It's nearing lunch time and I'm trying hard to stay focused on the noughts and crosses on the white board in front of me. The venue is the conference room of Fractal Analytics's Powai office in Mumbai, and I'm being given a quick lowdown by analytics head Awadhesh Kumar on Kernel Transforms and Support Vector Machines. Er...Support Vector what? Well, let's simply say that it's an advanced statistical technique for re-classification of variables. And it can help predict, for a bank, say, the likely rate of customers defaulting on a personal loan in a rare event that has no historical precedence.
A statistician and former scientist with GE's Global Research Centre in Bangalore, Kumar belongs to a rapidly growing breed of professionals - comprising statisticians, mathematicians and engineers - who're redefining the way business decisions get taken in the strategy rooms of corporations across the world, including India. They're bringing analytics, or the science of advanced mathematical-based decision making, into the mainstream of business.
Using data to arrive at business decisions isn't new. Simple correlation and trend analysis have routinely been carried out by managers before. But what's driving the new wave of analytics is its use for a higher, more rigorous level of analysis to yield patterns of behaviour that's giving companies a competitive edge in the market.
While businesses across sectors in India - financial services, retail, telecom, insurance, manufacturing and pharma - are beginning to use analytics to compete, financial services firms have been among the earliest adopters. Take ICICI Bank, which began investing in its analytics capabilities as early as 2001. Today its credit scoring model sifts through around 200,000 credit card applications received every month, using highly advanced algorithms to accept or reject, and set credit limits for each customer.
And the best part is that the model is a self-learning one, recalibrating itself based on actual available customer behaviour data. Today, banks issue credit cards without even seeing the customer, says Ramakrishna Reddy, vice president, Fractal Analytics.
In the past such analysis wouldn't have been possible as the cost of computing was prohibitive. That has changed as the cost of technology has dropped. Around ten years ago you needed a supercomputer. Now you can do the same work on your desktop, says Reddy.
Also, as Indian companies globalise and grow in size, it's becoming increasingly difficult for them to handle such large numbers of customers based on the old rules. Says Pankaj Chandra, professor of operations and technology management at IIM Ahmedabad: As the size of the business grows you realise that you can't do things on the back of an envelope.
The number of companies lining up to invest in analytics is only increasing as more players realise the benefits. Meanwhile, the experiences of the first movers can serve as a useful guide for those taking the plunge.
Making it work
Getting on to the analytics bandwagon is the easy part. Knowing what to do next is the difficult step. As with its predecessor, CRM, data analytics also tends to be sold as a IT package that promises to work wonders at the click of a few keys. That could be the worst way to go about it. Madhabi Puri Buch, head-customer service delivery & operations, ICICI Bank, has a word of advice - start small and keep testing and scaling up. Otherwise it could just turn out to be wasted expenditure: It'll be like learning to drive for the first time on a Ferrari, instead of a Maruti 800, she says.
Ajay Kelkar, head-marketing, HDFC Bank, recalls using simple excel sheet analysis for a year before he could get buy-in from everyone: You need to demystify analytics, by demonstrating performance.
It's also important to keep testing the system to ensure robustness of the results. ICICI Bank, for instance, carries out back testing of the credit scoring model, where past data is subjected to simulated conditions and compared with actual results that have taken place. In fact, this is an exercise most banks do this an annual basis, but Buch says ICICI Bank does it every six months due to the availability of adequate data points. We effectively have 10-12 years of learning. You need a test and control mindset, says Lalit Wangikar, vice president, Inductis, a firm that offers analytics services.
The most common application of analytics is in customer churn management where companies are able to predict, which customers are more likely to defect, and design retention strategies accordingly. This is used mostly by credit card issuers and telecom service companies. However, analytics can be as effective in areas like HR and supply chain optimisation.
For example, Infosys uses sophisticated mathematical tools to manage critical HR areas such as manpower planning, talent sourcing and retention and compensation. Compensation reviews are based on algorithms that take into account the market salary ranges and cost of living of relevant geographies and trends, the business objectives and internal data like present pay, role and performance to arrive at a competitive salary for employees across different levels, both offshore and onsite.
It has a central Systems Information Group, which works in collaboration with HR and other functions like corporate planning, on analytics. Our advantage is the predictability of our business model. We set all our targets by the models we have, says Bikramjit Maitra, head HR, Infosys.
