How to Use Data to Drive Online Video Success

streamingmedia.com
From content creation to subscriber retention, video publishers are embracing Big Data in a big way. These are the most important trends and challenges behind the video data revolution.
Ask any digital video publisher what its data strategy is, and you'll likely get an answer ranging from, "We're a highly tuned data-driven organization," to, "Well, we collect a lot of it, but don't really know what to do with it." Without solid analysis and interpretation, data is useless. So what kind of data is most valuable to streaming publishers, and how can they gain insights and generate outcomes from the data they collect?

Audience Profiling

Our first stop on the data trail uses data to identify what content the audience wants. Nonprofit broadcaster NPR collects data to deepen the relationships between stations, sponsors, and listeners. NPR comprises member radio stations all around the country, and it delivers personalized content through the NPR One app, which promises listeners, "Public radio made personal."
How do you make content personal? "The location of the end user is of great importance because it helps us to target not only which member station they're affiliated with, but also localized content that we would push to them," says Michael Dube, head of streaming media at NPR.
NPR developed the app and made it available for local stations to give them an affordable platform for distribution. Listeners from all over can tune in to hear locally produced content tailored to their interests. The result is data-driven content curated via a combination of machine learning and human assistance.
For deeper recommendations, listeners have the option to identify themselves and have their likes and dislikes tracked. "We do a tremendous amount of A/B testing around our media delivery," Dube says. This testing has been a way for NPR to design features and content, and then test to see what resonates. For example, if a user chooses not to listen to politics, then the app won't deliver political news to them, although it may present a piece that has a wider subject matter and includes political content.
"We're creating personalized and dynamic experiences," says Dube. Content is commonly 2–5 minutes long, with longer shows occasionally divided into short clips. The average unique session on the NPR One app is 40 minutes, which suggests its curation formula works well.
NPR stations raise funds from local listeners as well as sponsorships placed within the programming, and NPR's data helps them attract more sponsors. "Big Data is very important, because the more people are consuming our content, the more opportunities we have to include our sponsors within those deliveries," says Dube. The organization is selective about the types of sponsors it works with, and it uses its own ad network to perform dynamic sponsorship nsertion into content driven by audience data. Accurate analysis of audience data is vital to keeping media companies on the right side of their audience, since irrelevant sponsorships can send listeners somewhere else.

Subscription Data

Our next data point comes from the video platform's perspective. Vimeo's VHX subscription video-on-demand (SVOD) platform provides media companies with the technology they need to sell and serve subscription content. Vimeo's media customers use the VHX dashboard to see information about viewer conversion rate and location, along with which platform is driving the most traffic. This data helps their media customers create stronger businesses.
"One of the biggest things we've been rolling out this year is our branded apps on every device," says Casey Pugh, co-founder and head of product at VHX. "That data [collected via app use] helps sellers place their marketing dollars in the right place to reach customers." VHX's top SVOD seller is Black&Sexy TV, which bills itself as "the leading network for young and progressive black people since 2008." "It has some of the best conversion on the entire platform. It has niche comedy and drama series which you don't typically find on any other platforms," according to Pugh.

Vimeo’s VHX SVOD platform offers a dashboard that lets publishers analyze their data based on products, subscribers, and—shown here—traffic.
Data is nothing more than useless numbers and meaningless statistics until you gain insight about what works. Black&Sexy TV found that push notifications and releasing content over the weekend both drove increased engagement. The media company is also very active on social media, constantly engaging with its audience.
Another optimization technique VHX publishers use is offering promo codes for discounts to specific customers. "We released this new tool that allows sellers to create a promotional code to give out to the customer to get, for example, 30% off their subscription," says Pugh. "It's a marketing sales tool to drive more subscriptions by creating a lower barrier to entry and increase conversion. Black&Sexy TV has rolled out the promo feature, which has shown significant increase in subscribers already."
While Black&Sexy TV has identified the important items that drive their engagement, not all publishers are as focused on all of the distribution details as they should be. Pugh says VHX provides automated and human advice to all its publishers based on evaluation of customer data. "One [piece of advice] is a simple thing," Pugh says. "For SVOD, you need to upload content all the time. Some people forget to do it, and they are slowly churning customers." It seems obvious, but analytics can drive decisions about putting up new content, finding the right time to publish the content, and making sure viewers know about new offerings.

