Project business Case :
PDI is package pickup and delivery company which OUTSOURCES its ground operations to a third party
Inefficiency in processes and miscommunication at PDI -> rebates and idle time costs
Gross margin needs to be improved to sustain operations and for growth in future
Problem Statement :
Identify causes behind current inefficiencies (rebate and idle cost) impacting gross margin.
2. DEFINE – Project definition
Project business Case :
• PDI is package pickup and delivery company which OUTSOURCES its ground operations to a third party
• Inefficiency in processes and miscommunication at PDI -> rebates and idle time costs
• Gross margin needs to be improved to sustain operations and for growth in future
Problem Statement :
Identify causes behind current inefficiencies (rebate and idle cost) impacting gross margin.
Goal: to suggest improvements in the system for achieving at least 30% profit before tax with minimum errors and
defects
Out of scope
• Market-research data for the purpose of increasing market share.
• Improvement in Customer satisfaction metrics other than rebate.
• Details of an employee reward system for incentivizing employees.
3. DEFINE- CTQ
Characteristics of product or
service
On time pickup/delivery , no damage to package , convenience of
scheduling
Measures and operational
definitions
Rebates due to lateness or damage to package
Target value of measures Zero rebate
Specification limits Time window of 15 mins and undamaged package
Defect rate of measures Reduce rebate occurrence from 33.3% per total pickups to 5% or less.
*Idle time is an error in the process and an inefficiency between PDI and third party
*Rebate is a defect
4. 5. Suppliers 4. Inputs 1. Process 2. Outputs 3. Customers
sender Sender order details
order received from customer(package
sender)
Order details (name,
address, contact no.)
PDI
third party
dispatcher
Time window from
dispatcher
sender informed on estimated time
window
Estimated 15 min time
window for sender
sender
sender
Payment
Confirmation
pickup of package
Pickup confirmation from
sender
Third-party dispatcher/field
operator
Sales operator at
PDI
revenue calculation
bill generated and package recipient
informed on 15 mins time-window
Bill , estimated 15 min
time window for recipient
sender and recipient
Recipient
address details of
recipient
package delivered to recipient
Receipt confirmation from
recipient
Third-party dispatcher/field
operator
DEFINE- SIPOC
6. DEFINE – Assumptions
• Field operators are always available.
• Time window is negotiated between sender and Sales Operator and is an input to
dispatcher.
• Performance of field operator is out of our control.
• Package pickup and delivery are 2 isolated processes and we need to focus only on
package pickup.
• Payment issues from sender are not be considered.
7. Define – Initial business state
Sigma Measurement
Defects ( rebate occurrences) 706
Opportunities 2096
Defect Opportunities per unit 1
DPMO 336832.0611
Sigma Level 1.9
Financials
Revenues $125,335.40
Fields ops cost ( including Idle time cost) $42,753.90
Rebate cost $38,106.00
Gross margin $44,475.50
Fixed costs $35,000.00
Profit before tax $9,475.50
Profit before tax ( % of revenue) 7.6%
8. MEASURE – Data Collection requirements: Planning Sheet
Question To Be Answered Name of Data Required Operational Definition
how often are we early ? early frequency
Minutes To Customer (One Way < Trip Time Lower Spec
Limit
how often is rebate due to lateness ? late frequency
Minutes To Customer (One Way > Trip Time upper Spec
Limit
By how much is the dispatcher overestimating
the amount of forecast time in case of early
errors ?
amount of early error
(one-way distance*60)/ forecast speed))+7.5 mins –
(minutes to customer )
by how much is the dispatcher
underestimating the amount of forecast time
in case of late errors ?
amount of late error
(minutes to customer ) - ((one way distance*60)/forecast
speed) - 7.5 mins
Details about bike and Truck usage conditions
DMB, DMT, DEB, DET, SMB,
SMT, SEB, SET
Combined Indicator variables with rebate and idle time
9. MEASURE - Identify Sampling Bias and Measurement Problems
Sampling Bias
• Unequal weightage of data for bicycle and truck
• Unequal weightage of downtown and suburbia
• Seasonal fluctuation in data
• More/less incidents of damaged package considered than normal
• Higher proportion of data for particular range of package weight (can influence speed)
Measurement problems
• Estimation of route to be travelled by field operator
• Tracking of occurrence of a late pickup/delivery
• Variation in field ops cost with no relation to vehicle used.
