Leveraging Analytics for

Businesses
February 2015

Agenda
Introduction
Analytics: Future of Business Consulting
Business Cases: Examples from PwC US Advisory

How to enhance your analytical skills

How to pitch your analytical skills
Questions

PwC

February 2015
2

Analytics: Future of Business

Consulting

PwC

February 2015
3

The Top 6 Tech Skills You Need in 2015

The article appeared in the business magazine Inc. on 27th Jan

Coding

Big Data
Cloud Computing
Mobile

Data Visualization
UX Design Skills
PwC

February 2015
4

The need of analytics

Method 1: India is a developing country

Method 2: India is a developing country

PwC

February 2015
5

Theoretical Analytics vs. Application Analytics

Theoretical Analytics: Research in fields of topics like machine learning,
optimization, text mining, data visualization, regression, PCA, regression,
decision trees, linear programming etc.
Application Analytics: Solving problems of clients

Finding how many people are going to attrite from an organization in next 1
year: regression modeling

Clustering products from an inventory to optimize their transportation on

basis of size, volume, dimensions

Finding sentiments of users about a particular bank from social platforms

like Facebook: text mining and sentiment analysis

Analyst: Theoretical Analytics

PwC

World: Application Analytics

February 2015
6

Knowledge Transfer b/w The Two Domains

People in Theoretical Analytics

People in Application Analytics

Professors

PwC Diamond

Researchers

Fractal Analytics

IEOR

Google Analytics

Institutes (MIT, Stanford)

McKinsey & Co.

Opera Solutions

EXL Inductis

PwC

February 2015
7

Business Cases

Case 1: HR Analytics Predicting

Attrition for FY 14-15

PwC

February 2015
8

HR Analytics| Phase 1 - Program design

Outcomes:

Project plan with deadlines and responsibilities

Hypotheses inventory and mapped data elements

Hypotheses inventory
Area
Area
Current and Past Managers
Area
Current and Past
Managers
Current
andand
PastPast
Managers
Current
Managers
Current
and Past
Managers
Current
and
Past
Managers
Currentand
and
PastManagers
Managers
Current
Past
Current and Past Managers
Demographics
Demographics
Demographics
Demographics
Demographics
Education/
Credentials
Demographics
Education/
Credentials
Education/ Credentials
Education/ Credentials
Education/ Credentials
Education/ Credentials
Engagement Levels
Engagement Levels
Engagement
Levels
Engagement
Levels
Engagement
Levels
Engagement Levels

Hypotheses
Hypotheses
Current/past manager/coach Hypotheses
attrition will lead to attrition
Current/past manager/coach attrition will lead to attrition
Time
under the same
manager may
affect will
attrition
Current/past
attrition
lead to attrition
Time
under themanager/coach
same manager may
affect attrition
Number
of
employees
churned
under
the
manager
will affect
Time under
the samechurned
manager
may
affect
attrition
Number
of employees
under
the
manager
will affect
attrition
Number
of
employees
churned
under
the
manager
will affect
attrition
Tenure
will have a U curve for attrition
attrition
Tenure
will have a U curve for attrition
International
employees
have
higher
attrition
Tenure will have
a U curve
attrition
International
employees
havefor
higher
attrition
PartInternational
time education/MBA/credentials
in last
cycle will lead to
employees have higher
attrition
Part time education/MBA/credentials
in
last cycle will lead to
attrition
Part time education/MBA/credentials in last cycle will lead to
attrition
Specific
academic
background
/institutes
will
have higher
attrition
Specific
academic background /institutes will have higher
attrition
Specific academic background /institutes will have higher
attrition
Lowattrition
engagement score lead to attrition
Low engagement score lead to attrition
Employee
engaged inscore
an industry
having higher growth
Low engagement
lead to vertical
attrition
Employee
engaged in an industry
vertical having higher growth
willEmployee
have higher
attrition
propensity
jobhaving
availability)
engaged
in an
industry(more
vertical
higher growth
will have higher
attrition
propensity
(more job
availability)
will have higher attrition propensity (more job availability)
Filters
Attrition is higher for some particular vertical/ business unit
Filters
Attrition is higher for some particular vertical/ business unit
Filters
Some
officesiswill
havefor
higher
Filters
Attrition
someattrition
particular
Filters
Some
offices higher
will have
higher
attrition vertical/ business unit
Filters
Smaller
higher
attrition
Filters
Someoffices
officeshave
have higher
attrition
Filters
Smaller
officeswill
have
higher attrition
Hours
worked and vacation time High
one-offoffices
vacations
to past may lead to churn
Filters
Smaller
havecompared
higher attrition
Hours worked and vacation time High one-off vacations compared to past may lead to churn
Hours worked and vacation time High one-off vacations compared to past may lead to churn
Hours worked and vacation time Overall number of working hours would have higher attrition
Hours worked and vacation time Overall number of working hours would have higher attrition
Hours worked and vacation time Overall number of working hours would have higher attrition
Hours worked and vacation time Change in working hours compared to past year would have
Hours worked and vacation time Change in working hours compared to past year would have
Hours worked and vacation time
Change
in working hours compared to past year would have
impact
on attrition
impact on attrition
impact
on attrition
Hours worked and vacation time Change
in working
hours compared to peers would have impact
Hours worked and vacation time Change in working hours compared to peers would have impact
Hours worked and vacation time
Change in working hours compared to peers would have impact
on attrition
on attrition
on attrition
Hours worked and vacation time Change
in health condition will lead to attribution
Hours worked and vacation time Change in health condition will lead to attribution
Hours worked and vacation time Change in health condition will lead to attribution
Hours worked and vacation time Sabbatical will lead to higher attrition
Hours worked and vacation time Sabbatical will lead to higher attrition
Hours worked and vacation time Sabbatical will lead to higher attrition
Hours worked and vacation time Leave utilization will affect attrition
Hours worked and vacation time Leave utilization will affect attrition
Hours worked and vacation time Leave utilization will affect attrition

