Data Driven Decision Making

Data Driven Decision Making.
indd 1 24-02-2014 PM 05:42:29

Unlock
Insights on
any Data
Data Insights to big data, small data, all data.
To learn more, visit microsoft.com/server-cloud
Data Driven Decision Making.indd 2 24-02-2014 PM 05:42:30

Table of Contents
Chairman’s Message 2
Synopsis 3
About Authors 4
Data-Driven Decision Making 5
Introduction 5
Why Analytics 5
Introduction to Business Analytics 7
Descriptive Analytics 8
Predictive Analytics 10
Prescriptive Analytics 12
Predictive and Prescriptive Analytics Tools 13
Big Data Analytics 16
Social Media And Mobile Data 16
Framework for Data-Driven Decision Making 19
Creating A House of Analytics Excellence for Current Practitioners 21
Roadmap and Tips for Analytics Capability Building for New Entrants 22
Challenges in Data-Driven Decision Making and Future 24
Data-Driven Decision Making in Practice: Survey Results 24
Analytics: Expert View and Insights 31
Analytics in Action 33
The Akshaya Patra Foundation 34
Apollo Hospitals 34
Intelliassist 34
Shubham Housing Finance Development Company 35
Conclusions 35
Appendix 1: List of Software Tools in Analytics 36
Acknowledgements 37
References 37

Soumitra Bhattacharya
Chairman’s Message
In the past decade, every business has become dependent on Information Technology to drive
operational excellence, marketing, employee engagement, innovation or e-commerce to name a few.
It is safe to say that information technology has become essential not only to excel but to survive. In
such a world where businesses are global, employees and markets are diverse, legal and compliance
requirements hold significant risks, decision making has become more of a science than art.
CXO’s and heads at various levels in the organisation are realising that taking decisions without data is
a gamble. The report “Data driven decision making” is being released at this time to help organisations
understand relevance of Analytics in their businesses and how they can go about harnessing the
power of Analytics in their respective lines of work. The intention here is to create awareness on the
topic and deliver some insight through research, case-studies and surveys. I congratulate IT Panel for
taking up this initiative and I am certain, it will add value to CII member community at large.
Soumitra Bhattacharya
Chairman, CII Karnataka

Mr. Sudhir Kumar Reddy Mr. Narayan Rajan
Convenor, CII Karnataka IT Panel Co-Convenor, CII Karnataka IT Panel
& CIO, Mindtree Ltd & CEO, i-Vista Digital Solutions Pvt Ltd
Synopsis
Analytics has been around for some time delivering value in terms of reports and dashboards to
managers in enterprises. But, this was restricted to structured data and value derived was mostly in
the area of information synthesis (in other terms slice and dice by various parameters). Enterprises
today are sitting on terabytes and petabytes of information internally due to IT enabled processes.
Add to that, rapid adoption of technologies like e-commerce, social networks and mobility, the amount
of internal/external unstructured data which can be harnessed is enormous. Some companies have
recognised the value of mining such information to predict outcomes and create first mover advantage.
This led to the gold rush for analytics and newer skills/tool sets such as data scientists, big data
began to gain prominence. This report gives an insight into present trends in analytics, perception of
CIO’s and IT heads on the topic and tries to help guide the readers on how one can go about creating
analytics roadmap in respective organisations.

U Dinesh Kumar Sandhya Shenoy Arun Pandit
About Authors
Dr U Dinesh Kumar is a Professor of Quantitative Methods and Information Systems at IIM Bangalore.
U Dinesh Kumar holds a Ph.D. in Mathematics from IIT Bombay and M.Sc. in Applied Sciences (Operations
Research) from P.S.G. College of Technology, Coimbatore. Dr Dinesh Kumar has over 20 years of teaching
and research experience. Prior to joining IIM Bangalore, Dr Dinesh Kumar has worked at several reputed
Institutes across the world including Stevens Institute of Technology, USA; University of Exeter, UK;
University of Toronto, Canada; Federal Institute of Technology, Zurich, Switzerland; Queensland University
of Technology, Australia; Australian National University, Australia and the Indian Institute of Management
Calcutta.
Dr Dinesh Kumar has carried out consulting projects for organizations such as The Boston Consulting
Group (India) Private Limited, Hindustan Aeronautics Limited, Army Management Studies Board, Wipro
Limited, World Health Organization etc. Dr Dinesh Kumar teaches Business Analytics and Intelligence, Six
Sigma, Supply Chain Analytics and Stochastic Models at IIM Bangalore. Dr Dinesh Kumar has published 10
case studies on use of analytics among the Indian organizations at the Harvard Business Publishing. He has
authored more than 70 articles and 2 text books.
Ms. Sandhya Shenoy is a Research Associate (RA) at the Indian Institute of Management – Bangalore. She
comes with an overall Industry work experience of 8.5 yrs. Before joining IIMB in November 2013, she
worked for an American Retailing Company, Target Corporation for 4.5 years as a Senior Business Analyst,
prior to which she worked with Infosys BPO, a subsidiary of Infosys Limited for 4 years. Throughout her
stint, she has worked under various domains – Merchandise Planning, Financial Planning and Analysis,
Media and Entertainment, Financial Services and Mortgage banking. As a part of her Certification degree in
Business Analytics and Intelligence from IIM, Bangalore (2012-2013), she worked on projects pertaining
to Study of Brand Switching using Markov Chains and Study of Impact of Promotions on Sales using
Logistic Regression for an Indian Retail Organization. Her latest project as an RA was building a model for
predicting Probability of default (PD) for a Financial Institution. She aspires to continue working with IIMB
to enhance her knowledge in the field of research and analytics.
Mr. Arun Pandit is a Research Associate at Indian Institute of Management – Bangalore (IIM-B). Before
joining IIM Bangalore, he has worked with Hinduja Global Solutions Limited (HGS) for a decade. In HGS,
he started working as an Associate for International Voice Process and over next few years moved up
and became Manager – Management Information System. He was part of HGS’s India domestic business
team handling client relationship management, contract governance, commercials, operations metrics
and data management. Prior to HGS, Arun has worked for 4 years with YOKOGAWA, a Process Control
Instrumentation company and for 2 years with WIPRO (Fluid Power Division). He holds a Master of
Computer Applications degree from Bharathidasan University, Tiruchirappalli.

Data-Driven Decision Making
In God we trust; all others must bring data.
–W. Edwards Deming
Introduction
The epigraph, which is often attributed to the management guru and statistician, W. Edwards Deming,
encapsulates the need for a data-driven decision-making process at the organisational level. The
increasing complexity of business problems, the availability of several possible alternative solutions,
and the limited time available for decision making demand a highly structured decision-making
process for the effective management of organisations. Several theories on decision making such
as rational economic models (how decisions should be made), behavioural models (how decisions
are made), implicit favourite models (where decisions are made before the process even begins!),
and political models (where decisions are neither rational nor objective) have been reported in the
literature. In this report, our focus will be on rational economic models, that is, on how decisions
should be made. Although decisions are occasionally made using the “highest paid person’s opinion”
(HiPPO) algorithm and flipism (all decisions are made by flipping a coin), there is a significant change
in the form of “data-driven decision making” among several companies. A typical data-driven decision-
making process uses the following steps (Figure 1):
1. Identify the problem or opportunity.

2. Identify sources of data (primary as well secondary data).
3. Process the data for missing and incorrect data. Prepare the data for analytics model building.
4. Build the analytical models.
5. Communicate the data analysis output and decisions effectively.
6. Implement Solution / Decision.
Stage 3
Stage 1 Stage 2
Process the data for
Identify problems Identify Sources of Data
missing/incorrect data
/ Improvement Required for problem
and prepare for analytics
opportunities identified in Stage 1
model building
Stage 4 Stage 5 Stage 6

