A couple of weeks ago I got an opportunity to be a part of a panel discussion at ‘The Economic Times Business Analytics Summit‘. The topic of the discussion was ‘overcoming the challenges of bringing data science to the boardroom’. The panel had a well-balanced representation from both industry and academia. It was an interesting and thought-provoking discussion.
Later I thought about the topic in a greater detail and structured my ideas further. I believe the following four factors collectively play a pivotal role in bringing data science to the boardroom. In the tradition of marketing I have decided to call these factors the 4 Ps and they are:
- Platform – quality of data and data infrastructure in the organization
- People – quality of people representing data science in the organization
- Problems – questions data scientists try to answer in the organization
- Position – with whom the data scientists work within the organization
There is a general consensus that presence of data science brings objective thinking, and fact-based management to the boardroom. Moreover, in the new economy, competitive advantage for an organization is determined by its understanding of customers and market trends. Data science empowers organizations with this deep understanding and also provides sound strategic direction. Therefore, in short, the presence of data science in the boardroom determines the competitive superiority of the organization.
Now let’s discuss the four pillars or the 4 Ps to bring data science to boardroom in some detail:
1st P – Platform
Data science relies on easy access to good quality data for quick wins. In my experience, I have noticed that most mid and large size organizations are data rich because of fairly well managed transactional systems. The problem is usually with how this data is handled for analysis in these organizations. Traditionally data warehouses and more recently big data platforms are the suggested solutions for this problem. However, in the absence of scientific thinking coupled with business acumen these solutions rarely become effective. Gartner reports that more than half of data warehouse implementations completely fail or have limited acceptability within organizations. Primarily, since they are developed without the complete involvement of data scientists who understand business and analytics requirements. In the absence of an effective data platform, a lot of data science work happens in silos with unnecessary work for data gathering and processing. Hence,a scientifically designed data platform is the first step for an organization to reap the true potential of data science and excel in the digital economy.
2nd P – People
I believe the most important factor that will drive data science to the boardroom is the quality of data scientists in the organization. Most people confuse data science with software tools, however like all scientific endeavor data science is driven by creative and rigorous thinking of its practitioners. In addition, one of the crucial roles of data scientists in the organization is to communicate the power of their science to the management and the board members. This is not an easy task and requires expertise in critical business thinking, number crunching / mathematics, creativity, and interpersonal skills. At present, I see a massive shortage of professionals with the combination of these skills. This is where both academia and industry need to collaborate well to produce the next generation of professionals. There are small signs of this when Geoffery Hinton divides his time for research in neural networks and deep learning between Google and the University of Toronto. Similarly, Andrew Ng works with both Baidu and Stanford University. However, right now these collaborations are limited to extremely niche areas.
Academia, especially at the university level, needs to reassess its traditional ivory tower status and become more accessible to industry. At the same time, the industry needs to communicate the excitement of solving these practical problems to academicians to keep them interested. I believe, data science is a perfect ground for collaboration between academia and industry. This combined effort will speed up the process for companies to generate exponential financial benefits from data science, and keep the boardrooms interested in investing time and resources into data science.
3rd P – Problems
The third factor that drives data science in the boardroom is the nature of questions data scientists answers in the organization. Let me illustrate this by quickly discussing a couple of problem statements I have worked on in the recent few years. By the way, both these problems are really “cool” but one of them has a greater interest for the boardroom.
1. Artificial intelligence driven image analytics : extracting information from unstructured data such as images and text is one of the coolest problems for data scientists. Image analytics, in particular, has several applications including automated analysis of images from geostationary satellites, machinery parts, medical images, CCTV footage etc. It’s a fact that some of the best brains in the world are grappling with problems involving image analytics. But purely from the perspective of data science entering boardrooms, an image analytics solution has significantly lower chances of getting a full conversation in conventional boardrooms.
2. Improve P&L (income statement): the second problem that I have worked on has the statement around reducing cost pressure on the P&L. This open-ended problem didn’t have a crisp statement for specific data science tasks. The idea was to scientifically identify areas of revenue leakages and identify driving factors to cut down on cost. This required rigorous exploratory data analytics coupled with business consulting to generate actionable insights. Conventionally business and strategic consultants used to solve a problem like this however in the digital age data scientists have infiltrated in this territory. The objectivity measurement and fact-based insights that data science brings to the table have a greater appeal for the management and the boardroom to implement analytics solutions.
It is clear that anything that will directly impact the financial statements, and produce deeper business understanding will have a greater acceptance in the boardroom. An image analytics problem is a point solution, however, to enter the boardroom data scientists need to take the holistic view of the organization.
4th P – Position
Finally, with whom the data scientists work in the organization will ensure the presence of data science in the boardroom. In many organizations, each department has their individual data science / analytics team. Products, marketing, risk, procurement all run their analytics in silos. I think this is really dangerous since it is hard for data scientists to be objective in this scenario. Data science, to perform at its best, has to be seen as a strategic function. It has to directly roll up to the CEO or the board.
OK, I am going to cheat a bit here , I have a fifth P as well.
5th P: Process
Data science needs to get embedded in business processes for it to be a permanent figure in the boardroom. Data science is not just an intellectual exercise but needs to deliver on the ground as well. There is no alternative to bringing huge changes without experiencing ground realities of business operations in the organizations. Credit scoring is one of the successful examples where data science-powered business processes are delivering high value for organizations. The grand goal for data scientists is to empower every critical process in the organization through data science to deliver quantitative benefits. Data scientists need to be objective in measuring their own performance – as they are expected to be for the rest of the organization.