PwC

Mapped data elements

44 Onboarding survey All onboarding survey questions
onboarding
45 44 Onboarding surveyDateAllsurvey
taken survey questions
45
Date survey taken
46 Company
joined
Company
Company
47 46 Company joined Company
Location
Company
Location
48 47
Company
Industry
49 48
RoleCompany Industry
Role
50 49
Salary
Salary
51 50
Level
Category
Data
Elements
Level
52 51
Function
ID
EmplID
52 Client status
53 Post
Left Function
for competition?
ID
Scrambled
ID
Left for to
competition?
54 53 Post Client status
work as a contractor?
Demographics
Job Continued
Title
54
Continued to work as a contractor?
55
SexLeft for job in Federal sector?
55
Left
for job in Federal sector?
56 Filters
Role
Race
56
Filters
Role
57
Sub Role
Country
of origin
Role
58 57
LineSub
ofofService
Age/Date
Birth
of Service
59 58
SubLine
Level
Full/Part
Time of Service
60 59
Job Sub
CodeLevel of Service
Function
Job
Code
61 60
Level
Function
Descr
Level
62 61
Level
Descr
Client
facing
role
Level Descr
Hiring/Termination
HireRole
Date
63 62
Role
information
Status
64 63
Role(active/departed)
Descr
Descr
Termination
Date
65 64
GeoRole
Market
Geo reason
Market code
Termination
66 65
Region
Region
Interview
Interviewer
name
67 66
Country
67
Country
Performance
68 HIPO/
Pivotal
Identified
as HIPO (yes/no)
Interviewer
level
68
HIPO/
Pivotal
Identified
as HIPO
(yes/no)(yes/no)
69 identification
Identified
as Pivotal
employee
Interviewer
rating
Identified
as Pivotal
employee
(yes/no)
pay as HIPO
70 69 identification Starting
Identified
previous
years (yes/no)
Identified
as HIPO
previousprevious
years (yes/no)
Sign-on
bonusas Pivotal
71 70
Identified
employee
years (yes/no)
71
Identified
as Pivotal
employee
(yes/no)
72 Hours
worked andRehire
Total
hours worked
current
year previous years (yes/no)
72 Hourstime
worked Intern
andTotal
Total
hours
worked
current years
year
(yes/no)
73 vacation
hours
worked
in previous
Total
hours
worked
in previous
source
74 73 vacation time Hiring
Client
hours
worked
current
year years
Client
hours
worked
current years
year
Previous
Company
75 74 company Previous
Client
hours
worked
in previous
Client
hours (yes/no)
worked
in previous
years
Direct
college
76 75
Bizfrom
development
hours worked
current
year
Biz
development
hours
worked
current years
year
Direct
grad school
(yes/no)
77 76
Bizfrom
development
hours
worked
in previous
77
Biz
development
hours
worked
in previous
Worked
as
a
contract
employee
for
Client
(yes/no)years
78
Vacation hours current year
78
Vacation
hours
current
year
Years
of
experience
in
prior
company
79
Vacation hours in previous years
79
Vacation
hours
in
previous
years
Total
years
of
experience
prior
to
joining
80
Parental Leave Hours current year
Parental
Leave
Hours
current years
year
Previous
company
salary
81 80
Parental
Leave
Hours
in previous
Leave
Hours year
in previous years
Previous
company
location
82 81
SickParental
Leave hours
current
Leave
hours
currentyears
year
Previous
company
Industry
83 82
SickSick
leave
hours
in previous
Previous
company
role
Sick
leave hours
in previous years
84 83
Current
Vacation
Balance
PreviousCurrent
company
function
Vacation
Balanceyears
85 84
Vacation
balance
in previous
Referer
Referer Vacation balance in previous years
86 85
Hours over capacity (Month by Month) current year
Current Hours
Status of referrer
capacity
(Month
by Month)
current
year
87 86
Hours overover
capacity
(Month
by Month)
previous
years
If refererHours
no longer
firm, (Month
reason by
for Month)
departure
over with
capacity
previous years
88 87
Increase/decrease
from previous
years
If refererIncrease/decrease
no longer with firm,
date of departure
88
89 Commute
distance Current
work zipcode from previous years
Hiring
Manager
Hiring manager
89
Commute
distance
Current
work
zipcode
90
Past work zipcodes
Years worked
under
hiring manager
Past
work
zipcodes
91 90
Home
zipcode
Current Home
Status of hiring manager
92 91
Old home zipcode
zipcodes
If hiring Old
manager
longer with firm, reason for departure
homeno
zipcodes
93 92
Miles commute
(if available)
If
hiring
manager
no
longer
with firm, date of departure
Miles
commute
94 93
Miles
commute
past(if(ifavailable)
available)
94
Miles commute past (if available)

