Admissions Closing Soon. Limited Seats Available: Data Science: Waiting List | Economics: 2 | Psychology: 2 | Communication Design: 2 | Digital Business: 3 | Legal Studies: 3 Apply Now Download Brochure

Home > Blogs > Important Management Practices Driven by Data Science

Important Management Practices Driven by Data Science

Meet the Author

Prof. R. P. Suresh

School of Data Science, Vidyashilp University

Prof. R.P. Suresh is an award-winning and highly distinguished expert with over three decades of experience. He holds a Ph.D. in Statistics and was bestowed the ‘Young Statistician Award’ by the Indian Society of Probability and Statistics. During his academic and industry journey, Prof. Suresh wore many hats. He was a faculty at IIM Kozhikode for over a decade and also a researcher at General Motors R&D Centre, India. He has been visiting faculty and guest lecturer at institutes including Victoria University of Wellington, Indian Institute of Science, and Indian Statistical Institute. He is currently the Principle Director in Supply Chain Analytics at Accenture Applied Intelligence. Prof. Suresh is also a consultant with numerous organisations like the Department of Telecommunications (Govt. of India), WIPRO, Kirloskar Cummins, TATA Tea, Hindustan Newspapers and IES Officers Training among many others.

Prof. Suresh is a two-time gold medallist and has received awards including GM India President’s honour. He has refereed over 27 professional journals and contributed to several conferences and trade inventions.


Management plays a major role in meeting the needs of a business organization. It also enables the business to compete in the market effectively through the achievement of desired goals. Effective management practices have led companies from loss-making to profitable ones, and from small companies catering to local communities to global companies catering to a large population across the globe. For example, Crompton Greaves Ltd. (CGL) incurred a huge loss in 2000 and 2001. Consequently, the restructuring measures adopted by the company led to its spectacular turnaround (see [1]). CGL is one of the leading companies in power equipment and appliances manufacturing, based in India. Set up in pre-independent India, the company was able to establish itself as a strong player in the domestic market. When it ran into losses for two consecutive years in 2000 and 2001, many analysts and investors were ready to write it off. But CGL introduced effective restructuring measures which resulted in the company’s sprint back to profitability in 2002. CGL then focused on globalization, and has since increased its presence in the international markets.
Today, it has become one of the top five players in its industry, globally. The company used a number of Management techniques to stimulate growth such as Organizational Restructuring, Core Competence, Globalization, Productivity, Value Engineering, Treasury and Working Capital Management, Inventory and Debtors’ Management, Strategic Acquisitions, Corporate Communications, etc. Another such example is the success of Coca-Cola. The company used Global Marketing and Management principles (see [2]) such as forging human connections, remaining innovative while staying true to simple principles, and creating brand experiences established the company as industry leaders and kept them ahead of the competition for over 125 years. Despite its status as a global icon, Coca-Cola understands each country’s requirements at a personal, localized level. Each country’s offerings are customized to their local culture and language, with the most popular names of each region printed on cans and bottles in place of the company’s moniker. This campaign is the perfect example of effectively applying a localized positioning strategy to a global market. Most of the strategic decisions taken by these and many other successful companies are driven by data, which helps in understanding customer requirements, market segmentation, diversification of products, reduction of production and inventory costs, make or buy decisions, diversification of investment portfolio, mergers and acquisitions, etc. In this article, we will outline some of the important roles that managers play in shaping an organization, and the utility of appropriate data in decision making. We will also see how organizations can use data-driven approaches to reach their goal in a short period of time, while delving into how data science driven organizations can help them overcome redundancies in the system, and help them accelerate growth.

Decision Making Under Uncertainty

A manager’s role is to make appropriate decisions that lead to profit and growth of the organization under uncertainties. For example, a Production Planning Manager (PPM) needs to take a decision on what products to produce, how much of each product type to produce, how much of raw materials to be ordered, how to schedule different types of products in a production line, how many shifts to be operated, how many contract employees to be hired etc., for the following month, so as to minimize the overall cost of production for the company, while meeting the customers’ demand.

The PPM will have to take these decisions under uncertain circumstances such as

a) Unknown demand for each product (which is influenced by several external factors such as competing products, price, economic conditions, etc.)

b) Availability of quality raw material on time (which is influenced by vendor’s production environment, logistical issues, etc.)

c) Availability of machines and equipment in good condition (machine failure/uncertain downtime, etc.)

d) Availability of personnel with requisite skill set, etc.

A marketing manager takes a decision on the mode of advertisement / details of the campaign, target customer segment etc., so as to increase awareness of the products/ services and brand awareness. Here, there are uncertainties on the number of followers of the chosen channel/campaign, the consumer behaviour of the identified segment etc.

A bank manager in the loans department of a bank needs to take a decision on whether a loan should be sanctioned to a new entrepreneur. This decision is subject to several uncertainties, including the repayment capability of the borrower, uncertain business environment in the specific industry/product, etc.

An ATM manager has to decide the amount of currency to load into an ATM machine in a locality, which is subject to several uncertainties such as potential bank customers in that locality, number of customers likely to utilize the facility till the next re-fill, actual cash that the customers will be withdrawing from ATM, etc.