Build the best Communicate the Implement Solution /
analytical model Analytics Output Decision
Figure 1. Data driven decision making flow diagram
Why Analytics
Whether we accept it or not, humans are inherently not good at decision making. A great example of
this is the famous Monty Hall problem in which the contestant of a game show is shown three doors.
Behind one of the doors is an expensive item (such as a car or gold); there is an inexpensive item
behind the remaining two doors (such as a goat). The contestant is asked to choose one of doors.
Assume that the contestant chooses door 1; the game host would then open one of the remaining
two doors. Assume that the game host opens door 2, which has a goat behind it. Now the contestant
is given a chance to change his initial choice (from door 1 to door 3). The problem is whether or not
the contestant should change his/her initial choice. Note that the contestant is given an option to

switch choices irrespective of the item behind his/her original choice of door. In this problem, the
contestant—the decision maker—has two choices: he/she can either change his/her initial choice
or stick with his/her initial choice. When Marilyn vos Savant, a columnist at the Parade Magazine,
posted that the contestant should change the initial choice (Savant, 1990), 92% of the general public
and 65% of the university graduates (many of them with PhDs) who responded to her column were
against her answer.1 Although Marilyn vos Savant provided a simple decision tree argument to prove
that the probability of winning increases to 2/3 when the contestant changes his/her initial choice,
many scholars did not accept her argument that changing the initial option is the right decision.
The travelling salesman problem (TSP) is another decision-making problem—given a list of cities and
the distances between each pair of cities, what is the shortest possible route a salesperson should
take to visit each city exactly once and return to the original city? The Akshaya Patra Foundation,
which provides mid-day meals to approximately 1.3 million school children across India, faces
this decision-making challenge. Through their Vasanthapura kitchen in Bangalore, approximately
1,20,000 school children from approximately 700 schools in South Bangalore are provided mid-day
meals. Providing high quality food at an affordable price is one of the challenges faced by Akshaya
Patra. The Vasanthapura kitchen uses 35 vehicles to distribute the cooked food.2 To minimise the cost
of distribution, they need to solve a complex vehicle routing problem (VRP). To simplify this problem,
assume that they divide the number of schools equally among the vehicles; each vehicle would then
have to deliver food to 20 schools. For each vehicle, we need to find the best route. This problem can
be formulated as a TSP with a solution space of 20 factorial (20! = 2.4329 x 1018). If a computer can
evaluate one million routes per second, it would take more than 77146 years to evaluate all possible
routes! For Akshaya Patra, every rupee saved would enable them to add more children to their mid-
day meal programme. Given that the human brain lacks the ability to take the right decision in the
Monty Hall problem that has just two alternatives, a problem with 20 factorials is certainly beyond
the human brain’s processing ability. With approximately 270 employees, Akshaya Patra’s kitchen in
Vasanthapura falls under the category of a small and medium size enterprise (SME). Despite being
an SME, it has to handle a complex problem that many believe is relevant only for big organisations!
In today’s world, data-driven decision making through business analytics is not an option, but an
essential capability that every organisation should acquire irrespective of its size. Based on a survey
of 3000 executives, Hopkins et al. (2010) claimed that there is a striking correlation between an
organisation’s analytics sophistication and its competitive performance. Today, several multinational
companies are using Business Analytics as a competitive strategy. The biggest obstacle to adopting
analytics is the lack of knowledge about the tools and techniques that are needed.
The primary objective of this report is to discuss the importance of data-driven decision making
and to provide a roadmap and tips for creating in-house analytics capabilities for small and medium
enterprises (SMEs). Manufacturing holds a big share in SMEs across India. Many of the problems
pertaining to manufacturing such as inventory management, quality improvement, forecasting,
optimisation at various functional areas of manufacturing, supply and distribution management,
and after-sales service would need the intervention of analytics to improve the overall operational
effectiveness of the company. The India Brand Equity Foundation reported that the manufacturing
industry is the backbone of the Indian economy and that it strengthens employment, agriculture,
and the service sectors.3 India ranks second in the world as per the 2010 Global Manufacturing
Competitiveness Index (GMCI) prepared by the U.S. Council on Competitiveness together with
Deloitte.4 We will focus our discussion on the use of analytics within SMEs since for many SMEs,
analytics is still an alien topic.
1 Source: http://en.wikipedia.org/wiki/Monty_Hall_problem
2 Source: information provided by the Akshaya Patra Foundation.
3 Indian Manufacturing Sector Reports- Report Name: “SMEs-Role-in-Indian-Manufacturing Sector” accessed at (www.ibef.com)
4 Source: Indian Manufacturing: Overview and Prospects, available at www.ibef.com

Introduction to
Business Analytics

Introduction to Business Analytics
Go down deep enough into anything and you will find mathematics.
–Dean Schlicter
Business analytics is a set of statistical and mathematical management tools and processes used for
analysing the past data that can drive fact-based decision making in an organisation. According to the
theory of bounded rationality proposed by Herbert Simon (1972), the human mind lacks the ability to
choose the right decisions due to the complexity of the problems that the organisations are facing and
the limited time available for decision making. In the 1980s (and even today in many organisations),
the culture of data collection was poor. Organisations found decision making difficult due to the non-
availability of data that could be made available quickly. The introduction of enterprise resource
planning (ERP) systems in many organisations ensured the availability of data that can be called upon
whenever needed. However, the data sitting in the ERP systems needed to be analysed for arriving
at fact-based decisions; this need is now met using analytics. Business analytics (BA) is a subset of
business intelligence (BI) that helps organisations to manage the data-to-decision cycle. Business
analytics can be grouped into three types: descriptive analytics, predictive analytics, and prescriptive
analytics. In the following sections, we discuss the three types of analytics in detail.
Descriptive Analytics
If the statistics are boring, then you’ve got the wrong numbers.
–Edward R. Tufte
Descriptive analytics is the simplest form of analytics that mainly uses simple descriptive statistics
and data visualisation techniques to communicate the hidden facts and trends in the data. Dr. John
Snow’s cholera spot map demonstrates the power of descriptive analytics. Cholera claimed millions
of lives across the world during the nineteenth century. Medical practitioners did not have any clear
understanding of the causes of the disease. Between 1845 and 1856, over 700 articles were published
in London on the causes of cholera and how the epidemic could be prevented; however, none of them
offered any real solution. The breakthrough in cholera epidemiology was made by Dr. John Snow based
on the data of cholera outbreak in central London in 1854. Between 31 August and 10 September,
1854, over 500 people died of cholera in London. John Snow marked the locations of the homes of
those who had died during the epidemic and prepared a spot map (Figure 2 5). The spot map revealed
that the highest number of deaths occurred in the golden square area (Snow, 1999). The most striking
difference between this area and the other districts of London was the source of water (Brody et al.,
2000); thus, Snow established that water contamination was the main source of cholera.
5 Source: http://en.wikipedia.org/wiki/John_Snow_(physician)

Figure 2: John Snow’s Spot Map of Cholera Outbreak in London, 1854
Edward Tufte (2001), in his book titled The Visual Display of Quantitative Information, demonstrated
how innovative visuals can be used to effectively communicate data. Today, dashboards form the
core of business intelligence and are an important element of data-driven decision making. Indian
start-up companies such as Gramener6 have used innovative data visualisation tools to communicate
hidden facts in the data. For SMEs, descriptive analytics could be the initial stages of creating analytics
capability. Descriptive statistics could help SMEs to uncover inefficiencies and eliminate them.
Simple analysis of data can lead to business practices that result in financial rewards. For instance,
companies such as RadioShack and Best Buy found a high correlation between the success of individual
stores and the number of female employees in the sales team (Underhill, 2009). Underhill (2009)
also reported that the conversion rate (percentage of people who purchased something) in consumer
durable shops was higher among female shoppers than among male shoppers. Sometimes, a simple
query can lead to fraud detection. Recently, China Eastern Airline found that a man had booked a first
class ticket more than 300 times in a year and cancelled it before its expiry for full refund so that he
could eat free food at the airport’s VIP lounge7 (It is surprising that the airline took so long to uncover
this!). In 2007, the Times of India reported fake insurance claims of over INR 1 crore involving a car
that didn’t even exist!8
6 Source: https://gramener.com/
7 Source: http://articles.timesofindia.indiatimes.com/2014-01-30/mad-mad-world/46827501_1_free-meals-single-ticket-issued-ticket
8 Source: http://articles.timesofindia.indiatimes.com/2007-07-24/kolkata/27978551_1_accident-victims-insurance-firm-accident-cases

Predictive Analytics
If you torture the data long enough, it will confess.
–Ronald Coase
Predictive analytics is the most important class of analytics that aims to predict the probability of
occurrence of a future event such as customer churn, loan defaults, and stock market fluctuations.
The ability to predict a future event such as an economic slowdown, a sudden surge or decline in a
commodity’s price, and so on will help organisations to plan their future course of action. Anecdotal
evidence suggests that predictive analytics is the most frequently used type of analytics across several
industries. Irrespective of the type of business, organisations would like to forecast the demand for
their products or services and understand the causes of demand fluctuations. For example, when
Hurricane Charley struck the U.S. in 2004, Linda M. Dillman, Walmart’s Chief Information Officer,
wanted to understand the purchase behaviour of their customers (Hays, 2004). Using data mining
techniques, Walmart found that the demand for strawberry pop-tarts went up over 7 times during
the hurricane compared to their normal sales rate; the pre-hurricane top-selling item was found to
be beer. These insights were used by Wal-Mart when the next hurricane—Hurricane Frances—hit the
U.S. in August–September 2004; most of the items predicted by Wal-Mart sold quickly. Although the
high pre-hurricane demand for beer can be intuitively predicted, the demand for strawberry pop-tarts
was a complete surprise. The use of analytics can reveal relationships that were previously unknown
and are not intuitive. Closer to home, IIM Bangalore (IIMB) was associated with a project with Apollo
Hospitals, Bannerghatta Road, Bangalore to predict the demand for food and beverages with the
objective of minimising the wastage of food. The IIMB team developed several forecasting models
using the auto-regressive integrated moving average (ARIMA) technique for accurately forecasting
the demand for various food items in the hospital’s menu to minimise the wastage.
The most popular example of the application of predictive analytics is Target’s pregnancy prediction
model. In 2002, Target hired statistician Andrew Pole; one of his assignments was to predict whether
a customer is pregnant (Duhigg, 2012a). New parents are the holy grail of marketers since they are
price-insensitive customers who would like to buy the best things for their new born. In 2010, it was
reported that parents spent about USD 6800 on average on a child before his/her first birthday; the
North American new baby market was worth USD 36.3 billion (Duhigg, 2012b). At the outset, the
question posed by the marketing department to Pole may look bizarre, but it made great business
sense. Any marketer would like to identify the price-insensitive customers among the shoppers, and
who can beat soon-to-be parents? A list of interesting applications of predictive analytics is presented
in Table 1.
10