Optimizing Strategy

While Vimeo is ensuring its publishers have the building blocks to create strong businesses, Ooyala is providing strategy optimization to its customers through its data insights. "I come from the world of TV. When we had a new show come out we had to wait till 9:00 the next morning to hear how viewers had reacted to that show," says Steve Langdon, global director of the strategic media consulting group at Ooyala. "Now, publishers can see all sorts of details and immediately make changes to their video strategy." Langdon outlined a few examples where data has allowed publishers to uncover valuable information to optimize their businesses.
In the first scenario, a publisher gained insight by examining its licensing and user interface (UI) structure. The customer was a property delivering kids' content. "Their viewing data showed episode one of each of their TV shows was getting about 90% of the overall views, and episodes two through 10 were getting the other 10%. That went completely against our standard analysis of SVOD," says Langdon. "The reason for that is children enjoy the repetition." The insight meant the publisher could ask itself if it actually needed to buy a whole series or just the first episode of about 40 different series.
In the next scenario, Arsenal Media Group, the media property associated with the popular English football club, wanted to find a way to gain longer engagement periods and reduce what it calls the one-video bounce—viewers who watch a piece of content and then leave. Analysis of traffic showed that viewers started dropping off around 60% of the way through a program, so Arsenal decided to run a thumbnail saying which video was coming up next. "You would never see an episode of Downton Abbey finish on a commercial network and then the screen goes to black," says Langdon.
The last scenario shows the importance of context. In this case, Ooyala had two newspaper customers. At one, people were only logging on from 5:00 to 9:00 a.m. At the other property, viewers were checking in at various points during the day, with the peak between 5:00 to 7:00 p.m. The overall traffic numbers showed one peak period, but the exact content viewing time showed peaks and valleys. This key differentiator provided guidance for tweaks to programming strategy for each publisher.
"The result is the first publication can go about informing its viewers that there are lots more videos to be watched in the afternoon and the evening," says Langdon. "For the other company, it's about how do you maximize what's happening on the commute to promote content. That one single data point changes the way we speak to both of those companies."

Dailymotion uses data from third-party providers such as Tubular and Tableau to determine what content might do well on its platform.

Analyzing Trends

Our next stop on the data trail looks at how data is used for licensing rights. "Because we are an acquisition and business development group, we use data to identify what content creators we want to reach out to," says Anthony Layser, content director at Dailymotion. The ad-supported user-generated content (UGC) platform uses tools from two companies, Tubular and Tableau, to help it make sense of its data. Tubular posts a daily report of what is trending on about 30 social media sites.
Merging Tubular's information with internal data allows Layser to build projected revenue models. "We can get a rough idea about what potential revenue might look like if we were able to acquire that partner's content on our platform," says Layser. "If there's a content manager who is managing sports, they can see what sports-related accounts [on Tubular] were popular last week from a traffic standpoint."
If you need to analyze data for thousands of content partners, finding good tools becomes key to your business. Layser uses Tableau to help see and understand trends. "When you have a partnership you're handling that has many dozens of channels, you need some centralized location where you can categorize and manipulate data to get a clear understanding about how the partnership is performing," says Layser. "For example, I have over 800 Fullscreen accounts on the platform. If I want to see how all those accounts are performing collectively from a revenue standpoint or from a traffic standpoint, Tableau allows us to easily arrange that data."
On the partner side, Dailymotion offers two types of measurement tools. "The traffic dashboard is a lot of view data—minutes watched, geography, what the audience retention is, percent of video watched, domain breakdown [for embedded content], and traffic sources," says Layser. "The revenue dashboard is a day-by-day breakdown of impressions, average CPM [cost per thousand], and estimated revenue per day."
Partners with content on multiple platforms such as Dailymotion, YouTube, and go90 need to be able to grab the data from their partners and crunch the numbers themselves, says Layser. "It's really important to them to be able to  export spreadsheets from our dashboards to a CSV file and drop them into a centralized database [so they can compare how activity does on different locations], rather than having to manually see something in a dashboard and have an intern actually trying to create a spreadsheet," he says.
This brings up two very important points. First, the manual handling of data easily allows errors to be introduced. Second, standardized metrics are critical, so publishers can compare apples with apples.

Data-Driven Creative

Our next media company is going data first when it comes to designing content. Whistle Sports' focus is providing content to young millennial sport fans. "The truth is, sports happens billions of times a day in backyards, in parking lots, in playgrounds, in practice fields, sometimes even in office buildings, and that can be at least as compelling as that live sporting event if you aggregate it," says Brian Selander, EVP at Whistle Sports. Viewers are highly engaged by this content, so Whistle Sports is able to monetize it not by selling ads but by finding sponsors. This is the latest approach brands, sports leagues, and media companies are taking to combat ad backlash and ad blockers.
Competing with live content can be challenging, but Whistle Sports is creating evergreen branded entertainment. "If you're a brand or agency and the content you've created with us can be rediscovered in a month or 6 months and be as interesting to somebody, that's an investment that pays off repeatedly," says Selander. "This means it's important to figure out what's going to work before you make it. Sports that really engages young millennial audiences needs to be research and insight first, content second."
So far, Whistle Sports' insight has proved successful. Since its launch in January 2014, it has attracted 190 million followers in the 14–32 age range across all social platforms, and it attracts 2 million new followers a week. The company has approximately 400 content partners, such as Dude Perfect, a group of entertaining guys broadcasting a reality show, sporting event, and long-form commercial all rolled into one.
"I'll often say at conferences, we need to liberate brands and agencies from the tyranny of the view count," says Selander. "I think particularly in the digital space, people remain hung up on how many views something gets and are less concerned about the more important number, which is how engaging this content is and how often people are actively sharing it.
"Before we bring any creator into the environment we take a hard look at the data around their engagement, their viewability, and their scalability. We are data-driven from the beginning of any decisions," says Selander. "We get hundreds of millions of data points a month from interactions between creators, content, and fans that we analyze to see what works best."
Whistle Sports looks at data points like what time of day content is viewed, what brand integration is connecting most strongly, and what words in the video description may make viewers more likely to engage with content. All stories are crafted based on the insight Whistle Sports gains from its persistent focus on translating data into valuable analytics. "We may follow a particular video minute-by-minute after it launches, and we can also have insight across the network on an hour-by-hour basis," says Selander. "Then at the end of the month we can compile that to see if there are any emerging trends we might not have seen if we were too in the forest and not standing outside of it. Media has always been data-driven; it's just the immediacy and levels on which that feedback happens are now far more connected."