A hiring manager or a recruiter usually decides to hire a prospective candidate for a role based on their past experiences and response to the questions or situations presented to them. Even though the candidate has not worked in exactly the same environment, the hiring manager uses his intuition to assess whether the candidate will be able to discharge their role effectively, though there is no certainty that the candidate will be successful in their role and adapt to the organizational culture. In view of the uncertainties in decision making, there are always elements of risks involved in every decision taken by the manager.



The objective of management is to minimize risk. To bring this to fruition and enhance the robustness of a decision, most organizations tend to use relevant data to analyse the challenge before arriving at an optimum solution. For example, by collecting data on the financial discipline of an entrepreneur, and industry growth of the relevant industry, the bank loan manager can estimate risk. If the risk is very high, the manager may reject the loan, and vice versa, thus minimizing the overall risk for the organization. Similarly, by collecting various data such as historical demand data of different products, order fulfilment data of the vendors, failure data of equipment, the PPPM can estimate the risk involved in a production plan, and take a decision accordingly. Thus, a data driven decision-making approach would help understand and reduce the risk for a business. Below, we outline a few examples of data driven approaches and their advantages.


Many banks report large Non-Performing Assets (NPAs) mainly due to bad loans (see [4]). Some of the common problems of NPA management include highly divergent loan appraisal standards, absence of a common database, absence of timely intervention, excessive use of discretion, reliance on business models with suspect assumptions etc. Data driven management systems with a common big data platform - which takes, ingests and processes data related to a business from various sources, including social media, and deploys analytical tools to create effective dashboards with warning signals - can help monitor and take appropriate actions on loans. Also, the use of artificial intelligence driven rule-based tools can help eliminate or reduce subjectivities in appraisals and human errors in assessments, through validations and logical extrapolations, thus reducing the possible number of bad loans converting to NPAs.


In many companies, inventory cost is one of the major costs, which accounts for about 10-20% of sales. Inventory is the quantity of finished products and raw materials that companies keep in stock with a view to sell them or use them for uninterrupted production/manufacturing as and when needed. Inventory managers work towards maintaining the right amounts of products in inventory, since keeping large units in inventory would lead to a huge amount of unutilized working capital, whereas keeping less units in the warehouse may leave them out of stock and lead to loss of sales and customer dissatisfaction. As an example, consider a phenomenon that occurred in Big Bazaar in 2006 (see [5]) when they announced a “Sabse Sasta Din” on January 26, with huge discounts on many items including televisions, refrigerators, food items etc. All the stake holders from marketing, operations, IT and category managers had jointly planned for this event, and worked with several vendors across India to stock up on a good number of items in Big Bazaar outlets. The executives estimated and hoped that this event would trigger higher sales and clear the items they had specially procured for the event. However, what happened was totally different. Each store witnessed a turn-out in such huge numbers that the police had to be summoned to manage crowds. Moreover, they ran out of stocks. Many of their regular customers went home dissatisfied as they had gone through a lot of trouble to enter the store, only to realise that the product they wished to purchase was out of stock. Big Bazaar had to close down their stores by afternoon, and announce that the “Sabse Sasta Din” sale would continue for the next 2 days as well. By quickly procuring more items, they ran the event over the next two days, and realized record sales. It is here that they underestimated demand for the items at the price discounts offered during the event.

There are several factors, such as price, season, competitor/ substitute products etc., that influence the demand. It is important to collect data and estimate the effect of each of these factors effectively so as to estimate the demand and plan inventory accordingly. To see how Data Science can help in inventory planning, let’s look at an example of another retailer based in Europe, who was challenged by high inventory cost. The retailer had an inventory cost of 22% of sales, and also had a huge loss of sales. The analytics experts studied the profile of products in stock, and observed that large items in inventory are those which have competing products in the market, and the only way they could compete in the market was through price discounts. Also, it was noticed that the store ran out of stock for some of the essential products when the customers needed them, leading to loss of sales. The experts recommended an increase in the safety stock of such items. Based on the recommendation, the retailer announced price discounts on select items. This helped them realize an increased demand for these products in a short period of time. At this point of time, they developed an analytical demand planning solution using statistical and machine learning algorithms by taking into account the data of demand, store footfall, customer ratings of products (web-based ratings), marketing variables, weather, etc. This resulted in an improvement of demand forecast accuracy from 40% to 60%, leading to an inventory reduction of 42%, reducing the stock out situation by 50% from its existing level. All of this led to an overall cost savings amounting to about 10% of total sales.


In many companies, particularly in the resource industry, production is heavily dependent on the proper functioning of large machinery. For example, one of the large Oil and Gas manufacturing company based in Australia has over 2800 wells. It is important to ensure that these wells function properly. It was noticed that well failures are a major cause for production loss. It is typically much more cost effective to predict the failures in advance and take appropriate actions to avoid the consequences of failure, before failure occurs. With this view, an analytics project was undertaken by considering various data (downhole pressure, tubing pressure, water flow rate, gas flow rate, etc.). Some of these data are captured by electronic sensors. The first task was to bring all the data together on one platform. Using various statistical and machine learning algorithms, the analytics team developed a predictive model and predicted pump failures for a sample of 40 pumps based on the rate change/transient in rod torque, reduced water flow rate, reduced tubing pressure, spikes in gas flow rate, etc. Visual dashboards were developed to indicate which well needed immediate attention, and this dashboard was linked to the cell phones of the well operating managers, who received an alarm of a potential failure. This helped the technical team to proactively mobilize and take necessary action to avoid failures, and hence increased the productivity of the wells.


