suggestions for the improvement in hiring time was to organize the pre-screening and
skills assessment test in addition to job fair on weekends. Through this method, agency would be
able to attract more people. Also, by conducting the pre-screening and skills assessment test on
the same day as open house, the agency could expedite the hiring process and send the results to
the company at the earliest. Based on these results, the company can invite the successful
candidates for the interview on the following weekday. Through this improved process, the time
taken for hiring a person could be reduced considerably. The Pareto chart was developed after the
improvement phase as shown in figure 6 and it can be noticed that the time taken by the agency to
conduct open house, pre-screening test, and skills assessment has been reduced from 34.9 % to
21.2 %.
Figure 6: Pareto Chart after improvement process
Histogram
To study the improvement of the hiring process, the number of persons hired per week was
monitored across eight weeks i.e. 3/4/2016 – 5/2/2016. We introduced the improvement methods
to hiring process as stated earlier at end of week 4 i.e. 3/28/2016. Table 1 shows the number of
persons hired across the eight weeks.
11. 11
Table 1. Number of persons hired over the monitoring period (3/4/2016 – 5/2/2016).
Week
Number of persons
hired
1 61
2 85
3 112
4 78
5 97
6 136
7 145
8 202
To understand the distribution and the pattern of hiring process, the histogram was plotted for 8
weeks as shown in Figure 7. It can be seen that 3 weeks resulted in hiring of 80 – 100 persons per
week, 2 weeks each for 40 – 80 persons per week and 120 – 160 persons per week, and 1 week
yielded 200 – 240 persons per week.
Figure 7. The distribution of number of persons hired per week.
12. 12
To get better insight on the improvement process, the number of persons hired was plotted against
the weeks as shown in Figure 8. It was observed that the after the changes in process, we were able
to see the improvement in the hiring process. The mean of number of persons hired before the
improve phase was 84 with the standard deviation of 21.21. After the improvement phase, the
mean and standard deviation was found to be 145 and 43.33 respectively. The efficiency of our
improvement process was estimated by calculating the percent increase in the number of persons
hired and was found to be 72.62 %.
Figure 8. Plot of number of persons hired versus the respective weeks.
Like the X-Y graph, the bar graph was drawn to see study the impact of improvement process. As
shown in figure, the increase in the hiring process can be noticed. The first four weeks yielded the
output fewer than the target set at 120. Although we could not notice the immediate improvement
in number of persons hired, the subsequent increase in the process was evident from the figure 9.
13. 13
Figure 9. Bar graph of number of persons hired over the period of eight weeks.
Project Metric: Z-Score (Sigma score)
With the help of Z-score, we can measure the statistics of the hiring process. By calculating the Z-
score for hiring process before and after the improvement phase, we can evaluate the effectiveness
of improvement on the hiring process.
We can determine whether the number of people hired during first four weeks were below mean
or above mean or equal to mean through the Z-test. Similarly, we can use Z-score to determine the
number of people hired during the improvement phase. Refer to table 7 in appendix for more Z-
score details.
Since the target value to hire total employee every week was 120. Considering before and after
improvement phase, (Refer to Table 5 (a, b, c), we can calculate the Z-score.
𝑍 =
𝑋−𝑋̅
𝜎
- to calculate Z-score for entire population
𝑍 =
𝑋−𝑋̅
𝑆
- to calculate for sample (subset of the population)
14. 14
Using above equations we derive,
𝑍 𝐵𝑒𝑓𝑜𝑟𝑒 =
𝑋 − 𝑋̅
𝑆
𝜎 =
84 − 120
21.21
𝜎 = −1.697𝜎
𝑍 𝐴𝑓𝑡𝑒𝑟 =
𝑋 − 𝑋̅
𝑆
𝜎 =
145 − 120
43.33
𝜎 = 0.576𝜎
Figure 10: Z-Score (Sigma Score) Before Improvement Process
From figure 10, it can be seen that the Z-score is negative (Z= -1.697). This implies that the mean
of the number of people hired before the improvement is 1.697 times the standard deviation below
the target mean of 120. The probability of hiring (or the area) is equal to 44.85%
15. 15
Figure 11: Z-Score (Sigma Score) After Improvement Process
From Figure 11, it can be seen that the Z-score is positive (Z=0.576). This implies that the mean
of the number of people hired after the improvement phase is 0.576 times the standard deviation
above the target mean of 120. The probability of hiring (or the area) is equal to 71.77% which was
comparatively better than the previous hiring process.
Value Stream Map
We have reduced the total non-value added as indicated from the table. Before and after value
stream map can be found in Appendix.
Table 2. Summary of Value Stream Mapping for before and after improvement phase.
Before After Difference
Total Non-Value-Added Time 433 hrs or
18 days
77% 386 hrs or
16 days
75% 48 hrs or
2 days
Total Value-Added Time 127 hrs or
6 days
23% 127 hrs or
6 days
25% 0
Total Lead Time 560 hrs or
24 days
513 hrs or
22 days
48 hrs or
2 days
16. 16
Analysis and Findings: Analyze Phase
Ishikawa/Fishbone Diagram
Figure 12. Fish Bone Diagram to determine Root Cause analysis.
