The best part of running this blog has been, connecting with some of the best people in the Analytics Industry. I recently, got an opportunity to interview Srikanth Velamakanni, Co-founder and CEO of Fractal Analytics. The interaction was, undoubtedly, a great learning experience; something which I have tried to extend to all our readers, through this post.
Srikanth holds an MBA degree from Indian Institute of Management Ahmedabad and a bachelor’s degree in Electrical Engineering from Indian Institute of Technology, Delhi. Before co-founding Fractal in the year 2000, Srikanth was an Investment banker and had worked with ANZ and ICICI Bank, managing asset & mortgage based securities. Srikanth is passionate about learning and reading, especially about behavioral economics, positive psychology, and neuroscience. He loves listening to Audiobooks and is an analytics and technology evangelist.
In about an hour, which I spent talking to him, I developed deep respect for him as a professional and as a person. I hope that someday I would be able to make a similar impact in the Analytics Industry.
I plan to release the interview in three modules in the coming days. The structure for the same, would be as follows:
- Part 1 – Starting Fractal, initial challenges and how they were overcome.
- Part 2 – Building an Organization. Hiring, training, trusting your employees to build a world class Organization and dealing with attrition.
- Part 3 – Future bets for Fractal and evolution of Industry. His advice to people entering the industry.
Here is the first part of the interview:
KJ: You started Fractal back in the year 2000. At that time, data science and analytics were not as popular as they are today. You were one of the earliest start-ups in the industry. So, what was the idea when you started? How did it happen? How did you decide that this is the idea you would want to pursue full time in the future?
SV: Sure. When we started, we obviously didn’t have analytics in our mind. We (founders of Fractal) had quit our jobs and wanted to do something together. We actually did a little bit of search and thought that we should be in the IT space. However, after talking to people (senior veterans) in the industry, we realized that until and unless you are focussed (on a particular aspect), it is not a good idea to be in the IT space.
We were doing soul searching and looked at various things. After a while we realized that we are good at Financial Services and we are good at Maths. We understand probabilities & statistics and had worked on problems like hand-writing recognition and face recognition during college.
While talking to a few people in ICICI, we learn’t that they were facing challenges in identifying the credit worthiness of customers. This got us very interested, as we knew, credit risk could be identified with a lot of statistical tools. So it turned out, that we were actually the first ones to build a statistical scorecard (for identifying credit risk), in India. This score card, as mentioned earlier, was for ICICI Bank (to predict default on their personal loans). We were pretty excited because this could enable them to do real time risk scoring.
When we finished this project, we realized that this was something exciting.
To think of it, what do you need in life? You need three things:
- Pleasure – Whatever you are working on should be fun
- Meaning – It should add meaning and value to the world.
- Strength – It should be an area of strength
So it felt like this industry was meant for us.
What we did not know was how big this area was? Whether we could make money in this area or not? To be very honest, we were MBAs from IIM Ahmehdabad, but we did not do any major business planning in terms of how big this space was going to become.
The next thing we did, was to build a default prediction model for corporates using information regarding stock prices. We could simulate stock prices and see default ratios under various circumstances and could come out with pretty good models predicting when a company could default. In fact, we came out with models, which were better than Moody’s and S&P. Moody’s and S&P were both backward looking where as we were using predictive modelling to come out with default probability.
Another thing which was fascinating at that time, was the entire area of consumer behaviour. How do people change their behaviour? This was 2001 and it was recession time. We observed that consumer behaviour was changing dramatically in a few sectors. Specifically, when we started working with Hindustan Lever, their Chief Economist told us, “Here is all the panel data I have. Tell me what I am not looking at.” We looked at various aspects; we studied whether customers were changing products or brands they used to purchase before? And how were the purchasing patterns shifting?
Then we realized that this is it, this is the space we had to be in.
We did not have any role model. But we felt this was it was an exciting space to be in, and all the three elements (Pleasure, Meaning and Strength) looked to be in place, and we just kept going. We have never looked back since.
KJ: That almost brings me to the next question. This was a new area when you started. You were treading the path for the first time for yourself. When you start in that kind of scenario, there are challenges that you come across. The challenges could be, proving to customers that Analytics can or has added value. So, what type of challenges did you come across and how did you tackle them?
SV: Great question! Infact we did face a lot of significant challenges in our initial days. Right at the start while working with ICICI bank, we had to prove ourselves. We did not have any experience in building models. So we had to prove to them, that we know what we are talking about. Remember, this was about building a risk model. If you built a wrong risk model, it could cost the bank a fortune. It was not easy. ICICI bank asked us, “Why should we believe in you and not work with Fair Issac in the U.S.?” So we said that we will do some original research on the topic and come back to you with what we think about the space. We spent a couple of months writing a research paper on the subject. We took that to ICICI and told them that this is the state of the art research on the subject matter. They really liked what we had put together for them. This was the way we could earn their trust, so that they could share their data and we could build the first statistical scorecard.
Another instance was a conversation with Citibank. We met them in Mumbai and then in Chennai. They said that “We have massive team of Ph.Ds. sitting in New York, who know how to build all this stuff. You guys are 20 something and you have got no experience in the domain. So how can you beat what we are doing?”
So this time, we said “We understand that you might have a great team. We also believe in what we are doing. Why don’t we set up a challenge: You give us the data, we will not charge you anything. We will build the model and see if this adds value to what you are doing.” We have to give credit to them that they accepted the challenge and gave us all the data. We built a personal loan cross-sell model for Credit card customers and then also looked at the price sensitivity of the customers in the model. That was a new thing at that time. They had not done a lot of experimentation on various price points. So we had thin data to build the model.
Once we built this model, these models performed extremely well and all of us were surprised how accurate they turned out to be. At that time, they came back and said “You guys have clearly beaten what we were doing internally.” So that brought a lot of credibility to us. With clients like Citibank, ICICI bank and Hindustan Lever, we could now go and pitch confidently to other people about our services.
KJ: Interesting! This idea about pitching yourself against some of the best people in the industry sounds fascinating!
SV: In fact, we did a similar exercise with Capital One in the US. They challenged us whether we could predict when a customer would attrite (rather than just who would attrite)? We put together a survival analysis framework in terms of when would the customer attrite and what intervention could be made to retain a customer. This framework also provided answers like whether an intervention was effective or not and by how long would it increase the customer life-cycle. This framework was very fascinating and they loved what we had put together.
These are the kind of projects we have worked on. We have done a lot of stuff. We did not focus on making a lot of money. We focussed on doing a lot of cool stuff and focussed on building our credibility and expertise.
In next part of the interview, we will discuss Fractal’s hiring and training policy and how do they tackle employee attrition, which is one of the biggest challenges in the Indian Analytics industry today. For me personally, the key take-aways were the three elements which Srikanth focused on and some of the non-traditional pitches they made in the beginning of their journey. Stay tuned with us for more of this exciting and insightful conversation with Srikanth, which will be released in the next 2 parts.