can be separate data file

can be separate data file
Company joined after leaving Client
Company joined after leaving Client

List of variables from Datawarehouse/ HRIS

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
31
32
33
34
35
36
37
38
39
40
41
42
43

Comments
Client to scramble original IDs
Yes/no
Yes/no
Yes/no
Yes/no
Yes/no
Yes/no
To be
customized as per requirements
To be customized as per requirements

Yes/no

can be a separate file if multiple interviews

per employee

can be separate data file

be separate
cancan
be separate
datadata
file file
can be separate data file
can be separate data file
can
be separate
data file
Referral,
campus,
application,
etc
Company
s/hedata
worked
can bewhere
separate
file before
can be separate data file
can be separate data file
can be separate data file
can be separate data file
can be separate data file
can be separate data file
can be separate data file
can be separate data file
can be separate data file
can be separate data file
be separate
cancan
be separate
datadata
file file
can be separate data file

can be separate data file

can be separate data file
can be separate data file
can be separate data file
can be separate data file
can be separate data file

HR Analytics| Phase II Data extraction

Outcomes:

Core analytic dataset containing essential data elements

Data dictionary, data audit and hypotheses feasibility documents

44 Onboarding survey All onboarding survey questions

44 Onboarding survey All onboarding survey questions
45
Date survey taken
45
Date survey taken
46 Company joined
Company
46 Company joined
Company
47
Company Location
47
Company Location
48
Company Industry

Raw Analytic
Dataset

can be separate data file

can be separate data file
Company joined after leaving Client
Company joined after leaving Client

