Every business organization regardless of size faces a changing landscape of potential challenges and problems. Some of these problems can be attributed to a changing technological landscape: new technology tools, new web applications and potential competitors whose bar to enter a market is constantly being lowered by available technology. In such a dynamic business environment how does a business entity stay relevant and even grow in face of the myriad of challenges? The answer might be simpler than one can imagine. It is the possession of market and customer data that can keep a business entity relevant and help it thrive in the face of all these technological advances. In this article we outline the key steps a business entity can take to build a data analytic culture.
Every data strategy begins with a definition of key business problems. The two large categories of data science help define the problem:
Is the problem related to direct human needs such as which content to license, sales leads to follow, which medicine is less likely to cause an allergic reaction, what is the best strategy to use to clean a contaminated site, how to best monitor a farm, best investment strategy etc.
The second broad area is the application of data science intended for machines. Examples are technologies for self-driving cars, gyroscopes for Segway and other moving devices, recommendation systems such as Amazon and Netflix use to suggest next purchases for customers.
Both of these areas of data science cannot be initiated without the collection, cleansing and aggregation of credible data. Consequently every data science strategy begins with defining the key problems a company wants to solve with the help of data. The key stakeholders are also key in defining the data analytic strategy.
For the first type of problem, the key stakeholders tend to be executives (or the decision makers), product managers, designers, clinicians, governmental decision makers etc. These key stakeholders are responsible for helping define, design and implement key operational metrics. They help in defining which experiments to run and how to interpret the results, what dashboards to create and monitor and generate recommendations for modeling and measurement of overall business performance.
Examples of such business objectives for data analytical projects may try to address questions such as
a. How do we improving profit margins in an environment of increasing business competition?
b. How do we increase market share of a product or customer base?
c. How do we reduce business costs without applying the simplistic approach of laying off employees?
d. How do we gain better insights into customer preferences and improve customer retention?
e. How do you better select risks as an insurance company or how do you reduce default rates on loans for banks and other financial institutions?
f. How do you create an effective and efficient triaging system for a hospital, airport logistics, workers’ compensation claims systems etc.?
Answers to the above questions cannot be arrived at with simple anecdotal arguments or qualitative arguments based on individual biases. One needs credible quantitative and qualitative evidence to understand what the main variables or levers are that drive an outcome. Through this quantitative understanding of the problem, one can then measure the magnitude of the problem and devise strategies whose outcomes are measurable in addressing the problem at hand.
Because credible data is so important to running a successful business in a modern and highly technologically dynamic environment, data is considered the new gold of the 21st century. Companies that invest in gathering and maintaining high quality and rich customer and market data stand to remain competitive in the long-term. At the moment, companies such as Google, Facebook and Amazon are mostly into the business of data collection, mining and analysis. In fact most telecom companies are interested in harvesting consumer data. A telecom company may use the location of a client to send targeted marketing messages such as the proximity of restaurants, stores and services in the area of a customer. Such targeted marketing messages have been proven to be more effective than blanket messages sent to a wide segment of customers for which most clients will not be interested.
Data science is finding vast applications in the area of agriculture. Smart farming is the breakthrough application of science and technology in the field of agriculture. The application of technologies like the Internet of Things (IoT), Big Data and Analytics in addition with Machine Learning has brought insights into improving soil quality, crop yield and risk assessment of crops related to climate change. Different crops require different soil nutrients to flourish and data science technologies help refine and deepen that knowledge. Data science in general offers tremendous potential for improving cost to yield ratio, optimize crop yield and provides a platform to engage banks, insurance companies and government agencies to communicate and provide the needed services for the agricultural sector.
The educational sector is an area that can benefit greatly from data science application in understanding the drivers of student success such as economic factors, school income and other demographic factors. Understanding what makes a school and its students successful may influence how to fine tune the curriculum and funding at various schools to maximize the success rates of students and schools.
In future articles we will discuss in more depth how these applications can be developed for various industries and touch on a few technical models used in building these applications.
The second type of data science models are geared for machines. Machines can be built to becoming self-learning objects. Self-driving cars and gyroscopes are the most common examples. But modern cities are incorporating machine learning tools in traffic systems that automatically redirect cars using sensors and traffic systems to reduce congestion during heavy driving times in the day or week. Some modern weapon and rocket systems have been designed to self-destruct in areas that will minimize human casualties when the systems go awry.
A developing nation like Ghana can benefit tremendously in various sectors by incorporating the applications of data science into the decision making process. And this process can be built, developed and maintained by introducing data science into the curriculum of high school and university programs.
Dr. Albert Essiam is the Data & Analytics Lead at OZÉ, a Ghana-based business that helps businesses and banks use data to make more profitable decisions. Dr. Essiam holds a PhD from MIT. If you want to learn how OZÉ can help your bank better model risk, email firstname.lastname@example.org.