Narrative Science, a Chicago-based technology company, transforms data into stories and insights through its’ proprietary artificial intelligence authoring system. Narrative Science was recently among the top ten winners of the prestigious Chicago Innovation Awards, for which there were over 500 applicants. Narrative Science was founded by Stuart Frankel, Kristian Hammond and Larry Birnbaum. Professors Kris Hammond and Larry Birnbaum are from the Department of Electrical Engineering and Computer Studies at the McCormick School of Engineering and Applied Science and developed an early version of the Company’s core technology at Northwestern.
INVO interviewed Stuart Frankel, the company’s CEO. Mr. Frankel, an experienced technology executive was a member of the senior management team at DoubleClick and is the former CEO of Performics, a performance marketing services agency that is now owned by Publicis. Mr. Frankel is a lawyer and a CPA.
What is Narrative Science? How did it get started?
Narrative Science is a technology company focused on the automated creation of narrative content. Our technology generates news stories, business reports, tweets, texts, snippets and other kinds of text content purely from the analysis of data. We work with media publishers and businesses in a variety of industries. The product can be used to create editorial content across a variety of media verticals including sports, finance, real estate, and politics. In addition, our system can be used by businesses to create narrative reports that communicate information and insights gleaned from large data sources such as sales, and marketing data, or operations data and market research
The genesis for the idea began when my two co-founders, Kris Hammond and Larry Birnbaum, taught a course at the Medill School of Journalism, Media and Integrated Marketing Communications. Kris and Larry worked with one of the project teams in the class to develop a prototype of a technology that automatically generated editorial content from data. The project team focused on creating baseball-related content and one of the first stories generated by the technology was a story about a Northwestern Wildcats baseball game. The story was generated entirely from the game’s box score and play-by-play information.
How did you first learn about the work of Kris and Larry?
John Lavine, the Dean of Medill introduced me to Kris and Larry and they exposed me to their research lab that they ran at McCormick. Following that introduction, I met regularly with Kris and Larry to help them evaluate some of their research projects for commercial viability.
What compelled you to continue to spend time evaluating the core technology?
I was really blown away when I first saw a demo of a prototype of the early technology (initially called Stats Monkey). In particular, I was struck by the quality of the initial output -- it read shockingly well. I was also impressed with Kris and Larry as well as the two Medill students, Nick Allen and John Templon, who were on the original project team. Following some initial discussions with Kris and Larry, I spent several months doing diligence around the idea of building a company from the technology. Based on that work, it seemed pretty clear to me that if the technology could be developed beyond the prototype, there would be a very real opportunity to build a sizable business.
Can you share some important milestones?
The first significant steps were incorporating the company and negotiating and executing a license for the technology with Northwestern, both of which occurred in the early part of 2010. During April 2010, we raised a little more than $1M from angel investors. During this time, it became clear to us that the original application needed to be rewritten in order to create a horizontal platform that could be in used to write about any subject matter using just about any kind of data. This process took about six months and was a key inflection point for us because it really allowed the company to scale.
Another milestone was raising our first institutional round of $6M from Battery Ventures in January 2011.
We ended 2011 with about 25 customers and 24 employees, who are split between Chicago and New York.
Can you speak to the difficulty in transitioning the core application from what was initially research into something more commercial?
Like many technologies developed within a university, the initial version of our technology was not developed (and with good reason) with an eye towards things like scalability, reliability and security. When we decided to rewrite the core technology, we really made a significant investment in the future of the company, but this investment allowed us to move from a developmental stage to a commercial stage. I consider us fortunate in that it has primarily been customer demand that has required us to move rather quickly into an execution mode. We do not have time to think about the abstract very much. Every day it seems like we have more employees, customers and projects – we have to get things done.
Can you describe the company’s value proposition?
We create two broad types of content – media content and business reporting, although we use the same technology platform for each segment. For media publishers, our technology allows publishers to expand current content areas or enter new areas of coverage at a very reasonable cost. For instance, earlier this year, we partnered with GameChanger, a mobile scorekeeping application with a statistics management website for youth, high school and college baseball and softball. We have integrated our technology with GameChanger and now generate a story for every game scored through the GameChanger application. Our technology wrote 300,000 stories about youth baseball in 2011.
In terms of business reporting, we’re essentially helping businesses deal with the problem of data overload. We’re finding that companies across a wide range of industries are inundated with too much data and would benefit tremendously from tools and applications that help them understand and communicate insights from data in a more efficient and understandable manner. The narrative form not only makes data more understandable but also more consumable. For example, we are helping some advertising agencies and advertising technology companies present online advertising campaign data to their customers and employees in a simple, easy to understand narrative format. This allows these companies to leverage the investment that they have made in data and to communicate much more effectively with their customers.