10. MEASURE - FMEA
Key Process
Step or Input
Potential Failure
Mode
Potential Failure Effects
S
E
V
Potential
Causes
O
C
C
Current Controls
D
E
T
R
P
N
Actions
Recommended
Resp.
What is the
Process Step
or Input?
In what ways can the
Process Step or
Input fail?
What is the impact on the
Key Output Variables
once it fails (customer or
internal requirements)?
HowSevereisthe
effecttothe
customer?
What causes the
Key Process
Step or Input to
go wrong?
Howoftendoes
causeorFMoccur?
What are the
existing controls
and procedures that
prevent either the
Cause or the Failure
Mode?
Howwellcanyou
detecttheCauseor
theFailureMode?
What are the
actions for
reducing the
occurrence of the
cause, or
improving
detection?
Who is
Responsible for
the
recommended
action?
Dispatch time
calculation
dispatch time
calculation doesn’t
synchronise with
time window
rebate or idle time cost.
10
forecast speed
under/overestima
tion
10
none
10 1000
Estimate dispatch
time with accuracy
dispatcher
selection of
bike or truck
vehicle selected is
wrong as per time
window , location,
distance and time of
day considerations
rebate or idle time cost.
8
error by
dispatcher
4
none
10 320
dispatcher
package pickup package lost or
damaged
rebate
10
Error by Field
operator
2
none
10 200
field operator
11. ANALYZE – Data set 1-3
• No rebate due to damage
• regression reveals that
late rebate mainly due to
trucks used in Downtown
during morning
Increase reliance on bikes
in downtown morning
during morning
12. ANALYZE – Data set 1 to 3
• Truck performance improved
• Overall rebate increased
• Gross margin decreased
• Current formula for dispatch
mins calculation is:
(time to reach – 7.5) mins
Dispatch mins need to be increased to
reduce lateness. Lateness is more frequent
than idle time occurrence.
Improvement 1) (time to reach + 7.5) mins
Improvement 2) (time to reach + 15 ) mins
13. ANALYZE – data set 1 to 3
• Rebates decreased, GM improved
• No trucks used in evenings
• High rebates during bike usage in
morning
Dispatch time can be further increased
to (time to reach + 31 mins)
14. ANALYZE – Data set 1 to 3
• All lateness occurrences
are in morning and
mainly due to bike
bike
downtown suburbia downtown & suburbia
AM 25 25 20 - 0.3*(pounds)
PM 15 20 20 - 0.2 * (pounds)
FORECAST SPEEDS
truck
bike
downtown suburbia downtown & suburbia
AM 20 20 20 - 0.3*(pounds)
PM 15 20 20 - 0.2 * (pounds)
FORECAST SPEEDS
truck
15. ANALYZE – dataset 4 – 6
• Data set 4-6 consisted of
100% bikes as vehicle used
• Over 90% of the pickups
had idle time errors
• We focused only on
optimizing forecast speed
for bike and learnt that we
need to monitor both
dispatch mins and forecast
speed
• In data 6 Idle time error was
drastically reduced by
increasing dispatch time to
(miles/speed)*60 + 31 mins
16. ANALYZE – dataset 4 – 6
• But this was an
impractical
condition, so we
ignored dataset 4-6
for further analysis
17. ANALYZE – Data set 7 to 11
bike
downtown suburbia
downtown &
suburbia
AM 25 25 20 - 0.35*(pounds) (miles/forecast speed)*60 + 5
PM 15 20 20 - 0.10 * (pounds) (miles/forecast speed)*60 - 5
forecast speed
truck dispatch mins
bike
downtown suburbia
downtown &
suburbia
AM 15 15 20 - 0.40*(pounds) (miles/forecast speed)*60 + 5
PM 15 15 20 - 0.10 * (pounds) (miles/forecast speed)*60 - 5
forecast speed
truck dispatch mins
Rebates are quite low but idle costs are
hitting margins now, mainly due to trucks
18. ANALYZE – Data set 7 to 11
Total profit improved a bit to 26.4%, but
idle time cost still remains high
bike
downtown suburbia
downtown &
suburbia
AM 25 25 20 - 0.35*(pounds) (miles/forecast speed)*60 + 11
PM 15 20 20 - 0.10 * (pounds) (miles/forecast speed)*60 - 5
forecast speed
truck dispatch mins
19. ANALYZE – Data set 7 to 11
bike
downtown suburbia
downtown &
suburbia
AM 30 30 20 - 0.45*(pounds) (miles/forecast speed)*60 + 8
PM 15 20 20 - 0.15 * (pounds) (miles/forecast speed)*60 - 5
forecast speed
truck dispatch mins
profit before tax didn’t change
enough. Idle time cost for individual
scenarios needs to be examined.