You Might Also Like …

While Whistle Sports is experiencing great success curating its viewers' experiences, our next company is there to help when you want to watch something, but you're faced with too many choices.
Most of the major TV Everywhere (TVE) providers use tools from Digitalsmiths to identify trends and build on user preferences. "The old Amazon model is, ‘This show is similar to these three things, so you might be interested,'" says Billy Purser, VP of marketing at Digitalsmiths. "We take into consideration what that person typically watches, on what type of device, at a particular time, then make sure it's blended into the recommendation." The number of data sources on which Digitalsmiths bases its analytics varies from 10 to a couple thousand pieces of information, including scheduling data, what's on now, what's available to a subscriber, third-party critics' ratings, popularity, type of content, mobile location, and device type.
How do you make sense of thousands of pieces of information? Digitalsmiths does multivariate A/B testing, which is testing 60, 70, or 80 variables to identify which UI, features, or promotional campaigns have the best reception. It's like A/B testing on steroids. A/B testing typically tests a single variable to determine the effect of a change to that variable. With multivariate testing, several variables can be tested together to uncover the best combinations.
This type of testing helps providers segment subscribers to ensure the right audience is receiving the best recommendations. "I think where the market is going is segmenting the appropriate audience from a promotional standpoint and being able to leverage data to drive more viewership, more engagement, [and] more revenue," says Purser. "The Holy Grail is to deliver a really personalized offering so all the recommendations are unique to the person or the group that's in the room."

Measurement as a Platform?

Moving from the TVE world to the other end of the spectrum brings us back to Tubular. Tubular publishes daily rankings of content on the top advertiser-supported video platforms including Facebook, YouTube, Vine, AOL, Yahoo, and Twitter. Tubular has about 120 enterprise subscribers, and it offers a free version.

Tubular publishes daily rankings of content on the top ad-supported video platforms.
"People use data for content strategy, for sales support, for media planning, for executive reporting, for identifying influencers," says Allison Stern, co-founder and VP of marketing at Tubular. "There are a lot of different ways to use that intelligence layer, but it all [goes] back to being data-driven and using data to inform your content, promotion, and distribution decisions."
Tubular's data enables companies to track their own or competitors' content and see what is trending by views and engagement metrics. "If you are Scripps and you're tracking food content, you can have a dashboard that shows you what are the trends in food, what food videos are taking off on what platforms," says Stern. "It can [answer the question], ‘How is Tasty doing?' Or, ‘How is TipHero or Tastemade or all these other food channels doing in comparison to your food channels?' You can get a feel for how engaging your content is benchmarked across the universe."

Tubular also offers enterprise subscribers insights into how videos from particular creators are performing
Ranking results gives publishers an idea of best practices they should follow to see, for instance, what production style drives higher engagements. "For food videos right now, there's a lot of stop motion that makes the recipe in 10 different steps where it's pictures of the food, not of the person making it. In news, there are videos that have a lot of text because in Facebook the sound is off."

Driving Revenue From Data

The takeaway from the experts varies according to the business model, but there are three keys to making data drive revenue: Understand what the high value questions are, monitor the data that gives you answers, and gather insight and identify how to respond. Everyone wants higher engagement rates with their content, whether they are ad-supported or subscription-based. Some media companies need their data to tell them when to deliver and market new content, while others look at their content's engagement levels to see exactly what resonates with the audience. In each example, the experts tested what they were doing and tracked whether their ideas resonated or were something they should revise.
While our data trail ends here, the world of Big Data is only just beginning to impact video. Content creators and publishers need to find ways to integrate it into their businesses. "People who are analyzing data and are making decisions based off of it have one skill set, but that's not the skill set of a typical video creator or producer at a major media company," says Stern. "To me, that seems like a big challenge that people are facing. It's a misalignment of skills."
Does this mean video producers now need to become more adept in analytics? Will we see data analysts more involved in production? The solution may vary from publisher to publisher, but few other types of businesses have the ability to have an intimate conversation with each and every customer and then meticulously comb through the conversation to see how to find incremental value and improve the bottom line.