While the rigour of analytics lends itself to high-end applications, some, like ICICI Bank, are even using it to drive efficiencies in other smaller areas such as call centre rostering and ATM currency replenishment. One of the most interesting applications has been at the branch level, where it was facing a service quality issue on account of long queues and waiting time. It implemented a Dynamic Queue Management System that dynamically allocates customers and branch staff depending on the nature of transaction. As a result, average wait times and service quality levels are up.
A matter of culture
The transition from an environment where gut-feel and intuition are mostly used to arrive at a conclusion to one of fact-based decision-making can be a difficult one for any traditional company to make. HDFC Bank's Ajay Kelkar agrees, Analytics is really about the culture. It's about seeking and demanding evidence before you take a decision, he says. Marketing, however, tends to be a right-brained activity. This means changing the rules of the game for the marketing and sales teams and encouraging them to think not at a portfolio level but at the individual customer level. It's about marketing to a customer segment of one, says Wangikar. Effecting this shift in attitude can often be time-consuming and expensive. In fact, Kelkar says change management efforts for making analytics work could entail up to two to three times the cost of technology.
The other issue is getting people across teams to share data. Transparency and sharing is not something everyone likes to do, says Allan Russell, senior VP strategy, SAS International, a leading provider of business intelligence software. Therefore we like to get involved in missionary work. It's also useful to understand that analytics isn't just about technology - it's ultimately about the business. The key, says Reddy, is in understanding the domain.
It also requires a strong partnership between IT and the respective functional teams to ensure relevant data is made available. Most companies have established core analytics teams that are driving this initiative. ICICI Bank has 30 people across two teams - one on the credit and marketing side, and another on the customer benchmarking and operations side. This team consists of statisticians, MBAs, CAs and engineers, among others. HDFC Bank has a Marketing Information Group of 12 people and this is embedded within the marketing group.
Since most decisions are quantified, a positive fallout of an analytics-driven culture is that it helps set targets and motivate people better. At ICICI Bank, for instance, comparison data from the dynamic queue management system is sent out every month to all branch heads, and this motivates them to push their branch performance levels higher. Team spirit and accountability at branches has shot up, says Buch. There are measurable cost savings, as well. According to Buch, analytics based call-centre rostering could save up to 30% on headcount costs, and every additional conversion through cross-sell could result in savings of up to Rs 500 in cost of acquisition.
Not every company, though, can be considered ripe for analytics. Size does matter and so does the stage of maturity. For the big global Indian companies, topline growth is a priority and they are the right candidates to prepare for analytics. Fractal's Reddy explains that once this rapid growth plateaus and operations have stabilised, analytics will be able to provide a true competitive edge. Analytics is driven by business cycles and the state of the market, he says.
However, many companies make the mistake of investing too early in analytics, and then expecting quick results. It's a very time-consuming process, says Harsh Vardhan, director, Kandor Solutions, who advises use of the right tools for building decision rules, and bringing in experts with vast domain knowledge and experience in the real world. Software can do very little without this winning combination. Wangikar adds that this industry needs three kinds of people - those who understand business problems, like MBAs, those who're pure number crunchers, and those who're strong on statistical modelling and research.
Moving to the next level
Globally, analytics is being used by leading companies such as P&G, Wal-Mart and Amazon as a tool of competitive advantage. However, according to Thomas Davenport, professor and director of research at Babson College, most companies haven't yet fully leveraged the benefits it can bring. And tech companies are partly to blame.
Till now (analytics) was mostly query and reporting, admits SAS's Russell. Now there's a focus that helps understand the business rather than automate mundane tasks.
And though a lot of the data storage is being driven by regulatory requirements, Russell says those companies need to move beyond and start looking for more positive trends that can be uncovered. For instance European firms are already using analytics to justify their big IT expenditures.
Companies also need to move from a simple 'Yes-No' answer to a 'how much' type of decision that incorporates a variety of other dimensions from customer data. The next big step is online analytics, which banks like ICICI are already doing. This means that if a customer whose purchases are exceeding his credit limit swipes his card at an outlet, the system crunches through the customer's historical data to arrive at whether to give the extra credit, right on the spot. Analytics has to be driven online, in real time. Information is time-sensitive, says Vardhan. Kandor Solutions, for example, is developing analytic software customised for the Indian market for the first time, and Vardhan says his firm is also looking at service analytics, where integrating information across channels helps get a single view of the customer.
(The Economic Times (India) Via Thomson Dialog NewsEdge)