Table 1: List of Predictive Analytics Applications
Organisation Predictive Analytics Model

Polyphonic HMI Predicts whether a song will be a hit using machine learning
algorithms. Their product “Hit Song Science” uses mathematical and
statistical techniques to predict the success of a song on a scale of 1
to 10. 9
Okcupid Predicts which on-line dating message is likely to get a response
from the opposite sex (Siegel, 2013).
Amazon.com Uses predictive analytics to recommend products to their customers.
It is reported that 35% of Amazon’s sales is achieved through their
recommender system (Siegel, 2013).
Hewlett Packard (HP) Developed a flight risk score for its employees to predict who is
likely to leave the company (Siegel, 2013).
University of Maryland Claimed that dreams can predict whether one’s spouse will cheat. 10
Flight Caster Predicts flight delays 6 hours before the airline’s alerts.
Netflix Predicts which movie their customer is likely to watch next.
Capital One Bank Predicts the most profitable customer.
Google Predicted the spread of H1N1 flu using the query terms.
Farecast Developed a model to predict airfare, whether it is likely to increase
or decrease, and the amount of increase/decrease. 11
The examples shown in Table 1 represent a tiny fraction of the predictive analytics applications used
in the industry. Companies such as Proctor & Gamble use analytics as a competitive strategy—every
critical management decision is made using analytics (Devonport, 2013). If one were to search for
the reasons behind highly successful companies, one would usually find analytics being deployed
as the competitive strategy. Google—without which many people think the world would end—used
Markov chains for page ranking (Hayes, 2013). Google also developed accurate prediction models
that could predict events such as the outcome of political elections, the launch date of a product, or
action(s) taken by competitors (Coles et al., 2007). Devonport (2006) reported how companies such
as Amazon, Capital One, Harrah’s, and the Boston Red Sox have dominated their business by using
analytics. The application of analytics is not restricted to corporates alone; many sports clubs have
successfully used analytics to manage their clubs. The most famous application of analytics in sports is
by Oakland Athletics, which used analytics to put together a team with the limited resources available
for purchasing players (Lewis, 2003). Oakland Athletics had the third lowest payroll among the
major league baseball teams in 2002. The manager of Oakland Athletics, Billy Beane, used statistical
techniques to identify player qualities that made an impact on the match outcome and to also identify
relatively cheaper skill. Oakland Athletics revised their team management strategy and with a payroll
of USD 41 million, they were able to compete successfully in the league. In 2002, they won 20 games
in a row.
9 Source: http://polyphonichmi.blogspot.in/p/about-company.html
10 Source: http://elitedaily.com/news/world/dreams-can-predict-the-future-of-relationships/
11 Source: http://www.crunchbase.com/company/farecast
11

Prescriptive Analytics
Every Decision has a consequence.
–Damon Darrel
Prescriptive analytics assists users in finding the optimal solution to a problem or in making the right
choice/decision from among several alternatives. Unlike predictive analytics (which, in many cases,
provides the probability of a future event), prescriptive analytics in most cases provides an optimal
solution/decision to the problem. Operations research (OR) models form the core of prescriptive
analytics. Ever since their introduction during World War II, OR models have been used in every sector
of every industry. In this section, we describe some of the prescriptive analytics tools used by different
industries. The list of prescriptive analytics applications is long and several companies across the
world have benefitted from the use of prescriptive analytics tools.
Coca-Cola Enterprises (CCE) is the largest distributor of Coca-Cola products. In 2005, CCE distributed
2 billion physical cases containing 42 billion bottles and cans of Coca-Cola (Kant et al., 2008). CCE
developed an OR model that would meet several objectives such as improved customer satisfaction and
optimal asset utilisation for their distribution network of Coca-Cola products from 430 distribution
centres to 2.4 million retail outlets. The optimisation model resulted in cost savings of USD 54 million
and improved customer satisfaction. A similar vehicle routing problem was solved by the IIM Bangalore
team for Akshaya Patra. The Akshaya Patra Midday Meal Routing and Transportation Algorithm
(AMRUTA) was developed to solve the vehicle routing problem (discussed in the Introduction); this
was implemented at Akshaya Patra’s Vasanthapura campus, resulting in savings of USD 75000 per
annum (Mahadevan et al., 2013). A major challenge for any e-commerce company is to improve the
conversion of visits to transactions and order sizes. Hewlett Packard (HP) established HPDirect.com
in 2005 to build online sales. HP Global Analytics developed predictive and prescriptive analytics
techniques to improve sales. The analytical solutions helped HP to increase conversion rates and
order sizes (Rohit et al., 2013).
Inventory management is one of the problems that are most frequently addressed using prescriptive
analytics. Samsung implemented a set of methodologies under the title “Short Life and low Inventory
in Manufacturing” (SLIM) to manage all the manufacturing and supply chain problems. Between 1996
and 1999, Samsung implemented SLIM in all its manufacturing facilities, resulting in a reduction in
the manufacturing cycle time of random access memory devices from more than 80 days to less than
30 days. SLIM enabled Samsung to capture additional markets worth USD 1 billion (Leachman et al.,
2002). In the next section, we discuss the more frequently used predictive and prescriptive analytics
tools.
12

Predictive and Prescriptive
Analytics Tools
13

Predictive and Prescriptive Analytics Tools
The most frequently used predictive analytics tools are regression, logistic regression, classification
trees, Markov chains, and neural networks. The frequently used tools in prescriptive analytics are
linear programming, integer programming, multi-criteria decision-making models, combinatorial
optimisation, and meta-heuristics. In Table 2, we provide a brief description of some of these tools
and the problems that are solved using these tools. There are a large number of tools used in analytics
and it is impossible to list all of them in this report. We have highlighted a few tools that are frequently
used by analytics companies. The list of software tools that are used to solve the problems discussed
in Table 2 are provided in Appendix 1.
Table 2: Predictive and Prescriptive Analytics Tools
Analytics Tools Applications
Regression Regression is the most frequently used analytics tool. In management and
social sciences, almost all hypotheses are validated using regression. In
business, irrespective of the sector, the management would like to know how
the key performance indicators (KPIs) of the business are related to macro-
economic parameters and other internal process parameters. Regression
is an excellent tool for establishing the existence of a relationship between
a response variable (KPI) and other explanatory variables. For example,
if one would like to check whether drinking Complan (a powdered milk
energy drink) would increase the height of children or whether using Fair &
Lovely (a face cream that is marketed as having skin-lightening properties)
really makes people fair and/or lovely, one would have to use regression
techniques. Unfortunately, regression is one of the most highly misused
techniques in analytics.
Logistic and Logistic and multinomial regression techniques are used to find the
Multinomial probability of an event. Both these techniques are used for solving
Regression classification and discrete choice models. Classification problems are
common in many businesses. For example, banks and financial institutions
would like to classify their customers into several risk categories. Companies
would like to predict which customer is highly likely to churn in the next
quarter. Marketers would like to know which brand a customer is likely to
buy and whether promotions can make a customer change his/her brand
loyalty. Credit scoring and fraud detection are other popular applications of
logistic regression.
Classification Trees Chi-Squared Automatic Interaction Detection (CHAID) and Classification

Trees (CART) are frequently used for solving classification problems.
14

Markov Chains Olle Haggstrom wrote an article stating that problem solving is often a matter
of cooking up an appropriate Markov chain. 12 One of the initial applications
of Markov chains was implemented by the American retail giant Sears. They
used a Markov Decision Process to decide the optimal mailing policy for their
catalogues (Howard, 2002). Today, Markov chains are one of the key analytics
tools in marketing, finance, operations, and supply chain management.
Neural Networks A neural network is a machine learning algorithm inspired by the functioning
of the human brain and is one of the very popular tools in predictive
analytics, especially in forecasting applications.
Linear Since its origins during World War II, linear programming is one of the most
Programming frequently used tools in prescriptive analytics. Problems such as resource
allocation, product mix, cutting-stock problem, revenue management, and
portfolio optimisation are some of the problems that are frequently solved
using linear programming.
Integer Many optimisation problems in real life may have variables that can take only
Programming integer values. When one or more variables in the problem can take only an
integer solution, the model is called an integer programming model. Capital
budgeting, scheduling, and set covering are a few problems that are solved
using integer programming.
Multi-Criteria In many cases, the decision makers may have more than one objective (or
Decision-Making KPIs). For example, a company may like to increase the profit as well as
Model the market share. It is possible that the various objectives identified by the
organisation may conflict with one another. In such cases, techniques such as
Analytic Hierarchy Process and Goal Programming are used to arrive at the
optimal decisions.
Combinatorial Combinatorial optimisation involves choosing the optimal solution from a

