Bootstrapping a Big Data Company Using Services: DataSong CEO John Wallace (Part 4)

Sramana Mitra: Can you talk about that?

John Wallace: The problem now has a name. It’s not an ideal name but it has a name. It’s called marketing attribution. It’s looking at the effectiveness of marketing spend. The field closest to that would be approaches of this in Statistics in the past 20 years – by week, how much we’ve spent and see if we can sort out changes in our revenue based on changes in spend. We chartered a model like that. They just couldn’t fall in love with it. We asked them why. They said, “It doesn’t take into account which consumers have been exposed.” They had this catalog modeling background where they’re used to looking at households and whether or not to spend money or not. That was a problem we decided to address.

Sramana Mitra: What year was that?

John Wallace: That was between 2010 and 2011.

Sramana Mitra: That’s when you found the problem that helped you move from services to product.

John Wallace: Correct. Then we took a computing approach that would have been a little bit crazy to follow. We would have needed to deal with the NSA to run the kind of analysis we were doing for them without being on this current generation of Big Data.

Sramana Mitra: So, Hadoop made a difference in terms of infrastructure?

John Wallace: As an enabling technology, yes.

Sramana Mitra: Is there any other new-born technology that you use from the current stack of stuff that’s available out there?

John Wallace: We’re experimenting with a platform called H2O. You had Hadoop. People talk a lot now about Spark out of Berkley as a replacement. Then in the analytics field, there’s a package called H2O.

Sramana Mitra: This is what has got you these key customers from the retail world?

John Wallace: It’s that intersection of software and services to be able to on-board and rationalize a wide variety of data. We are intentionally going after the hardest problems to solve. The more we look at it, the bigger the problem gets and the harder it gets.

Sramana Mitra: The other thing that’s really great with the way you’re doing it is you have a lot of domain knowledge that you are building into your approach. This is like hard-core omni-channel retail solution. That has its own applicability.

John Wallace: Being bootstrapped, we’ve been able to make experiments that make sense to us. We didn’t have to buy in from someone.

Sramana Mitra: Except for customers. That’s the only thing that matters. Our philosophy in 1M/1M is entrepreneurship equals customers, revenue, and profits. Everything else is optional including financiers and investors.

John Wallace: We’ve taken up consulting, so we are giving roles as account managers to people with a background in Statistics or even retail on our team. In our target market, we send someone out who already has the domain expertise to fill that role as opposed to someone who’s more about the process and organization. These people are just deep on the problem. It has an interesting payoff from the customers.

Sramana Mitra: Who do you see in deals in terms of competitors?

John Wallace: There were three teams that worked on this problem – visual IQ out of Boston and Adometry out of Austin.

This segment is part 4 in the series : Bootstrapping a Big Data Company Using Services: DataSong CEO John Wallace
1 2 3 4 5 6 7