Online video providers are awash in data these days. Sometimes we should listen to what it has to say, and sometimes it might lead us astray.
At NAB 2017, I moderated a discussion entitled When Video Is King, Data Rules: How Analytics Puts Companies Ahead of Competitors. I asked the panelists to give an example of where data helped in an unexpected way, and where they expected it to help and it didn’t. Their answers reveal how tricky it can be interpreting the data.
Data relationships may not be obvious, but they’re there
Michelle Boockoff-Baidek, VP of Global Marketing at The Weather Company, points out that no matter how strange the story told by the data, you should trust it:
“Data doesn’t lie. There are these really unintuitive correlations that you can make between weather and behavior. For example, we know that juice sales spike when it’s windy and we have lower than average temperatures.”
Soumya Sriraman, President of the new BBC/ITV US SVOD service BritBox, agreed that the data relationship may not be obvious, but is powerful once revealed:
“The thing that’s been the most useful is attribution data. Someone’s seen your information in the New York Times. They come back 3 days later and you’ve generated a subscription from a completely different platform.”
Shira Lazar, Co-Founder/Host, What’s Trending, also says the cause and effect relationship may not be obvious at first, but digging deeper into the data will reveal it:
“Sometimes I look at our stats for the week and it’s a video I’d forgotten about. It’s because we got connected to another video that’s trending. If that video is getting hundreds of thousands of views, and you’re the first one connected to it you’re going to draft off that one.”
There’s gold in that back catalog!
Dave Mowrey, Head of Product – Watson Media at IBM, says that Watson is mining “dark data” to deliver revenue gold. The technology is helping marry library content with trending media:
“It’s really interesting to be able to pull monetization where you didn’t have it before. 80% of data is unstructured and dark. Being able to pull that out <of the library content> and monetize it is really powerful.”
Social data disappointing, and contradictory
Ms. Sriraman pointed out that social media is turning out to not be quite as helpful as hoped:
“Social data hasn’t been that it was made out to be. We’re using social triggers but it’s just not doing what it’s supposed to.”
Ms. Boockoff-Baidek went further in the criticism of social media:
“We’re on social and everyone is consuming data, but the monetization is really a struggle.”
Ms. Lazar says that sometimes social can deliver a viral a hit, despite the publisher ignoring all the common-sense data about the optimal time to release a video:
“We had a video that got 14M viewers. It was Elmo Getting Fired, based on the Trump cuts. It was released in our off hours of publishing, on a Friday at 5PM Pacific. We went against all the rules and yet it somehow worked. That’s one of those viral moments. How do we tap into that again? It’s hard to know.”
When bad data reveals something good
Mr. Mowrey points out that bad data can also point out something good.
“You think that when your quality of a streaming is poor, you’re buffering a lot. Obviously, there’s going to be drop off. We’ve found the more the viewer goes back and tries to watch the content, (the more it) gives you a good indication about how engaged they are with it.”
Why it matters
Data is revolutionizing the business of video.
Sometimes following it can deliver big benefits for the service.
Other times, ignoring it can be the best thing you can do.