Optimisation large number of finite solutions. The travelling salesman problem (TSP), the
vehicle routing problem (VRP), and the minimum spanning tree problem
(MST) fall under this category. Many industry problems are analogical to TSP,
VRP, and MST.
Non-Linear Large classes of problems faced by the industry have non-linear objective
Programming (NLP) functions and/or non-linear constraints. Many engineering design
optimisation problems fall under this category. NLP are also difficult set of
problems to solve due to limitations of the existing algorithms.
Social Media Social media analytics is a collection of tools used for analysing unstructured
Analytics Tools data such as texts, videos, photos, and so on. With the exponential growth
in the use of social media by the general public, tools designed for analysing
unstructured data will be frequently used by organisations.
12 Source: http://math.uchicago.edu ~shmuel/Network-course readings/Markov_chain_tricks.pdf
15

Big Data Analytics
The world is one big data problem.
- Andrew McAfee
Big data is defined as a big volume of data (in excess of 1 terabyte) generated at high velocity and high
variety. That is, big data is identified using 3 Vs, namely, volume, velocity, and variety. For example,
10 billion photos are uploaded on Facebook every hour. Google processes 24 petabytes (1 peta byte
= 1015) every day (Mayer-Schonberger and Cukier, 2013). Google was the first to exploit big data for
targeted advertising using clickstream data. Google also predicted the spread of H1N1 flu based on
the search terms entered by Google users (Ginsberg et al., 2009). According to Richard Kellet, Director
of Marketing at SAS, we (human) created 500 times more data in the last 10 years than what we had
done prior to that, since the beginning of humanity (Scott, 2013). Every B787 flight creates half a
terabyte of machine-generated data. For large banks, the automatic teller machine (ATM) transactions
themselves will run into several billion transactions per month. Telecom call data, social media data,
banking and financial transactions, machine-generated data, and healthcare data are a few examples
of the sources of big data. Alternatively, any problem that can challenge the existing computing power
and IT systems constitutes a big data problem. Big data problems need innovative ideas in order to
handle the 3 Vs for effective decision making. The volume of the data generated is increasing every
day, and most of this data is user generated, mainly from social media platforms. With the increase in
Internet penetration and autonomous data capturing, the velocity of data is also increasing at a faster
rate. It is estimated that 2.5 exabyte of data is created every day; this figure is likely to double in a
year (Mayer-Schonberger and Cukier, 2013). As the velocity of the data increases, traditional models
such as regression and classification techniques may become unstable and invalid for analysis. The
variety of data is another challenge within big data. Even in the case of text mining, users may use
different languages within the same sentence, especially in a country such as India, which is home to
hundreds of languages. Natural language processing (NLP), which is an essential component of big
data, is challenging when multiple languages are used in the text data.
Innovative parallel processing capabilities such as Apache Hadoop (that comes with the ability to
process large-scale datasets in multiple clusters using the Hadoop Distributed File System) and
MapReduce (which enables parallel processing of large datasets) are used by organisations to
handle big data. However, big data technologies are still in a nascent stage and are far from maturity.
Many operations-related analytics problems such as the travelling salesman problem and dynamic
scheduling cannot be completely solved using these technologies. However, these big data technologies
do provide better computing power compared to existing technologies.

Social Media And Mobile Data
Social media and mobile devices such as smart phones are becoming an important source of data for
all organizations, small and big. Social media is also an important marketing channel for marketers
since it helps to create buzz or electronic word of mouth (WoM) effectively. Stelzner (2013) claimed
that 86% of the marketers indicated that social media is an important channel for their business in
2013. Stelzner (2013) identified the following questions as the most relevant for any marketers when
dealing with social media engagement (also valid for mobile devices):
1. What is the most effective social media tactics?

2. What are the best ways to engage the customers with social media?
16

3. How to calculate the return on investment on social media engagement?
4. What are the best social media management tools?
5. How to create a social media strategy for the organization?
From the literature on social media and our interviews with the industry experts, it is evident that
social media is important for both sources of data as well as marketing the products and services.
However, the effectiveness of the social media marketing is still an understudied subject. A study by
IIMB team (Suhruta et al 2013) claimed that there is a relationship between social media engagement
and the box office collection of movies based on the data obtained from the Bollywood movie, “1920
evil returns”. Social media has several advantages over conventional media as given below.
• Social media is measurable in terms of impressions, visits, views, clicks, comments, shares,
likes, followers, fans, subscribers, etc. Impact of conventional media cannot be measured, for
example, views of a hoarding or newspaper ad cannot be measured.
• Social media is less expensive than conventional media and has the potential to reach to a wider
audience. Social media can create viral impact in a short duration and can reach larger number
of people.
A key challenge in social media strategy will be assessing the return on investment. Return on
Investment (ROI) should be calculated by the formula ROI = (Gain from Social Media Marketing – Cost
of Social Media Marketing)/Cost of Social Media Marketing. However, it is difficult to quantify the
actual gain from social media marketing. Hence, several variations are used to calculate ROI as given
below:
• Return on Engagement (ROE) 13 – This measures the impact of social media marketing on users’
engagement on the premise that higher engagement leads to higher awareness and thus greater
likeliness to make a purchase decision. ROE calculation for some of the social media platforms
are given below:
o Facebook – (Number of likes, comments, and shares on a post)/(Total number of Facebook
page likes)
o Twitter – (Number of replies, re-tweets)/(Number of followers)
o YouTube – (Number of comments, ratings, and likes)/(Number of video views) OR (Number
ofcomments, ratings, and likes)/(Number of subscribers)
• Return on Influence – This tries to measure how social media activity changes the behavior of
users.
• Anecdotes – This measures verbal sharing of sales activity or intent of purchase on the social
media platforms.
• Correlation – This measures the relationship between any social media engagement activity,
and actual sales
• Multivariate Testing – This measures the relationship between multiple social media engagement
activities and actual sales and enable providing the right kind of offers and promotions to
different users
• Linking and Tagging – This approach provides links on the social media to the buyers to make
his/her purchase and thus it is possible to relate sales and social media engagement. Another
way is to embed ‘‘Cookies’’ (a piece of software), which track consumers’ online activity, thus
providing the connect between social media engagement and actual sales. However, this
approach is more effective when the sales are conducted online.
13 Hemann, C and Burbary, K., Digital Marketing Analytics, Que Publishing, USA, 2013.
17

• Social commerce approach – In this method, sales are directly conducted through social media
– for example, a store front is set up on Facebook page.
• Share of conversation – (Volume of conversation for a particular brand)/(Volume of conversation
for entire industry)
• Sentiment Analysis – Tracks overall brand perception by crawling through all the data available
on the net
Kumar and Mirchandani (2012)14 have proposed new measures such as customer influence effect
(CIE), stickiness index and Customer Influence Value (CIV).
14 Kumar V and Mirchandani R, Increasing the ROI of Social Media Marketing, MIT Sloan Management Review, 2012, 54, 55-61.
18

Framework for Data-Driven
Decision Making
19

Framework for Data-Driven Decision Making
The framework for data-driven decision making and problem solving can be divided into five
integrated stages: problem and opportunity identification; collection of relevant data; data pre-
processing; analytics model building; and communication of the data analysis for decision making.
The various activities carried out during these different stages are described in Figure 3. The success
of analytics projects will depend on how innovatively the data is used by the organisation as compared
Problem or Opportunity Identification

• Domain knowledge is very important at this stage of the analytics project. If one looks at
Target’s pregrancy prediction model, this problem statement used a lot of insights about
the retail business such as price-insensitive customers, understanding customer needs,
possibility of long-term relationships, and the market size. This will be a major challenge for
many companies who do not know the capabilities of analytics.
Collection of relevant data
• Once the problem is defined clearly, the project team should identify and collect the relevant
data. This may be an interative process since “relevant data” may not be known in advance in
many analytics projects. The existence of ERP systems will be very useful at this stage.
Data Pre-processing
• Anecdotal evidence suggests that data preparation and data processing forms a significant
proportion of any analytics project. This would include data imputation and the creation
of additional variables such as interaction variables and dummy variables in the case of
predictive analytics projects.
Model Building
• Analytics model building is an iterative process that aims to find the best model. Several
analytical tools and solution procedures will be used to find the best analytical model in this
stage.
Communication of data analysis
• The communication of the analytics output to the top managmenet and clients plays a crucial
role. Innovative data visualisation techniques may be used in this stage.
Figure 3: Framework for Data-Driven Decision Making
20

to the mechanical use of analytical tools. Although there are several routine analytics projects such
as customer segmentation, clustering, forecasting, and so on, highly successful companies blend
innovation with analytics.
Creating A House of Analytics Excellence for Current Practitioners