Forecast speed can be increased,
dispatch mins can be decreased
20. ANALYZE – Data set 7 to 11
dispatch mins
bike
downtown suburbia downtown &
AM 27 30 20 - 0.35*(pounds) (miles/forecast speed)*60 + 7
PM 15 20 20 - 0.2 * (pounds) (miles/forecast speed)*60 - 5
FORECAST SPEEDS
truck profit before tax didn’t change
enough. Idle time cost for individual
scenarios needs to be examined.
Forecast speed can be increased,
dispatch mins can be decreased
21. ANALYZE – Data set 7 to 11
Profit margin is 34 %
Sigma level is 3.8
Higher sigma level is desired since
target for profit margin has already
been achieved Profit margin is 37%
Sigma level is 3.4
23. ANALYZE – Overall progress
Reduce lateness occurrence and rebates reduce idle time cost to improve profit
24. ANALYZE – INSIGHTS
Idle Time Lateness
WHEN ? Speed of field operator is underestimated Speed of field operator is overestimated
HOW ?
Forecast Speed < Actual Speed
OR
Dispatch mins > Mins to customer
Forecast Speed > Actual Speed
OR
Dispatch mins < Mins to customer
SO ?
Dispatch mins OR Forecast Speed Dispatch mins OR Forecast Speed
BUT,
Reducing dispatch mins more than required can
cause lateness
Increasing dispatch mins beyond certain limit can
cause idle time
15 MINS TIME WINDOW
field operator
dispatcher
early late
Minutes to customer with actual speed
Dispatch minutes calculated using forecast speed
Lower
Spec
limit
Upper
Spec
limit
28. CONTROL - Non-Statistics Process Control
• Standardized Operation Procedures –
• Dispatcher need to follow our final rule, as below, to calculate bike estimated speed and
dispatch time
• Field Operator must follow suggested route from Dispatcher
• Documentation –
• What kind of documentation is available? Improved Process Map, Mainframe process
documentation.
• Where is the documentation located? Change Management folder on shared drive
• Who has access to the information? Change management team
• Who will be responsible for updating the information? Change Management Team
• How is documentation / file change control managed? Change management team updates
the version changes in the document.
29. CONTROL - Non-Statistics Process Control
• Poke Yoke –
• Before dispatching field operator, we have to check there is parking spot at
destination for truck delivery.
• Demanded 3rd party regularly implement bike and truck maintenance.
• Routes should be decided by dispatcher and followed by field operator.
• Monitoring –
• Conducted a regular monthly meeting to review our gross margin, price rebate
and idle time cost, and perform root-cause analyses.
31. CONTROL - Idle Time Control Chart
Idle Time Control Chart 5% outliers
are above UCL
32. CONTROL - Idle Time Control Chart
1% outliers
are above UCL
33. Summary
Proposed system should regularly meet these criteria
• Profit before tax should be above 30% of revenue
• Gross Margin should not be lower than LCL ($1392)
• Not more than 5% of idle time outliers should be above UCL (13.76 mins)
• Not more than 1% of lateness occurrences(defects) should be above UCL (0.173 mins)
Separate tracking mechanisms for bike and truck
Truck speed doesn’t depend on weight
Based on our detailed analysis we realized that we can use the same formula for calc. bike forecast speed in either case
Data 10 meets both criteria – increasing profit before tax and reducing rebates
Sigma level increased from 1.9 to 3.8
Gross margin has improved from 7% to 34%
Rebate occurrences have reduced to around 1%
Although idle time occurences have remained fairly same, the effect of them have reduced