Sramana: Who were your early adopter customers and what is it that made them converge to you? How did you clinch those deals?
Suresh Shankar: Essentially we are building a choice engine in a way that gives you multiple ways to apply it to your needs. In one case we have been working with a very large hotel company in the UK. They really don’t understand what type of customer comes and makes a booking at their hotel. They don’t know what the customers purpose is or what kind of influence the customer has. Now, every time the booking happens at the hotel we are able to go out into the larger world and try and understand that customer through larger social media datasets of similar people. We are able to build a larger profile of the customer references. They can then reach out to customers and provide personalized services better even before that person walks into the hotel.
Banks have a good play here as well. If you have a private relationship with a bank there is not a lot of behavior data about that one customer at the bank, especially if they are new to the bank. There is a lot of information available about that type of customer outside of the bank. We can use that data to suggest choices to that customer out of the millions of financial instruments available.
We are also doing some work with a large global payments company that wants to expand its footprint in Asia. They know the types of people who would typically sign up for their payment processes, but they needed to find 50,000 of them in the world. How can they find those people? In the past they would try to buy lists from local providers. We are able to look at the enterprise layer to find out what kinds of customers are good for them and then match that profile against public sources of information. We were able to give that customer a list of highly refined profiles of potential customers.
Sramana: I think you have a layer of your algorithms, models, infrastructure, and frameworks, that are horizontal. It seems like you can apply that to every single vertical. I strongly recommend that if your goal is to build a product company that you should build vertical by vertical and not try to do so many of them at once. As you start developing heuristics and business processes that are domain specific you will find that it is custom to the vertical. You need to leverage that work over and over again.
Suresh Shankar: That is extremely good advice. It is good to learn from other peoples experiences. I have heard that from several different people. We believe that ultimately you are right. We also believe we are going to have to use a lot of different business partners who will build verticalized applications for their industry. This year is a learning year for us. We are building across categories and we want to understand different types of enterprise data sets. We need to look at multiple types of data to further develop out the base platform.
Sramana: If your focus is to operate at the horizontal layer and to recruit a set of partners to handle the verticals then you will likely have success. Marketing to the various verticals is a complicated task.
Srikant Sastri: This is a live debate that we are going through all the time. We are always debating the number of verticals that we should be engaged with at any one time.
Sramana: Good luck with Crayon Data. Keep me posted with how it evolves!