Data team, not an individual, drives Scripps Networks
Conference Blog | 27 March 2015

Data science.
That’s an admittedly broad, almost nebulous, term, but a few of its key components are that it encompasses both business understanding and analytical capabilities, including statistical expertise, data architecture, coding, and development skills.
It’s hard to find all of the above skills in one person. It may be easier to find a unicorn.
But news media companies are in luck: Laura Evans, vice president of audience development and data science at Scripps Networks, assured delegates at INMA’s Big Data for Media conference at Google London on Friday that there are lots of benefits in using a team.
The easiest way to make this clear is with an example: Site recommendation engines, for instance, are useful, but, Evans said, “they can take up a lot of a site’s real estate.” Moreover, organisations don’t always know how effective the engine is at warding off missed-clicks, and, according to Evans, “what people click on a site is just as important as what they don’t click.”
Here are six steps that Scripps Networks used to surmount this particular obstacle:
- Understand and be clear about what the company wanted to achieve.
- Create a data warehouse and extraction options for data processing.
- Review and analyse the data collected in order to determine what key metrics and associations drove the desired behaviour.
- Create an algorithm that reflected the optimal content path.
- Built out an algorithm so it could be applied to user experiences.
- Iterate base-level formulas to outpace the current vendor performance.
If that’s too much to remember, then keep in mind these three things: “Capture and collection; report and analyse; and use and grow,” Evans said. Reorganising Scripps Networks’ strategy around the data development lifecycle allowed the company to sidestep challenges and get on the path to success. Specifically, it clarified what works best, where, and with whom, which unearthed how to make site recommendation engines both more effective and more personal to users.
Those replicating this process may learn a few more things along the way, such as patience and the importance of actually starting at the beginning in order to reach the end (no skipping steps!).
The takeaway? “I don’t have a data scientist,” Evans said. “I have a data organisation.” Each step of the process allowed Scripps Networks to tap into individuals’ areas of expertise, which could range from statistical modeling to front-end development and anything in between.
Don’t wait around for mythical, omniscient “data scientists.” They’re probably not coming.