48
Company Industry
List
49 of variables
Role from Datawarehouse/ HRIS
49
Role
50
Salary
50
Salary
51
Level
Category
DataLevel
Elements
51
52
Function
1 ID52
EmplID
Function
53 Post Client status Left for competition?
2 ID53 Post Client statusScrambled
Left for ID
competition?
Continued
to work as a contractor?
354Demographics
Job
Title
54
Continued to work as a contractor?
Left for job in Federal sector?
45555
Sex
Left for job in Federal sector?
Filters
Role
55656
Race
Filters
Role
Sub Role
65757
Country
origin
SubofRole
Line of Service
75858
Age/Date
Birth
Line ofofService
Sub Level
of Service
85959
Full/Part
Time
Sub Level
of Service
Job Code
960
Function
60
Job Code
Level Descr
1061
Function
61
Level
Levelfacing
Descrrole
1162
Client
62
Level Descr
Role
1263Hiring/Termination Hire
Date
63
Role
Role Descr
1364information
Status
(active/departed)
64
Role Descr
Geo MarketDate
1465
Termination
65
Geo Market
Region reason code
1566
Termination
66
Region
Country name
Interviewer
1667Interview
67
Country
HIPO/ Pivotal
Identified level
as HIPO (yes/no)
1768Performance
Interviewer
68 HIPO/ Pivotal
Identified as HIPO (yes/no)
Identified rating
as Pivotal employee (yes/no)
1869 identification
Interviewer
69 identification
Identified as Pivotal employee (yes/no)
Identified
1970
Starting
payas HIPO previous years (yes/no)
70
Identified as HIPO previous years (yes/no)
Identified
as Pivotal employee previous years (yes/no)
2071
Sign-on
bonus
71
Identified as Pivotal employee previous years (yes/no)
Total (yes/no)
hours worked current year
2172 Hours worked and Rehire
72 Hours worked and Total hours worked current year
Total(yes/no)
hours worked in previous years
2273 vacation time
Intern
73 vacation time
Total hours worked in previous years
Clientsource
hours worked current year
2374
Hiring
74
Client hours worked current year
Client hours
worked in previous years
2475Previous company Previous
Company
75
Client hours worked in previous years
Biz development
worked current year
2576
Direct
from college hours
(yes/no)
76
Biz development hours worked current year
Biz development
hours(yes/no)
worked in previous years
2677
Direct
from grad school
77
Biz
development
hours
worked in previous years
2778
Worked
as hours
a contract
employee
Vacation
current
year for Client (yes/no)
Vacation
hours current
287978
Years
of experience
in prior year
company
Vacation
hours in previous
years
Vacation
in previous
years
298079
Total
years Leave
of hours
experience
prior to
joining
Parental
Hours current
year
Parental
Leavesalary
Hours current year
308180
Previous
company
Parental
Leave Hours
in previous years
81
Parental
Leave
Hours in previous years
3182
Previous
company
Sick Leave
hourslocation
current year
Sickcompany
Leave hours
current year
328382
Previous
Sick leave
hoursIndustry
in previous years
Sick leave hours
in previous years
338483
Previous
Currentcompany
Vacation role
Balance
Current
Vacation Balance
348584
Previous
company
Vacation
balancefunction
in previous years
85
Vacation balance in previous years
3586Referer
Referer
Hours over capacity (Month by Month) current year
Hours over of
capacity
(Month by Month) current year
368786
Current
referrer
Hours Status
over capacity
(Month by Month) previous years
Hours
capacity
(Month
by Month)
previous years
378887
If referer
noover
longer
with
firm,
reason
for departure
Increase/decrease
from
previous
years
Increase/decrease
from date
previous
years
388988
longer
with firm,
of departure
Commute distance If referer
Currentnowork
zipcode
89 Commute
Current work zipcode
3990Hiring
Managerdistance
Hiring
Pastmanager
work zipcodes
Past workunder
zipcodes
409190
Years
Homeworked
zipcode hiring manager
Home
zipcode
419291
Current
Status
of hiring manager
Old home
zipcodes
Old home zipcodes
429392
If hiring
with firm, reason for departure
Miles manager
commuteno
(if longer
available)
Miles commute
(if available)
439493
If hiring
longer
with firm, date of departure
Miles manager
commuteno
past
(if available)
94
Miles commute past (if available)

Comments
Client to scramble original IDs
Yes/no
Yes/no
Yes/no
Yes/no
Yes/no
Yes/no
To be customized as per requirements
To be customized as per requirements

Learning
Management
System

Yes/no

can be a separate file if multiple interviews

per employee

can be separate data file

can be separate data file
can be separate data file
can be separate data file

Recruitment
System

can be separate data file

can be separate data file
Referral, campus, application, etc
can be separate
dataworked
file before
Company
where s/he
can be separate data file