Although many companies have successful analytics verticals within their organisation, many are still
in the process of creating one. In this section, we discuss the pillars of building a centre of analytics
excellence, especially within a small and medium enterprise (SME). Like many other initiatives,
creating a data-driven decision-making process requires a change in organisational culture, and
without the support of the top-level management, it may be difficult to create a strong data-driven
decision-making culture within an organisation. Data-driven decision making may not result in
immediate benefits, especially when the process is new to an organisation. The second important
factor in creating a house of analytics success is the talent. It is important that organisations identify
the right talent and nurture them within the organisation to avoid attrition. The organisation should
have the ability to differentiate the true analytics talent from the mediocre analytics professionals.
Devonport and Patil (2012) listed 10 ideas for finding the right data scientists such as recruiting from
top universities, using social media such as LinkedIn, looking for evidence, and so on.
Information technology (IT) plays a crucial role in implementing analytics. Data capturing, data storage,
data transfer, data analysis through analytical models, and finally, communication of the model output
cannot be achieved without proper data architecture supported by other IT infrastructure. In addition
to data handling capabilities, software tools such as SAS, SPSS, STATA, and so on are an essential part of
IT support. A large number of organisations, including multinational companies, prefer open-source
software such as R instead of proprietary software.15 Organisations expect real-time decisions; thus,
in-memory computing is becoming popular among analytics companies. The fourth pillar of analytics
excellence is innovation. Without innovation, the analytics function may not achieve its full potential.
All these pillars need to be integrated with the domain knowledge of the business; otherwise, the
analytics may end up solving non-value-adding problems. The house of analytics is shown in Figure 4.
House of Analytics
Excellence
Integration
Top
Manpower Information
Management Innovation
(Talent) Technology
Support
House of Analytics
Figure 4: House of Analytics Excellence
15 Many universities across the world teach analytics using R due to its capability, and not simply because it is an open-source software.
21

Roadmap and Tips for Analytics Capability Building for New Entrants
Many organizations whether small or big have a large number of low hanging fruits that can be
targeted with simple analytics tools such as descriptive analytics statistics, data visualization, pivot
tables, correlation analysis, basic quality tools, lean and six sigma. Tools such as pivot tables that can
be obtained using Microsoft Excel can be used for targeting several small improvement opportunities.
Lean and Six Sigma concepts are usually a good way to start analytics practices in an organization if
Define Analytics Strategy
• Develop long term plan for the role of analytics within the organization.
• Identify key functional areas within the organization where the organization would like to
kick start the analytics process.
• Communicate analytics strategy across the organization.
Build Talent
• Success of analytics projects heavily depends on the human resource. Get the recruitment
strategy right - Manpower should include skills on technology as well analytics model building.
• Outsource the tasks if necessary. In house team is still needed to validate the models
generated by external consultants.
Build Infrastructure
• Hardware and software can be built incrementally. There are many open source software
tools available. SMEs may exploit open source software such as R.
• Explore cloud option for hardware.
Identify Sources of Data and Develop Data Collection Plan
• Access to relevant date is important for the success of analytics project. Identify relevant
data and sources of data.
• Design the data collection, storage and retrieval plan.
Analytics Implementation
• Start with simple applications targeting small improvements. Use lean techniques such as
GEMBA to identify opportunities for analytics projects.
• Innovation plays a major role in success of analytics implementation.
• Build effective communication strategy for analytics output.
• Calculate return on investment (ROI) on analytics projects.
Figure 5. Roadmap for Analytics Capability Building for SMEs
22

they do not have strong analytics expertise and would like to initiate analytics practices within their
organization. Companies who are planning to start analytics divisions to support decision making
may use the framework shown in Figure 5.
Sample industry wide applications of analytics are captured in the following Table 3.
Table 3. Industry wise analytical problems and data resources
Industry Sector Sample Analytical Problems Data Sources
Manufacturing Supply chain analytics: Inventory Procurement, sales and

management, procurement, vendor selection, production data.
distribution management Warranty and after sales
Quality and Process improvement: Product service data.
Quality, Manufacturing quality, process Commodity price data
improvement Manufacturing data.
Revenue and Cost Management: Revenue Macroeconomic parameter
maximization and cost minimization. data.
Warranty Analytics: Manage end customer
warranty and after sales support data.
Retail Assortment Planning: Category and SKU Price data.

(stock keeping unit) management that will Demand data at SKU and at
maximize the revenue and improve loyalty. category level.
Promotion Planning: Decide promotion SKU level sales data with and
strategy such as temporary price cuts, without promotions.
markdowns, bundling etc Planogram
Demand Forecasting: Forecast demand at SKU Customer demographics
level for managing supply chain. data.
Market Basket Analysis: Association among Point of Sales (PoS) data.
SKUs in customer purchase. Loyalty program data.
Customer Segmentation: Identify the
customer segmentation for target marketing.
Healthcare Clinical Care: Data related to clinical care and All patient care related data
treatment required for improving quality of care. Hospitality related data
Hospitality related data: Data related Patient feedback data
to issues such as registration process,
housekeeping, nursing, utility, diagnostic data
etc.
Service Demand Forecasting: Forecast demand for the Transactional and feedback
service data
Service Quality Analysis: Analyse quality for
benchmarking and improvement Pricing and demand data
Customer Segmentation: Used for Target
Marketing Promotional data
Promotion: Data related to promotion and its
impact
23

Banking and Service Demand Analysis: Demand for Customer transactional data
Finance different services
Customer Transaction Analysis: Used for Loan originating data
many different Analytics and decision making
insights. Credit scoring data
Credit Scoring: Important for managing
different portfolios.
IT and ITES Demand for Analytics Services: Identify Customer interaction and
(IT enables demand for analytics products and services market research data
services) Software Development Cycle Time: Cost and Internal product
Time reduction development data
In addition to the primary data, the companies need to use data from secondary sources such as social
media and sources such as centre for monitoring Indian economy (CMIE), Capitaline, Bloomberg,
Thomson Reuters, A C Nielson, Indiastat.com etc.
Challenges in Data-Driven Decision Making and Future

Analytics requires a cultural change in an organisation and managing this change will be the most
significant challenge for many companies. This is true for any new initiative and is not specific to
analytics. However, unlike many other initiatives, developing analytics skills—if these skills are not
already present in the organisation—can be a major hurdle. Employees who are not skilled in analytical
tools would need to be trained. Unlike other training programmes, analytics training can be long and
expensive. Sirkin et al. (2005) identified duration, performance integrity, commitment, and effort as
important factors that would determine the outcome of the change initiative. A short duration to bring
about the change would be better. Performance integrity refers to the ability to complete the change
initiative on time. Commitment from the top management as well as the employees affected by the
change and the additional effort required by the employees to bring about the change are the other
two factors.
If analytics talent is not available internally, the company should establish a system to identify the right
talent. Building the right talent pool and retaining the talent would be a key challenge. The third big
challenge would be the investment—the IT infrastructure required for advanced analytical techniques
can be expensive. However, small and medium enterprises can achieve significant improvements by
using simple tools such as MS Excel.
Analytics will become an integral part of organization and majority of the decisions will be made using
data. Innovation will be the key success factor for analytics and technologies will provide answer to
questions such as what should be the best strategy for big data and social media? How to calculate the
return of investment?
Data-Driven Decision Making in Practice: Survey Results

A survey questionnaire was designed to capture the extent of the use of “data-driven decision-making”
processes among Indian firms. The survey respondents included senior management leaders, middle
24

level managers, consultants, and analysts. We received 63 responses for the survey from various
functions and business units such as finance, academic institutions, IT advisory function, technical
teams, sales & marketing, retail analytics supply chain, QM & IS, analytics, sales, license compliance,
corporate, technology group, management, marketing, sales and accounts, global business services,
consulting function, financial services, engineering services, PMO & BD, software implementation –
project management, corporate quality, advisory function, products, business and product development,
outsourcing, HR, operations management, display, documentation, digital analytics, operations, MIS,
web analytics, credit risk management, research & analytics, quality engineering, delivery, staffing,
retail and logistics, merchandise planning, general management, engineering industrial services,
insurance, and healthcare. The analyses of the responses to the survey questionnaire are provided
below.
Which sector does your company/organisation belong to?
Educational Banking
Institution 3%
5%
Retail
8%
Other
Healthcare
30%
8%
Products
9%
Manufacturing Consultancy
12% 25%
How many employees work in your company/organisation?
100%
80%
60% 49%
40% 32%
16%
20%
3% 0%
0%
>5000 <500 500-1000 2000-5000 1000-2000
25