Fishbone diagram is used to determine the root causes for low hiring of employees.
17. 17
5 Whys Diagram
The 5-WHY tool is used to find specific and systemic root causes of a problem. It can also be
used to estimate the various causes which lead to failure rate.
Figure 13: The 5-WHY Diagram to estimate the detection of various Root Causes
By applying 5-Why, the various causes of failure could be linked to the root causes, which were
estimated by Fishbone diagram in Figure 12.
18. 18
Recommendations: Improve Phase
Table 3.1. The chronological table of different improvement suggestions over the time.
0-3 months 3-6 months 1 + years
Increase weekly meeting
Reduce Attrition
Improve Retention
Targeted Training for
Operators
Standardize
Process/Procedure
Improving
screening test
Computerized testing systems
Improve Forecasting model
Z-test
The Z-test was executed to evaluate the mean, standard deviation, and 95% confidence interval
of hiring process before and after the improvement phase. From the table 3.2, it can be stated the
implementation of improvements has increased the number of persons hired per week by 72.6 %.
Table 3.2. Summary of z-test showing the parameters before and after the implementation
of improvement process.
Parameter Before improvement
phase
After improvement
phase
Mean 84 145
Standard Deviation 21.21 43.33
Lower 95% CI 50.24 76.04
Upper 95% CI 117.75 213.96
19. 19
Monitoring and Control: Control Phase
Improvement Methods
To study the effect of improvement phase statistically, the box plot was plotted for before and after
the improvement. As seen in figure 14, the shift in the median, first quantile, and third quantile
suggests that the efficiency of hiring process has increased through the implement of the
improvement phase. In addition to shift in their quantiles, the efficiency of hiring process was well
below the target line and improved later. It is believed that the box plot could be plotted in the
upcoming weeks to study the variability.
Figure 14. Box plot providing the changes in hiring process statistically.
20. 20
Summary/Conclusion
The hiring process of contingent employees has been significantly improved after the
implementation of the suggested process improvements. The reduction in time is taken from the
pre-screen interviews to the day a contingent work starts. The agencies are also able to manage a
larger candidate pool in the reduced time developing a steady pipeline of candidates who are on
standby as the businesses request additional headcount or backfills due to attrition. This improved
process gives the Staffing department the ability to scale the hiring process for future business
need ramp ups and deliver a higher quantity of manpower (if necessary) to the assembly line in a
shorter turnaround time. In the future the potential aspects to focus on are identified as improving
the forecasting methodology, increasing the frequency of weekly meeting to identify better the
need for employees, reduction of attrition rate and improving the retention rate.
24. 24
Table 4:
Whole Data
Number of persons hired
N total 8
Mean 114.5
Standard Deviation 45.39666
Lower 95% CI of Mean 76.54748
Upper 95% CI of Mean 152.45252
Sum 916
Minimum 61
1st Quartile (Q1) 81.5
Median 104.5
3rd Quartile (Q3) 140.5
Maximum 202
Interquartile Range (Q3 - Q1) 59
Range (Maximum - Minimum) 141
Table 5 (A): Before Improvement Phase Table 5 (B): After Improvement Phase
Before Data: After Data:
N total 4 N total 4
Mean 84 Mean 145
Standard Deviation 21.2132 Standard Deviation 43.3359
Lower 95% CI of Mean 50.24503 Lower 95% CI of Mean 76.04286
Upper 95% CI of Mean 117.75497 Upper 95% CI of Mean 213.95714
Sum 336 Sum 580
Minimum 61 Minimum 97
1st Quartile (Q1) 69.5 1st Quartile (Q1) 116.5
Median 81.5 Median 140.5
3rd Quartile (Q3) 98.5 3rd Quartile (Q3) 173.5
Maximum 112 Maximum 202
Interquartile Range (Q3 - Q1) 29 Interquartile Range (Q3 - Q1) 57
Range (Maximum - Minimum) 51 Range (Maximum - Minimum) 105
25. 25
Table 5 (C): Deduction from Table 5 (A) & 5 (B)
Before After
Total Hiring
Target
Mean 84 145
120
Standard Deviation 21.21 43.33
Table 6:
Sub-group size (n)
Run
Total People
Interview
People Hired
No Show Up
(Defects)
Defects per Unit
(DPU)
1 49 43 6 0.12244898
2 25 18 7 0.28
3 55 49 6 0.109090909
4 39 36 3 0.076923077
5 48 48 0 0
6 80 64 16 0.2
7 40 39 1 0.025
8 40 39 1 0.025
9 31 28 3 0.096774194
10 83 69 14 0.168674699
11 57 49 8 0.140350877
12 90 87 3 0.033333333
13 60 59 1 0.016666667
14 86 80 6 0.069767442
15 164 135 29 0.176829268
16 92 85 7 0.076086957
Total 1039 928 111
Average DPU (Ū)=111/1039= 0.107=Centreline (CL)
26. 26
Table 7: Z-Score Description
Z-Score Description
Values Explanation
0 it is equal to the group mean
Positive above the group mean
Negative below the group mean
1 it is 1 Standard Deviation above the mean
2 it is 2 Standard Deviations above the mean
-1 1 Standard Deviation below the mean
-2 it is 2 Standard Deviations below the mean