HR
Information
SystemPeopleSoft

ERP Finance
& Operations

can be separate data file

can be separate data file
can be separate data file
can be separate data file
can be separate data file
can be separate data file
can be separate data file
can be separate data file
can be separate data file
can be separate data file

can be separate data file

can be separate data file
can be separate data file
can be separate data file
can be separate data file
can be separate data file

Macroeconomic/I
ndustry data

Employee
Survey
Data

910 data fields

Historic Data from

2006

PwC

Core Analytic
Dataset

Computed
Predictors

can be separate data file

can be separate data file

Data extracted from

different
databases/surveys

652 total predictors

Including 246 derived

predictors

Core analytic dataset

created

71 individual data files

10

HR Analytics| Phase III Model development

Methodology
Phase III Model development
Outcomes:

Predictors identified, refined and validated

Final predictive model

1. Core analytics dataset

2. Variable Treatments

3. Modeling dataset

4. Data segmentation analysis

Flooring

Core Analytic
Dataset

Capping

Missing
value
imputation

Modeling dataset

10. Model validation

5. Visualization and transformation

Attrition pattern by performance based award in the current year

Lift from the model (ROC Curve)

KS value
Maximum KS of
41.55% indicate that
the model is able to
discriminate
between attrites and
non-attrites

100%

Area under the curve

is 72.5%

90%

80%

80%

70%

70%

60%

60%

50%

50%

40%

40%

30%

30%

20%

20%

10%

22.2%

20.0%

20.0%

Attrition rate

100%

90%

25.0%

% Cumulative Percentage

% Cumulative Attrite

15.0%
10.0%

5.6%

5.2%

5.0%

4.4%

4.8%

0.0%

10%

0%

0%
0%

20%

40%

%Cul mulative Attrite

60%
80%
100%
% Cumulative Non Attrite
%Culmulative Non-Attrite

0%

10%

20%

30%

40%

% Cumulative Attrite

50%

60%

70%

80% 90% 100%

% Population

% Cumul ative Non-Attrite

9. Final model
Predictor Category

8. Model fine-tuning

Attrition Predictors

Demographics

Age at the time of hiring

Tier

Tier at the end of the year

Awards

Performance award during the current year

Training

Average hours per training

Multivariate visualization
Variable interpretation
Multicollinearity checks

7. Iterative multivariate
analysis
Iteration 1
Param eter
DF
Estim ate Standard Error Wald Chi-Square Pr > ChiSq
Intercept
1
-2.1 7 85
0.3284
44.01 66 <.0001
A ttrtn_under_mgr_flg
1
0.2265
0.0831
7 .4239
0.0064
srv y _Q7 _y 0_flg
1
0.927 4
0.1 867
24.6622 <.0001
Iternation 2
prtm_flg
1
1 .4231
0.801 5
3.1 528
0.07 58
Param eter
DF
Estim ate Error
Chi-Squ are
Pr > ChiSq
dpndnts_cnt_y 0
1
-0.1 338
0.0482
7 .7 1 29
0.0055
Intercept
1
-2.1 1 02
0.3329
40.1 904 <.0001
age
1
-0.01 99
0.00687
8.3956
0.0038
A ttrtn_under_mgr_flg
1
0.2223
0.0831
7 .1 591
0.007 5
awrd_cat1 _flg_y 0
1
1 .7 1 27
0.1 486
1 32.909 <.0001
srv y _Q7 _y 0_flg
1
0.9601
0.1 888
25.8642 <.0001
bnft_chng_cnt_y 0
1
0.97 84
0.3528
7 .6902
0.0056
prtm_flg
1
1 .4261
0.8067
3.1 251
0.07 7 1
lv l_mod_y 0
1
-0.1 34
0.07 06
3.6033
0.057 7
dpndnts_cnt_y 0
1
-0.1 281
0.0483
7 .0263
0.008
age
1
-0.01 85
0.00695
7 .1 1 1 8
0.007 7
Percent Concordant
7 0 Somers' D
0.41 9
awrd_cat1 _flg_y 0
1
1 .6325
0.1 638
99.2895 <.0001
Percent Discordant
28.1 Gamma
0.428
ltst_tnr
1
-0.00251
0.0021 9
1 .3088
0.2526
Percent Tied
1 .9 Tau-a
0.055
bnft_chng_cnt_y 0
1
0.967 8
0.3527
7 .532
0.0061
Pairs
1 1 57 285 c
0.7 1
lv l_mod_y 0
1
-0.1 324
0.07 05
3.5283
0.0603
Percent Concordant
Percent Discordant
Percent Tied
Pairs