Which category does your organisation fall into?
100%
80%
63%
60%
40% 29%
20% 6%
2%
0%
MNC Domestic/ Startups Government
Indian Company (Owned/Controlled)
What are the possible challenges your organisation may face when
adopting analytics capabilities?
Budget
Constraints
14%
Right Skill Set
/Manpower
32%
Management 14%
14%
Technology
26%
Awareness of
Analytical Capability
What tool(s) is predominantly used for decision making in your organisation?
SPSS
R 8%
9%
MINI TAB 9%
45% MS Excel
13%
Other
16%
SAS
26

How would you rate the analytics capability in your company/organisation?
Poor 8%
Average 17%
Good 34%
Very Good 29%
Excellent 12%
0% 10% 20% 30% 40% 50%
How would you rate your own understanding of analytics?
Poor 2%
Average 25%
Good 43%
Very Good 26%
Excellent 4%
0% 10% 20% 30% 40% 50%
How would you rate your satisfaction level with regard to your usage
of data for your team?
Poor 7%
Average 23%
Good 42%
Very Good 23%
Excellent 5%
0% 10% 20% 30% 40% 50%
27

Do operational decision makers in your Are these business rules derived on the basis
company/organisation have pre-de�ined of data analysis/business analytics?
business rules for some processes?
100% 91% 100%
80% 80%
66%
60% 60%
40% 40% 34%
20% 9% 20%
0% 0%
Yes No Yes No
How frequently are the business rules revised in your organisation?
Once in 5 years
Once in 2 years 2%
11%
Quarterly
37%
Annually
29%
Semi-annually
21%
Is there a digitised platform that supports Are you aware of the analytics functions
the enterprise’s key business processes? that support your businesses?
100% 100%
85%
62%
50% 50%
38%
15%
0% 0%
Yes No Yes No
28

Have you ever used analytics or data Do you have unstructured data
analysis in your projects? in your business?
100% 100%
82% 80%
50% 50%
18% 20%
0% 0%
Yes No Yes No
Does your organisation have an enterprise Does your organisation intend to expand its
resource planning (ERP) system in place? focus on a data-driven approach to make
decisions in the next two years?
100% 100%
85%
62%
50% 50%
38%
15%
0% 0%
Yes No Yes No
Your company/organisation follows evidence-based decision making.
Strongly Disagree 2%
Disagree 2%
Neutal 15%
Agree 50%
Strongly Agree 31%
0% 10% 20% 30% 40% 50% 60%
29

Your company/organisation follows analytics practices.
Disagree 5%
Neutal 23%
Agree 51%
Strongly Agree 19%
0% 10% 20% 30% 40% 50% 60%
All decisions taken in your organisation/business unit should be supported by appropriate data.
Disagree
Neutal 7%
Agree 58%
Strongly Agree 33%
0% 10% 20% 30% 40% 50% 60%
Analytics as a functional unit is important and has tremendous in�luence on the pro�itability
and success of your organisation.
Disagree
Neutal 6%
Agree 42%
Strongly Agree 51%
0% 10% 20% 30% 40% 50% 60%
30

Your organisation encourages analytics practices and has provisioned the appropriate budget
to enhance the analytics capabilities required to meet the business needs.
Disagree 6%
Neutral 20%
Agree 48%
Strongly Agree 23%
0% 10% 20% 30% 40% 50% 60%
The survey revealed the following important aspects on “data-driven decision making”:
1. 82% of the respondents mentioned the use of analytics in their current project(s).
2. 85% of the respondents mentioned that their organisation was planning to expand the analytics
function.
3. Only 62% reported that their organisation has an ERP system.
4. The majority of the respondents felt that analytics is important for decision making.
Analytics: Expert View and Insights

We interviewed several experts while preparing this report. In this section, we summarize the findings
from our interviews with different experts.
1. The majority of the practitioners believed that there is a significant shift in the decision-making
process towards data-driven decision making compared to the early 2000s. The shift towards
analytics was aimed at understanding consumer behaviour, leading to the development of
new tools and techniques. However, there were also concerns that organisations can get lost
in huge volume of data.
2. Supply chain management issues such as inventory management, vendor selection, and so on
dominate the majority of the manufacturing firms’ analytics solutions.
3. Many practitioners agreed on a significant shift in the decision-making process with greater
emphasis on using data to understand customer behaviour.
4. Analytics for analytics sake can be expensive and ineffective; a business should look at the
outcome before embarking on the analytics journey.
5. Data accuracy is a major concern for many analytics teams.
6. For manufacturing companies, data capturing plays an important role since there could be a
large number of data sources and real-time information is extremely important. For facilities
such as oil and natural gas, the number of parameters can run into several thousands and are
captured every second; an appropriate IT system is necessary to make meaningful decisions.
7. Business intelligence is a frequently used tool by many organisations to continuously monitor
the important key performance indicators (KPIs). Data visualisation is becoming a popular
tool among many enterprises.
8. Before starting an analytics project, it is important to know what improvement the organisation
is looking for.
31

9. It is advisable for SMEs to outsource their analytics requirements since in many cases,
significant investments in IT systems is required for in-house capability.
10. Talent plays a major role in the success of analytics initiatives.
11. Big companies are investing heavily in new technologies to meet the analytics needs across
the different functional areas of management such as supply chain management, sales and
marketing, finance, and social media functions.
12. Today, many companies have the ability to capture data, which was missing 10 years back;
now, it is also economically feasible to analyse the data for driving decisions, which was not
the case until recently.
13. Strong technical background to fetch the appropriate data, analytics background, and business
knowledge are the desirable qualities of data scientists.
14. It is important to have a good reporting system about the data with good insights when it
comes to advanced analytics.
15. There is a lot of user-generated data, which is unstructured. This form of analytics is still at a
nascent stage and is growing at a fast pace, with many companies getting engaged. With the
large volume of user-generated data, there is a need to think whether it is relevant to store the
data (tweets, for instance) from the previous year(s). And would it make any sense now (in
the current context)? More developments will be seen in the years to come.
16. SMEs are becoming a big focus area for large MNCs in the analytics domain, with many
analytics solutions being developed for the SME sector.
32

Analytics in Action
33

Analytics in Action
In this section, we discuss four small and medium enterprises (SMEs) and how they are using analytics
in managing their organisations.
The Akshaya Patra Foundation

The Akshaya Patra Foundation (TAPF) provides free mid-day meal (lunch) to school children across
India. TAPF provides freshly prepared nutritious food to over 1.3 million students across 9 states of
India through their 20 kitchens. TAPF has used several Continual Improvement methodologies and
analytical tools such as Six Sigma, Lean to improve their Food Safety, Quality and Service delivery.
TAPF identified several critical to quality (CTQs) parameters such as: 1. Cooking to consumption time,
2. Temperature of the food at the time of delivery at the school and 3. Cold Storage Temperature.
Cooking to consumption time is an important quality parameter for TAPF since food may become stale
beyond six hours. Six Sigma methodology was used by TAPF to reduce the cooking time from 6 hours
to 4 hours, resulting in reduction of cooking to consumption time. Vehicle routing problem was solved
using clustering and mathematical programming to minimize the distribution cost (Mahadevan et al
2013).
Apollo Hospitals
Apollo Hospitals was established by Dr. Prathap C. Reddy in 1972 to contribute to the healthcare
system in India. Apollo Hospitals had been using a clinical score card called ACE@25 (Apollo Clinical
Excellence), which measured and monitored clinical excellence among the group’s hospitals. ACE@25
was launched on September 18, 2008 and was used across 32 hospitals of the group. ACE@25 measured
25 clinical parameters every month, which were benchmarked against global standards. Apollo has
been using several analytical tools to improve various hospitality-related processes, resulting in higher
patient satisfaction. Lean and Six Sigma tools were used to minimise the number of patient complaints
related to housekeeping, food and beverages, facilities, and so on. A linear programming model was
used to manage the linens and to minimise the total cost of linen management. Forecasting methods
such as auto-regressive integrated moving average (ARIMA) was used to minimise food wastage.
Owing to better processes, the cost of quality (in terms of re-work and wastage of consumables) was
reduced, which helped in improving the bottom line. Further, owing to better service and higher levels
of satisfaction, the patients acted as brand ambassadors for Apollo and provided word-of-mouth
publicity, which improved the top line.
Intelliassist
IntelliAssist is focused on providing a comprehensive range of marketing communication services
to business and consumer brands. The company is a full-service digital and social media marketing
agency; they provide print media and direct marketing support also. They service clients across the
entertainment, financial services, information technology, non-profit, pharmaceutical, and real estate
industry sectors. They design solutions to build and manage the online reputation of a brand, create
social engagements to build communities, and encourage positive word of mouth to fuel demand.
34

