Mar 6, 2020

The Next Big Thing in Data According to Dativa’s Michael Collette


If there is anyone who has built his career around data, it would have to be Michael Collette, CEO of Dativa, a fast-growing data science services start-up providing custom TV measurement strategy and solutions to brands, agencies and media companies.

Collette is a serial entrepreneur with four start-ups, including Dativa, dedicated to maximizing the value of data for use in media. “Most of my career was in technology – set top box software, the early days of digital satellite and multi-room DVR,” he explained. His professional crossroad came as a result of a shift in thinking about interactive television.

Charlene Weisler: How did you make the pivot from interactive TV into data?

Michael Collette: About 7 or 8 years ago, I was consulting with Canoe trying to figure out how to get interactive television working once again on set top boxes when I met with people at Turner who basically said, ‘eh, it’s never going to happen, but let us show you this thing.’ I was introduced to Zeev Neumeier (currently SVP of Product, Inscape) who was the technical founder of what was then TV Interactive Systems which became Cognitive Networks. That experience really started a significant pivot for me. We thought we were building an interactive television company, but our biggest investor said, ‘You’re not. You’re building a data company and that is why I am investing.’ I walked out of that meeting and turned to Zeev and said, ‘Do you know anything about data?’ He said, ‘Nothing.’ I said, ‘Me neither! Let’s go!’ (laughs) It was very much like that. 

Weisler: Which led to Cognitive. (Disclosure: I consulted with Cognitive from 2013-2015)

Collette: Yes, we started to scale the company about 7 years ago. We did a bunch of really fun interactive TV stuff with LG as a customer but we quickly found that our investor was right: the most interesting and valuable output of the ACR system was the data. As the dataset began to scale, and as we started to get Vizio launched alongside LG, we saw some profound things happening. So I spent several years focused on making great data.

Weisler: I know you have a specific philosophy regarding data.

Collette: Yes. We took a different approach compared to other companies in the space. They were taking the data and creating services with it. We thought the more important thing to do was to make great data and to build trust in the data by being really transparent as to how was it made and what it was like. Data is complex.  It has to be transparent so people can use it properly. There are always pros and cons.  Raw data is rough.  We also wanted to empower an ecosystem so we partnered with lots of folks to let them chase different aspects of the use of that data. That is where the iSpot relationship came in, along with VideoAmp and Alphonso.  They came out of that early effort. The strategy has proven out well in that Vizio acquired Cognitive and created Inscape. Inscape has flourished with its focus on making the data and being transparent about how it’s made. 

Weisler: So what is the next big thing in data?

Collette: From my work with Inscape and my partnership with Tom Weiss, Dativa co-founder and CTO, it looked like all the emerging companies in the data-driven TV space were pivoting into each other - all around measurement.   It looked a lot like the ad tech train wreck with an oversupply of similar offerings.  We believed that a lot of companies were going to raise a lot of money. A few of them would get good access and a lot of them would struggle.  Our fundamental observation was that eventually what will happen is, as the market gets smarter about granular data, the endpoints in the markets - the brands, agencies and networks - are going to start to bring the data in-house. Some will bring that talent in but, at the end of the day, there is a real need for a professional services company to help those companies implement their data strategies. This is where Dativa comes in. It has been a professional services company from the start. We are not here to sell a platform. We operate as a service to help our customers build a platform that they want to use. We will help them extend it as they wish but they own it. We just build what our customers want, implementing the data they are licensing and not licensing to them what we have.

Weisler: Any company working in the data space has to be concerned about privacy.

Collette: This underscores why insourcing is so important. When our customer brings Inscape data into their own domain, for example, in that setting they can bring whatever other first or third data they have. So, if you’re a brand and you have customer data, you can match the consumption data you have to the viewing data privately without giving up possession of your data. In today’s privacy world, that is really important. You haven’t shared your customer’s data with anyone. You’re just using it for your own insights. From a privacy standpoint, that use of data is much, much more permissible. 

Weisler: How do you think this will play out in the media ecosystem?

Collette: Information is power. When an organization that never had access to fundamental data starts to have access to it, that organization starts to change.  Five years ago, we knew that this data was going to be transformational. But five years ago almost no organization had the skills to use it. That was a real limiter to growth which was why the third-party ecosystem players were necessary because you needed a skilled team between the data and the user to put it to work. That has changed. Now what we are seeing is that the media organizations are starting to build up their own data mining muscles. They are building inventory optimization and forecasting, which is a tricky data science undertaking. That kind of activity is growing and becoming more important. And each organization does it differently – they have different inputs, concerns and specific questions. For that reason, implementing data in-house is going to be increasingly important as it begins to make granular data close to currency class.

This article first appeared in Mediapost.com
  

No comments:

Post a Comment