As discussed earlier in this article, availability of appropriate data, and extracting insights from the data improves decision making capabilities. However, in most organizations, different departments such as production, supply chain, marketing, finance, human resources, IT, etc., use data that are appropriate to them in arriving at decisions. Most of these decisions are taken independently at a department-level, which leads to inconsistencies and redundancies in the system, conflicts, and increased costs. Hence, it is important for organizations to create a common data warehouse with the integration of data across the organization and develop an organizational culture so that all departments make use of the data available on a common platform. This will help bring-in transparent decision making and a trust-based culture across the organization. It will also encourage teamwork to achieve the common goal of organizational growth and profitability. It was found that companies that rely on data in their organizations and employ a data driven culture have better financial performances (see [6]). Studies show that data driven organizations not only make better strategic decisions, but also enjoy high operational efficiency, improved customer satisfaction, and robust profit and revenue levels. Recent research also shows that data-driven organizations are twenty-three times more likely to acquire customers, six times as likely to retain those customers, and nineteen times as likely to be profitable as a result. It has been observed that companies which use data in their decision-making process are able to increase their profit (see [6]). In fact, many business executives are enhancing their skills to allow them to integrate analytical tools into their business decision-making practices. While intuition does provide an initial hunch or spark and guide you down a particular path, it is data that helps you to verify and explain that idea or hunch that you started with, in a better way, and implement it to benefit the organization. Today, we see that the e-commerce industry uses data to drive profits and sales in an effective manner. All of us would have received a product recommendation while visiting the Amazon website or through email. This is a typical example of a data driven business decision. Here, Amazon uses the data of past purchases, items viewed by the customer and the items in the customer’s shopping cart etc., and using key engagement metrics such as click-through rates, open rates and opt-out rates to further decide what recommendations to push to which customer. By integrating recommendations into nearly every aspect of Amazon’s purchasing process, from product browsing to checkout, the company has found that product recommendations, in fact, do drive sales and increase the bottom line.


i) Understand the Business and Decision-Making Process: For most organizations, lack of data is not a problem. In fact, it’s the opposite: there is often too much information available to make a clear decision. With so much data to sort through, organizations need to have a clear understanding of their business decision-making process and a better data science strategy to support that process. For example, to promote a product, the use of data on customer segments and product preferences as observed in product reviews.

ii) Establish Performance Metrics: In order to successfully translate this vision and business goals into actionable results, the next step is to establish clear performance metrics. The metrics is defined considering each business unit’s objective and designed to improve their performance.

iii) Architect the End-to-End Solution: Organizations need to think more organically about the end-to-end data flow and architecture that will support their data science solutions. Data architecture is the process of planning the collection of data, including the definition of the information to be collected, the standards and norms that will be used for its structuring and the tools used in the extraction, storage and processing of such data. This stage is fundamental for any project that performs data analysis, as it is what guarantees the availability and integrity of the information that will be explored in the future. It can be said that at this point there is an intersection of the technical and strategic visions of the project, as the purpose of this planning task is to keep the data extraction and manipulation processes aligned with the objectives of the business.

iv) Build Data Science Tool Box: Based on the requirements, the organization needs to develop toolkits such as demand forecasting, marketing, human resource planning, warranty planning, spare parts planning, predictive asset maintenance, financial planning etc. Since all these tools use commonly available data on one platform, it will be easy for all the stake holders to trust the solution developed in this manner.

v) Unify Your Organization’s Data Science Vision: Develop interactive dashboards which provide a unified view of all the aspects of the organization. The advantage of such a tool is to understand how a change in any particular function affects other functions, and hence leads to a decision which is easily acceptable by all functions and easily implementable as well. For example, if the marketing department complements a promotion (price discount), this would impact revenue, and would also involve increased inventory, and probably increased production as well. Using the unified tool, one can visualize the production and storage constraints, and then take appropriate decisions collectively.

vi) Keep Humans in the Loop: One of the most important steps is to ensure that the human inputs are considered at each step of the process. For example, the Data Science tool may recommend the reduction of price with reduced features for a product considering external market related data. However, the company’s brand or reputation may be sacrificed if they offer a flagship product at a lower price with reduced features. Human intervention is an important input to be considered to effectively fine tune the recommendation and implement it in the organization.


The success of an organization depends heavily on the decisions taken by the management of the company at various levels. Every organization will have different departments/functions, and the decisions taken by managers of each department depends on the availability and visibility of data. However, these decisions taken by one department could heavily impact other departments, and hence the overall company’s growth and profitability. By following a unified data science approach and having an organizational culture of data driven decision-making process, as explained in this article, companies would be able to achieve higher growth and profitability in an efficient manner.

Study at VU