Krishna Pera was the Head of Global Business Services for the Olam Group. He was the COO of Murugappa Group-owned digital-content services KPO and held multiple leadership roles at companies like Satyam and Siemens.

Krishna currently runs a consulting entity focused on analytics and data-driven decisions, building a couple of Tech-products – a data-driven marketing solution based on micro-market data from 600,000 villages in India & and the other for building a network of online Food Co-ops. 

Furthermore, he is the author of the book named “Big data for Big decisions - Building a Data-driven Organization”

What inspired you to write this book? What were your early challenges when writing this book?

You need to figure out how many among your readers know this: Gartner estimates anywhere between 60-85% of all Big Data AI projects fail to deliver the intended value.

This book describes my journey in setting up a data-driven organization and devising data strategy and governance for global firms. But as I mentioned, I had to supplement it with substantial research: which was more gruelling than I imagined. I had to read dozens of books, hundreds of articles, and whitepapers. I must have cited over a hundred articles in the book. The lockdown helped.

The biggest challenge was that there needed to be books or articles on creating a roadmap and a business case for a data-driven organization or re-engineering the data-to-decisions value chain, except for a few that I published myself in Data Science Central between 2016-19. So a good part of it came from my own experiences.

What is the core message of your book: Big Data for Big Decisions- Building a Data-driven Organization?

This book is a practitioner’s handbook for creating a transformational roadmap and business case for a data-driven organization prioritizing the big decisions, the 10% of the decisions that account for 90% of business outcomes. 

The acid test for a data-driven organization is when all CXO-level strategic decisions are taken based on data and, where possible, with an audit trail. This book intends to Guide enterprises in their journey towards becoming data-driven organizations, prioritizing analytics investments that vastly improve the insights and thereby substantially improve the quality of each Big-decision.

This book is about choosing and prioritizing the right kind of analytics projects, understanding where and what sort of Big data and AI applications need to be built in an enterprise to cover 90% of business outcomes…. while not failing like Gartner predicts and delivering the intended value. Furthermore, this book helps identify and prioritize the most significant opportunities for cost-saving, selling more, building visibility and control, and foreseeing and preventing problems. The emphasis is on creating a repeatable self-priming process model to support big decisions for analytics.

The book promises to provide a fail-safe methodology to benefit from investments in analytics. Please elaborate.

Yes, the book provides a fail-safe methodology to transform any enterprise into a data-driven organization and derive real $ value from analytics. The book also comprehensively covers the data-to-decisions lifecycle transformation, Data-driven IT strategy, Data strategy, Data-driven Marketing, and the all-important ‘Integrated Data Governance’.

Who is this book intended for? What kind of readers will benefit from reading this book?

I think it’s a must-read for CDOs and CIOs with a mandate to set up a data-driven organization. 

This book is written for everyday managers, not just data scientists and techies. Organizations have two kinds of stakeholders, those who consume information (essentially data + insights) and those who produce, curate and distribute information – The IT guys. This book is helpful for both.

The book will be helpful for severe data science & AI students in graduate and post-graduate courses, besides researchers.

Would you like to comment on the rate of adoption of Big data and AI in the industry? What, according to you, is going right, and what is not? 

The pace at which Big data & AI tools are being developed is amazing. While there is increased availability of Big data and AI tools, there are mixed signals regarding adoption. While some industry sectors like Finance and Healthcare are early adopters, others are still cautious. There has been substantial progress in Manufacturing & Auto industries as well. Automation of routine tasks, self-driving cars, and increased use of robots are visible.

The biggest impediment, in my view, is the ‘quality of data’, apart from ‘privacy’ concerns. Almost all transnational companies still have petabytes of legacy data at each geography and thousands of legacy applications that do not talk to each other. The quality of ERP implementations, lack of centralized data governance etc., is a long way to go for most enterprises.

Most CXO needs to understand there is a direct correlation between the data quality and overheads ($) as a percentage of revenue. Poor data quality almost always means ‘far higher than normal’ overheads. 

Data quality in most global multinational enterprises needs to be improved. The larger and more complex the enterprise is, the more likely you end up with isolated, disconnected islands of information. It is because data lakes in most companies are used as data dump yards. Fortunately, it is a different story for new-age Internet companies. They operate as a global single-instance application, so they have no crippling ‘data quality issues.

Do you have any advice for Universities and Engineering colleges designing courses on Data Science & AI?

The universities should notice a few things.

The importance of Decision Sciences: We need to understand the core purpose of data is to aid in decision-making. Serious AI/Data Science students need to study the basics of decision sciences and how exactly to identify the data behind the decisions and build decision models. Unfortunately, most students I meet do not even know the meaning of the ‘DMN’ notation. 

The Data Lifecycle: No one teaches this subject in Universities. Students need to understand the core of data science & AI is ‘the data’… quoting from the book, “ All decisions, big and small, can be data-driven, assuming there is the right data to analyze and draw actionable insights from. The “right data,” however, cannot be produced instantaneously. One could not go back in time to capture the correct data if it was not designed to be captured in that instant. In the case of multinationals, one cannot blackjack the IT systems of each country to produce the standardized-granular data if it was not part of their original design. In a nutshell, systems across organizations must be designed to capture the right data”.

What advice would you provide to students and professionals interested in a career in artificial intelligence?

More than learning to code is required. You will gain confidence only when you learn to build applications, real business applications, and as many as you can. 

I advise the students to intern with Analytics and AI product development companies as and when possible. Engineering colleges need more practical and workshop courses, ideally designed in collaboration with Tech companies. And the colleges must encourage the students to build real-life applications as individual projects, not group projects.