BW Communities

Betting Big On Big Data

Rajeev Dubey Nov 08, 2014

<div>As of 2012, humans and machines generated almost 2.5 quintillion bytes of data every day. For every byte of structured data that is being properly mined, there are 4 bytes more of unstructured data. According to IBM, of all the digital data in the world today, almost 90 per cent was generated in the last two years alone. There is close to 3 zettabytes (around 3 billion TB) of data in the world today, and this is throwing up unprecedented challenges for organisations trying to make sense of it. So much so, that an entire industry is developing around ‘Big Data’. On the sidelines of the Nasscom BPO Summit 2012 in Gurgaon, BW|Businessworld and the National Association of Software and Services Companies (Nasscom) organised a roundtable to discuss the new ecosystem around Big Data. The panelists included Som Mittal, president, Nasscom; Mohit Thukral, senior vice-president, Genpact; Roopa Kudva, CEO and MD, Crisil; and Sundar Ramaswamy, CEO, AbsoluteData. Excerpts from the roundtable: Rajeev Dubey: The best explanation I have heard about Big Data is that when GBs become ZBs — leapfrogging TBs and PBs (petabytes) — you are talking ‘Big Data’. But the rate at which it is being generated, this could be a dated definition faster than we think. Also, there is a lot more to Big Data than just the data itself, such as the 4 Vs —  volume, velocity, variety and variability. So, can we have Som Mittal to start off with whether what we are talking about Big Data is completely new or is it an amalgam of many things? Som Mittal: All of what you said. In the early 1990s, the PC was 40 MB. What you were computing was not really data. All that has changed (with) 10 GB pen drives today. And what has enabled this change is the cloud. The cost of storage has become very low, (and) that is an opportunity for Indian companies...because of something new. Earlier, (a) data was not there, (b) when data was there, there was no technology to mine it, (c) when both options were available, there were no resources, and (d) it wasn’t affordable. All the four parameters are now coming together. It was a natural extension for us to leverage this. (Our customers are now) saying that you handled our supply chain, our transactional process and customer interaction; you have gathered some data, now what more can you do for us? That’s the driver. Rajeev: Mohit, so is it old wine in a new bottle, or is Big Data really different from the data centre and analytics combination that has always existed? <img align="middle" alt="" src="/image/image_gallery?uuid=a58dfcc4-c9e4-48f8-82b0-4c6bcf167b02&groupId=520986&t=1362656076757" /> Mohit Thukral: It is new. The magnitude of data is a lot. We have over 1 trillion devices today, from 300 million devices over eight years. Firms in India and around the world see so many transactions going through the system. What you do with these transactions is going to be the key. Today, a lot of firms don’t do market research; they use social media network and tools to do market research, which never happened five years ago. As more and more of us get on to more mobile devices, more data will flow, and you will require more computing. So, I think, we are at the cusp of a change. Rajeev: Roopa, how real is it, especially from the point of view of the 4 Vs? Roopa Kudva: It’s not that we are talking about stuff that’s already being done, (and) which we are seeking to do more efficiently with more value addition. This is the first time this is going to happen. Traditionally, companies, government and enterprises of all kinds always captured data and the focus was internal. Like, to improve efficiency, for forecasting in engineering firms, etc. Why this is transformational and new is because for the first time, it is about understanding the customer better and figuring out new opportunities to grow. The other new element is the variety of sources from which the data is being captured — internal sources, security cameras, other devices and tweets — diverse and numerous. Capturing data is new, quantum of data is new and, more importantly, what you are seeking to use this data for, is new. Rajeev: Sundar, what’s the big challenge in Big Data? Is it handling the data itself, or making sense of it? Sundar Ramaswamy: For me the Big Data challenge is taken away with technology trying to capture that data pretty well. The bigger challenge is when you look at the variety of data and how you can get those sources together to make a consistent meaning in a compressed manner; to be able to assimilate different types of data to make sense out of it. It is going to be unstructured, so you are looking into the quality of data. The bigger challenge is how you overlay the analytics over it (technology) to make a consistent, actionable, real-time meaning for people to react to. Rajeev: There is a need for faster processes, storage and retrieving. Can computing, and the other areas around it, keep pace with Big Data? Som: I don’t think today we have enough computing power. (But since) cloud is enabling sharing of both computing resources as well as the storage sources, it is not going to be a constraint. Mohit: There is so much data that you can keep juggling and computing it. But, at the end, companies will have to get smarter to pull out what (they) are looking for. Sundar: Many companies are making investments in technology and are asking consultants about what they should do with all of this. I really don’t know how I am going to make a difference to getting the analytics out to make meaningful decisions. Clients would like non-human intervention in algorithms to go. They want to create a Gen-I algorithm, which has human intervention but the improvements of those algorithms should be built into the technology we have overlaid, until there is a disruptive need for human intervention on those models. We need people with a background in computer science and psychology to come back to analytics and help this field to grow. The second piece is IP (intellectual property). People who can create IP in this field will differentiate the winners from the losers. Roopa: Where technology and Big Data (come together), it is called ‘visualisation’, that is, visually depicting what you see from the data. Not just simple crafts, charts, but heat maps. However, the question of what to look for in the heap of data needs to be addressed. Rajeev: How much of the unstructured data are we trying to make sense of?  