PwC

7 0.4
27 .8
1 .8
1 1 57 285

Somers' D
Gamma
Tau-a
c

6. Variables clustering

X12 X7
X14 X1 X6
X9 X3 X2
X5
X16 X4 X8 X13
X11 X10 X15

Cluster 3

Cluster 1

Set of all
variables
Cluster 2

0.426
0.434
0.055
0.7 1 3

11

HR Analytics| Phase IV Model deployment

Outcomes:

Final summary report of key insights

Heat maps depicting attrition risk

Individual employee level scoring

Final Predictive Model

Current
Employee
Dataset

Where

Three logistic regression models

for the three different segments of
the population

22,474 employees scored*

Employee profiles created using

departure probability deciles

Though model formulation is same

across the three models, coefficients
and predictors are different

*We have scored all active firm employees with

tenure greater than 1 year

PwC

12

Business Cases

Case 2: Trade Area Mapping and

Retail Store Positioning

PwC

February 2015
13

Retail Store Positioning

PwC

Define the trade area

map

Identify the impact of

retail store on digital

Identify the target

customer profile

Identify the recommended locations and the

associated trade area for given customer profiles
Define the future state of retail-network and the
expected market-coverage / revenue impact

Analyze and quantify the relationship between

retail locations and online transactions
Identify potential areas of overlap with the tradearea map

Identify key attributes of the customers that are

part of the target market for respective client
products sold via retail channel

Defining the trade area is dependent on target customer

market
Target Customer footprint around a store Category A[1] Market
25 mile radius:
no incremental
benefit from
trend fit curve

Traditional PC Users
Family Fortunes
Heavy buyers of PC/
Laptop/ Software
Online channel
preferred
Power Couples
More preference
towards PC/Laptop
High preference for
tablets

18 mi

Core market area

Low competition
effect and high
incremental benefit
(540 customers/mi)

Incremental benefit :
100 customers / mi

Demand + Sales Index

Low

Heavy Gamers
Family Sprawl
Large family
size(4+)
High PC preference
as well
School Daze
High propensity for
game systems
Medium Online
preference

High

[1] Category A : High Demand, High Competition 1+ client stores

PwC [2] Refer ence : There is a slight decrease in trade area radius due to competition

client Store (current)

Slide 15

How do we optimize the trade area..

Boundaries of trade area are defined based on the projected revenue / customer growth
per incremental zip code from the stores location

10000
9000

Absolute distance
threshold is around 13
miles after which the
increment is marginal

Sales vs distance

8000

Force fitting gives an

optimum trade area
radius of 25 miles
around the store

# Customers

7000
6000
5000

Gaming
Tablets
Software
PC/Laptops/Tablets

4000

Sales
Poly. (Sales)

3000
2000
1000
0

10

20

30

40

50

Data used : sanitized sales data (mocked up from a different project) by zipcode

PwC

Slide 16

Capturing the impact of competition and xchannels

Comparison analysis of scenarios with and without competition (including alternative

retail channels), to quantify the projected impact of competition on the total revenue
attributable to the store

Competition Analysis
10000

Slight decrease in the trade area

radius due to competition

9000

Without competition

8000

With competition

# Customers

7000
6000

Sales volume decreases

due to competition

5000
4000
3000
2000
1000
0
0

10

20

30
Distance from store

40

50

Sales

60

Data used : sanitized sales data (mocked up from a different project) by zip code

PwC

Slide 17

Quantifying the impact of stores location on

customers propensity to buy online
Online Sales analysis for Category A market
Segments defined by
propensity to use digital
channel but wander stores for
checking out new products are
the ones whose buying
behavior is affected by
presence of a store.
Top Segments
Midlife Highlife
Power Couples
Online Living
Family Fortunes

18 mi

Within the defined trade

area measure the baseline
for online sales and channel
preferences
Incremental benefit :
100 customers / mi
Demand + Sales Index
Low