Since their client portfolio is diverse and every client is in a different stage in terms of their social
media presence and usage, it is imperative to outline the marketing objectives and the success
measurement criteria before launching any campaign. To enhance efficiency, the company uses tools
like TweetDeck and HootSuite. When it comes to ROI measurement, the company relies on Facebook
Insights and Google Analytics. To monitor online conversations, the company uses Google Alerts and
Social Mention. To measure performance on Twitter, the company uses FollowerWonk. To understand
the performance of a video on YouTube, the company uses VidIQ.
Shubham Housing Finance Development Company

Shubham Housing Development Finance Company commenced its lending operations in May 2011
through a single branch in New Delhi, with a vision to provide mortgage products and housing
improvement loans to families that were excluded by traditional housing finance institutions. Led by
Sanjay Chaturvedi as the CEO and Ajay Oak as the COO, Shubham became a pioneer in offering formal
housing credit to low-income families from the informal sector. Within two years, Shubham had over
40 branches across several cities in India. In November 2012, Gurgaon-based Shubham raised an
additional USD 7.8 million (approximately INR 50 crore) from venture capitalists (Bruhadeeswaran,
2012), two years after the housing finance firm had first raised around USD 2 million (approximately
INR 12.5 crore). By September 2013, Shubham had disbursed loans amounting to over INR 125 crore to
around 2,300 applicants. An interview-based approach was used by Shubham to assess the applicants
based on their daily or monthly cash earnings as observed by Shubham’s staff at the applicant’s
workplace, instead of relying on formal documentation for proof of their income. In addition, logistic
regression and CHAID trees were used to evaluate the credit risk associated with a customer.
Conclusions
Data-driven decision making is not a fad; what had been believed by many organizations for several
decades is now getting strengthened and formalised. The use of data in decision making was optional
until recently, but it has now become an essential element of running a business. Many companies
such as Amazon, Capital One, Google, Hewlett Packard, Proctor & Gamble, Target, and Walmart,
among others have benefitted through the use of analytics. During this research, we unearthed several
companies that are using data innovatively. With increasing computing power and the availability of
new technology for capturing and processing data, there will be sufficient opportunity for companies
to predict what their customers want even before their customers indicate their choice.
Significant improvement in managing the organizations can be achieved by using simple analytical
techniques. Many organizations are plagued by inefficiencies in operations and supply chain
management and use of analytical tools such as lean and Six Sigma can help organizations remove
the inefficiencies. Major success factor of data driven decision making will be innovation, Each stage
of analytics such as what questions to ask, what data to use, what models to build and how to use the
model outputs need innovative approach and that will differentiate high performing companies from
mediocre companies.

35

Appendix 1: List of Software Tools in Analytics
Sl. No. Software Description
1. CPLEX Cplex is known for efficiently performing many types of mathematical optimisations.
It has been used to solve problems based on linear programming, mixed integer
programming, quadratic programming, and quadratically constrained programming.
http://www-03.ibm.com/software/products/en/ibmilogcpleoptistud
2. Excel Excel is a commonly used spreadsheet application with excellent reporting and dashboard
features. It is widely used for its graphical and calculation abilities, and supports a macro
programming language, VBA. Its advanced version can handle up to 1 million rows.
http://office.microsoft.com/en-in/excel/
3. FICO Xpress Fico provides easy and robust methods to solve business optimisation problems. It has 3
versions: Xpress-Mosel, for powerful and quick optimisation modelling; Xpress-Insight,
which allows users to deploy optimisation models as applications without doing any
additional development; and Xpress-NonLinear, which solves non-linear optimisation
problems with great accuracy.
http://www.fico.com/en/products/fico-xpress-optimization-suite
4. Gurobi The Gurobi Optimizer is a state-of-the-art solver for mathematical programming. Solvers
are included for: linear programming (LP), quadratic programming (QP), quadratically
constrained programming (QCP), mixed-integer linear programming (MILP), mixed-
integer quadratic programming (MIQP), and mixed-integer quadratically constrained
programming (MIQCP).
http://www.gurobi.com/
5. Matlab Matlab is a high-level fourth-generation computing language that allows data analysis,
matrix manipulation, programming, and the creation of models as well as the creation
of user interfaces. Using add-on toolboxes, it can be extended to specific areas of
functionality, such as statistics, finance, image processing, etc.
http://www.mathworks.in/programs/nrd/buy-matlab-get-price-request.html?
6. Minitab Minitab lets users set the order in which the values appear, so that the results are always
presented clearly and logically. Minitab makes it easier to visualise the data and to
recognise important patterns and relationships. Minitab’s layout tool helps to assemble
individual graphs into a single display.
http://www.minitab.com/
7. RapidMiner RapidMiner is an easy-to-use visual environment for predictive analytics. The RapidMiner
Studio studio has application wizards for predictive maintenance, sentiment analysis, and direct
marketing.
8. R R is a free programming language and software environment that provides a wide variety
of statistical and graphical techniques. It is highly extensible and provides effective data
handling, data manipulation, and storage facilities.
http://rapidminer.com/products/rapidminer-studio/
9. R Studio R Studio is a powerful and productive UI for R. It is a comprehensive data recovery

program renowned for its ability to recover mission-critical data that was lost to viruses,
malicious attacks, hardware failure, and OS crashes.
http://www.r-studio.com/
10. SAS SAS is one of the largest market shareholders for advanced analytics. It provides a great
platform for data mining, managing, and retrieving. It also has features that support
statistical analysis and predictive analytics.
http://www.sas.com/en_us/home.html
11. Salford Salford specialises in providing predictive analytics and data mining tools for businesses.
Its four main data mining products are CART, Tree Net, MARS (Multivariate Adaptive
Regression Splines), and Random Forests.
http://www.salfordsoftware.co.uk/
36

12. SPSS Modeler The SPSS Modeler builds and deploys predictive models directly into the business
processes with decision management. The SPSS Modeler can handle text analytics, entity
analytics, and social network analysis, along with automated modelling and preparation
techniques, which help to unlock insights from almost any type of data.
http://www-01.ibm.com/software/analytics/spss/products/modeler/
13. SPSS Statistics SPSS Statistics is a software package for statistical analysis, which is extensively used
for predictive analytics. Its applications include analysing survey data, data mining, text
analytics, prediction of numerical outcomes, and identification of groups.
http://www-01.ibm.com/software/analytics/spss/products/statistics/buy-now.html
14. Stata The Stata statistical software provides fast and easy-to-use tools for data analysis
and management and for graphics. It can also be used to perform statistical analysis,
simulations, and custom programming.
http://www.stata.com/
17. WEKA Weka is used extensively for performing data mining procedures. It includes tools to
perform functions such as data pre-processing, classification, and visualisation. It also
supports the development of new machine learning schemes.
http://www.cs.waikato.ac.nz/ml/weka/
18. QlikView QlikView provides a flexible platform that enables the easy consolidation and analysis of
data. It is easy to learn and has a lot of great data visualisation features.
http://www.qlikview.com/
19. Tableau Tableau facilitates the easy analysis and visualisation of any data variety with its major
products—Tableau Desktop, Tableau Server, Tableau Online, Tableau Public. It also has
tools for the quick creation of powerful dashboards.
http://www.tableausoftware.com/
20. Targit Targit is the largest developer of business intelligence and analytics software that excels
in natural language processing, data visualisation techniques, and database retrieval
processes.
http://www.targit.com/en
21. SAP Lumira SAP Lumira redefines self-service BI and provides the fastest ways to get answers from
the data (without IT help). It creates informative visualisations that combine data from
multiple sources into a single view.
http://www.sap.com/index.html

Acknowledgements
We interviewed several senior managers and leaders while preparing this report. We would like to
thank the following thought leaders for their time and insights.
1. Ami Shah, Founder, IntelliAssist

2. S. Anand, Chief Data Scientist, Gramener
3. Ananth N. Rao, Head of Business Development, Apollo Hospitals
4. Cyrus Dhabhar, Head, Process and IT, Volvo Buses
5. Pramod Singh, Director, HP
6. M.G. Raghuraman, Senior Vice President & Chief Information Officer, Mphasis
7. Satish Kumar, Co-Founder and Director, Axcend Automation and Software Solutions
8. Satyajit Sarkar, Head, Global IT, DTDC

References
Brody, H., Rip, M.R., Johansen, P.V., Paneth, N., and Rachman, S. (2000). Map Making and Myth Making in Broad Street:
The London Cholera epidemic 1954. The Lancet, Vol. 356, pp. 64–69.
Bruhadeeswaran, R. (2012). Shubham Housing Development Finance raises $7.8M from Elevar, Helion, Others.
Vccircle. Available at: www.vccircle.com/news/2012/11/12/shubham-housing-development-finance-raises-78m-
elevar-helion-others.
37