How much of it is being captured and how much is going waste? Mohit:  A lot of the data today doesn’t get captured. There will be a lot of unstructured data available which is critical and you need to pull out the relevant information, and make it more adaptable to your business (and) your customers. How to improve efficiency in a business? How to increase your market share? What customers are saying about your company, product or services? Social networks, where people are blogging or tweeting about you, are unstructured data because they are all over the place. Sundar: To answer your question straight, we haven’t done enough. Having said that, I think there is a lot of investment being made in trying to make that happen. People will need to look at our traditional algorithms and re-engineer them. To give you an example, if we did market mix models right now, we would never have considered social media as an input. Now people have to rethink ways of doing it. Is it moving at the right speed, ‘No’. Som: There is a big distinction between all of us in our industry. We have been working for ‘clients’. Every time someone uses the term ‘customers’ we are not talking about our customer. When we are talking about Big Data, (it) actually (belongs to) our client’s customers. The domain is no longer a technology or an expertise domain; it’s understanding our customers’ business so that we can advise them on the data being generated. Rajeev: How is the business process outsourcing (BPO) industry trying to leverage Big Data and where has it been more successful than elsewhere? Mohit: The business process management (BPM) industry is pretty ahead in the game. The two main users of data are financial services and retail. Because they are B2C (business-to-consumer) businesses and more consumer-oriented, we have built a scalable business. We have mathematicians, statisticians, econometricians, etc, and built a Big Data or analytics business. The next evolution for firms in the business analytics area is going to be to understand the end customer, their business needs and domains. You have to look at how you build products that are used by your customers and then how do you service those products. Roopa: The estimate is that the (global industry) size was $5.3 billion in 2011, and by 2015 it will be $25 billion. That’s a 46 per cent CAGR (compound annual growth rate). For India, that number will go from $200 million to about $1 billion. The three sectors that will account for the bulk of this are financial services, telecom and retail. But, for India, we are saying that in this path from $200 million to $1 billion, in the first year (2012) the mix is going to be 84 per cent technology and 16 per cent analytics, but five years down, analytics will be 30 per cent. Initially, you need to invest to set up the data architecture, technology, the backbone, the hardware and in designing the whole system. Sundar: The global stage is on an equal footing. Nobody’s really cramped the court there. Given our talent, our experience to really take a lead in shaping how Big Data gets played out, we have the right kind of investment and scale to run those experiments, to leapfrog, create our IP and lead the charge. That is the opportunity I would exhort India as a service industry to do. Roopa: With one nuance though…we’re now talking about cutting-edge IT stuff in terms of analytics and I believe that initially the Indian BPO industry will also have to learn to leverage global talent for this in a way it hasn’t done before. It would not be unusual to see Indian firms putting together a team of 15-20 people of PhD source from all over the world, from the US, eastern Europe, India. That’s good for the Indian BPO industry because it’s all about the shift from BPO to BPM. Rajeev: What’s the weakest link in the Indian Big Data ecosystem and abroad, and why is that lagging? Sundar: One, is talent. While we say we have a great mathematical logical structure in our education system, there’s a long way to go to make them analytics-ready. Second, is changing our own mindset about upfront investment and IP, and leverage top talent to be able to do it. If we miss that boat, we will lag. Mohit: Why you would need global teams to work on this to create cutting-edge changes is because a lot of the domains in those industries lie outside the country. <img align="middle" alt="" src="/image/image_gallery?uuid=50cd1e7e-1da9-419c-9149-d73f9db3aaa1&groupId=520986&t=1362656289099" /> Som: We’re talking about a new class of people being hired. The fact is that we are catching this opportunity early, and trying to carve it out later. Rajeev: What is triggering the wave of M&As in the industry? Is it the innovation or lack thereof? Som: The main reason for M&As is (that) you cannot build domain expertise through on-boarding and training in classrooms; it’s such a slow process. You need to do acquisitions so you get domain experts. A majority of acquisitions — while there are some that may have been done for revenue accretion — are to get people with knowledge. Mohit: One is to acquire expertise. The M&As in this space are niche firms, not massive players like IBM. These are $20-30-40-million firms and you are just plugging them into your larger pool for speed to market. Rajeev: What’s going right for Big Data is that regulators in financial services, healthcare and telecom are backing it. So, what’s in it for the regulators? Mohit: Regulators want more information to share, so they want companies to be more transparent. Companies are getting swarmed with regulators asking for more and more information. Specifically in the US financial services today. Companies will have to get smarter to make data available, else costs are going (to go) up. Compliance costs have gone up 15-20 per cent. Rajeev: Coming back to the $200-million-to-$1-billion opportunity, where do we stand in exploiting this? Roopa: We’re early in the game. It’s new for everybody. We are there, which is good; understanding that there’s an opportunity and (that there are) initial investments. Many new firms have come up, that is testimony to the fact that India is moving to capitalise on this opportunity. Rajeev: But, if we look at it right now, the Big Data work here  is mostly with the captives? Roopa: I would not agree. You see a nice mix of captives and third-party firms. It’s gone beyond an arbitrage. It’s about knowledge, capability, skills. It will continue to be a mix of in-house as well as sourcing from third parties. (This story was published in Businessworld Issue Dated 25-03-2013)</div>