High

client Store

PwC

Slide 18

Effect of store on online sales

Quantifying the channel preferences and influence of store wandering on transaction
volume and value will help us quantify the impact of retail store on digital-transactions

Sales vs Distance
Online sales+ Retail PoS
10000

Sales volume increased due to

presence of store
Retail PoS

# Customers

8000

6000

Sales
Competition Presence

Poly. (Sales)

4000

Poly. (Competition Presence)

2000

Effectively, marginal increase

in trade area observed due to
online sales

0
0

10

20

30

40

50

60

Distance
Data used : sanitized sales data (mocked up from a different project) by zip code

PwC

Slide 19

Business Cases

Case 3: Exploring consumer demand

and preference for bundled services

PwC

February 2015
20

Market Overview - Dallas

Market Overview : High Speed Internet Demand

Penetration of Client
Suburbs of Dallas show a higher
demand for High speed Internet as
compared to regions around
Downtown

Downtown and northern suburb of

Dallas have higher penetration and
product mix along with having a high
demand

These regions have high demand for

only one product type
Demand

Size : Penetration
Low
Product A*

High
Product B*

Product C*

*Product mix is calculated using Mock Data

PwC **Demand & Penetration is derived from Claritas survey data

November 2014
21

Market Overview - Dallas

Market Overview : Competition Analysis

Majority of the zips show low
competition in the high speed internet
space. However, these regions are
dominated by two of the competitors

These regions have high demand and low

This region has
competition
with
high
a decent
demand
population
and low base
competition
which
suggest
with
Client
decent
should
population
focus onbase
entering these zips

Demand

Size : Population
Low
Low

High
Medium

High

**Demand & Competition is derived from Claritas survey data

PwC

November 2014
22

Customer Profiles Selection - Dallas

Selected Zips based on certain profile attributes

Market Segmentation can be done by choosing from the different scenarios. Similar markets can be focused
with similar strategies. After selecting markets with such filters , strategies should be created on the basis of
customer profile and preferences.

PwC

November 2014
23

Customer Profile Details - Dallas

Customer Profile
The following segment Economizers dominates the Dallas area. This segment consists of the poorest financial
groups. Consists of racially mixed singles and single-parent families, watching wresting and listening to gospel
radio

PwC

24

How to enhance your analytical skills

PwC

February 2015
25

Gain expertise or working knowledge of

theoretical analytics

Coursera: Data Scientist Toolbox

MIT Courseware (courses offered under Sloan School of
Management): http://ocw.mit.edu/courses/sloan-school-ofmanagement/
MA 106: Linear Algebra
IC 102: Probability and Statistics for Engineers, Sheldon Ross
Kaggle: For problems in the domain of business analytics
Visualization super awesome website: d3js.org

PwC

February 2015
26

How to pitch your analytical skills

PwC

February 2015
27

Process flow for a good problem solving

Step 1: Understand the problem (without the bias that you have to
solve it analytically)
Step 2: Put yourself in the client shoes and realize what you would
have wanted as a solution if you would have been the client. Nobody
is asking you to achieve a correlation coefficient of 0.999 but rather a
probably lower value of 0.7 and an implementable solution.
Step 3: Give the first try to solving the problem imagining that you
know nothing about analytics, putting business sense to it
Step 4: Attack! Attack! Attack! Put all your analytical skills,
techniques you have learned into solving the problem. Go in pure
analyst mode

Step 5: Now rephrase the analytical solution found out by you in a

layman interpretable form: graphs, charts, numbers, comparisons
PwC

February 2015
28

Questions...

This publication has been prepared for general guidance on matters of interest only, and does
not constitute professional advice. You should not act upon the information contained in this
publication without obtaining specific professional advice. No representation or warranty
(express or implied) is given as to the accuracy or completeness of the information contained
in this publication, and, to the extent permitted by law, PricewaterhouseCoopers LLP, its
members, employees and agents do not accept or assume any liability, responsibility or duty of
care for any consequences of you or anyone else acting, or refraining to act, in reliance on the
information contained in this publication or for any decision based on it.
2015 PricewaterhouseCoopers LLP. All rights reserved. In this document, PwC refers to
PricewaterhouseCoopers LLP which is a member firm of PricewaterhouseCoopers
International Limited, each member firm of which is a separate legal entity.