Coles, P.A., Lakhani, K.R., and Mcafee, A.P. (2007). Prediction Markets at Google. Harvard Business School Case
(No 9-607-088).
Devonport, T.H. (2006) Competing on Analytics. Harvard Business Review.
Devonport, T.H. and Patil, D.J. (2012) Data Scientist: The sexiest job of 21st century. Harvard Business Review, pp. 2–8.
Devonport, T.H., Iansiti, M., and Serels, A. (April 2013). Managing with Analytics at Proctor and Gamble.
Harvard Business School Case (Case Number 9-613-045).
Duhigg, C. (2012a). How Companies Learn Your Secret. New York Times, February 16, 2012.
Duhigg, C. (2012b). The Power of Habit: Why we do what we do in life and business. William Heinemann, London.
Hayes, B. (2013). First Links in the Markov Chain. American Scientist, Vol. 101, pp. 92–97.
Hays, C.L. (2004). What Wal-Mart Knows about Customers’ Habits. New York Times, November 14 2004.
Hopkins, M.S., LaValle, S., Balboni, F., Kruschwitz, N., and Shockley, R. (2010). 10 insights: A first look at the New
Intelligence Enterprise Survey on Winning with Data. MIT Sloan Management Review, Vol. 52, pp. 21–31.
Howard, R. (2002). Comments on the Origin and Applications of Markov Decision Processes. Operations
Research, Vol. 50, No. 1, pp. 100–102.
Kant, G., Jacks, M., and Aantjes, C. (2008). Coca-Cola Enterprises Optimizes Vehicle Routes for Efficient Product
Delivery. Interfaces, Vol. 38, pp. 40–50.
Leachman, R.C., Kang, J., and Lin, V. (2002). SLIM: Short cycle time and low inventory in manufacturing at Samsung
Electronics. Interfaces, Vol. 32, pp. 61–77.
Lewis, M. (2003). Moneyball: The art of winning an unfair game. W W Norton & Company.
Mahadevan, B., Sivakumar, S., Dinesh Kumar, D., and Ganeshram, K. (2013). Redesigning Mid-day Meal Logistics for
the Akshaya Patra Foundation: OR at work in feeding hungry school children. Interfaces, Vol. 43, No. 6, pp. 530–546.
Mayer-Schonberger, V. and Cukier, K. (2013). Big Data: A revolution that will transform how we live, work and think.
John Murray.
Scott, N. (2013). Big Data: What is the big deal? Technology, pp. 61–65.
Siegel, E. (2013). Predictive Analytics: The power to predict who will click, buy, lie or die. John Wiley and Sons,
Hoboken NJ.
Sirkin, H.L., Keenan, P., and Jackson, A. (2005). The Hard Side of Change Management. Harvard Business Review, Vol.
83, No. 10, pp. 109–118.
Snow, S.J. (1999). Death by Water: John Snow and cholera in the 19th century. Liverpool Medical Institution. Available
at: http://www.lmi.org.uk/Data/10/Docs/11/11Snow.pdf.
Savant M V (1990). Ask Marilyn, Column in “Parade Magazine”, Page 16, 9 September 1990.
Simon, H. (1972). Theories of Bounded Rationality, in “Decisions and Organizations” by McGuire C. B. and Radner R
(Eds.). North Holland Publishing Company, 1972.
Stelzner, M A. (2013). 2013 Social Media Marketing Industry Report – How Marketers Are Using Social Media to Grow
Their Businesses. Social Media Examiner Report 2013.
Tandon, R., Chakraborty, A., Srinivasan, G., Shroff, M., Abdulla, A., Shamsundar, B., Sinha, R., Subramaniam, S., Hill, D.,
and Dhore, P. (2013). Hewlett Packard: Delivering Profitable Growth for HPDirect.Com Using Operations Research.
Interfaces, Vol. 43, No. 1, pp. 48–61.
Suhruta K., Makija K., and Dinesh Kumar U. (2013). 1920 Evil Returns – Bollywood and Social Media Marketing”, IIMB
Case Number IMB437.
Underhill, P. (2009). Why We Buy: The science of shopping. Simon & Schuster Paperbacks.
38

39

Is there a new
customer hiding
in your data?
Most organisations only use 1% of the data available to them. Smarter
enterprises are analysing more of their data to find more of their customers.
By uncovering entirely new markets and improving experiences for existing
customers, some businesses are getting more than `450* back for every
rupee spent on analytics.
This is Big Data & Analytics on a Smarter Planet.
O&M 07
www.ibm.com/in/findcustomers
IBM, the IBM logo, ibm.com, Let’s Build A Smarter Planet, Smarter Planet and the planet icon are trademarks of International Business Machines Corp., registered in many jurisdictions worldwide.
A current list of IBM trademarks is available on the Web at www.ibm.com/legal/copytrade.shtml. © International Business Machines Corporation 2013.
*Conversion rate as of 15th January 2014, 1 USD = 45 INR
40

Download the Salesforce1 Mobile App now


Data Driven Decision Making

Hochgeladen von

Dokumentinformationen

Originaltitel

Copyright

Verfügbare Formate

Dieses Dokument teilen

Dokument teilen oder einbetten

Freigabeoptionen

Stufen Sie dieses Dokument als nützlich ein?

Sind diese Inhalte unangemessen?

Copyright:

Verfügbare Formate

Data Driven Decision Making

Hochgeladen von

Copyright:

Verfügbare Formate

Data Driven Decision Making.

indd 1 24-02-2014 PM 05:42:29

Data Insights to big data, small data, all data.

To learn more, visit microsoft.com/server-cloud

Data Driven Decision Making.indd 2 24-02-2014 PM 05:42:30

Data Driven Decision Making.indd 1 24-02-2014 PM 05:42:31

Data Driven Decision Making.indd 2 24-02-2014 PM 05:42:31

Data Driven Decision Making.indd 3 24-02-2014 PM 05:42:31

Data Driven Decision Making.indd 4 24-02-2014 PM 05:42:32

1. Identify the problem or opportunity.

Stage 4 Stage 5 Stage 6

Figure 1. Data driven decision making flow diagram

Data Driven Decision Making.indd 5 24-02-2014 PM 05:42:32

Data Driven Decision Making.indd 6 24-02-2014 PM 05:42:32

Data Driven Decision Making.indd 7 24-02-2014 PM 05:42:32

Data Driven Decision Making.indd 8 24-02-2014 PM 05:42:32

Data Driven Decision Making.indd 9 24-02-2014 PM 05:42:33

Data Driven Decision Making.indd 10 24-02-2014 PM 05:42:33

Organisation Predictive Analytics Model

Data Driven Decision Making.indd 11 24-02-2014 PM 05:42:33

Data Driven Decision Making.indd 12 24-02-2014 PM 05:42:33

Data Driven Decision Making.indd 13 24-02-2014 PM 05:42:33

Table 2: Predictive and Prescriptive Analytics Tools

Analytics Tools Applications

Classification Trees Chi-Squared Automatic Interaction Detection (CHAID) and Classification

Data Driven Decision Making.indd 14 24-02-2014 PM 05:42:34

Combinatorial Combinatorial optimisation involves choosing the optimal solution from a

12 Source: http://math.uchicago.edu ~shmuel/Network-course readings/Markov_chain_tricks.pdf

Data Driven Decision Making.indd 15 24-02-2014 PM 05:42:34

1. What is the most effective social media tactics?

Data Driven Decision Making.indd 16 24-02-2014 PM 05:42:34

Data Driven Decision Making.indd 17 24-02-2014 PM 05:42:34

Data Driven Decision Making.indd 18 24-02-2014 PM 05:42:34

Data Driven Decision Making.indd 19 24-02-2014 PM 05:42:34

Problem or Opportunity Identification

Collection of relevant data

Communication of data analysis

Figure 3: Framework for Data-Driven Decision Making

Data Driven Decision Making.indd 20 24-02-2014 PM 05:42:34

Creating A House of Analytics Excellence for Current Practitioners

Figure 4: House of Analytics Excellence

Data Driven Decision Making.indd 21 24-02-2014 PM 05:42:34

Define Analytics Strategy

Identify Sources of Data and Develop Data Collection Plan

Figure 5. Roadmap for Analytics Capability Building for SMEs

Data Driven Decision Making.indd 22 24-02-2014 PM 05:42:34

Table 3. Industry wise analytical problems and data resources

Industry Sector Sample Analytical Problems Data Sources

Manufacturing Supply chain analytics: Inventory Procurement, sales and

Retail Assortment Planning: Category and SKU Price data.

Data Driven Decision Making.indd 23 24-02-2014 PM 05:42:35

Challenges in Data-Driven Decision Making and Future

Data-Driven Decision Making in Practice: Survey Results

Data Driven Decision Making.indd 24 24-02-2014 PM 05:42:35

Which sector does your company/organisation belong to?

How many employees work in your company/organisation?

Data Driven Decision Making.indd 25 24-02-2014 PM 05:42:35

What tool(s) is predominantly used for decision making in your organisation?

Data Driven Decision Making.indd 26 24-02-2014 PM 05:42:36

Very Good 29%

0% 10% 20% 30% 40% 50%

How would you rate your own understanding of analytics?

Very Good 26%