Winning the Data War: Accessing and Leveraging Streaming Analytics
In today’s world of media insights, there’s a tug-of-war going on between those who can access data intelligence that identifies customers’ viewing patterns, demographic details, interests, and much more. The companies that are able to gather this data use it to answer questions that every media business wants to know more about: Are viewers churning after a month? Is advertising consumption down? Should we be doing bingeable full-series releases versus scheduled/staggered content drops? Is personalization working? Do recommendations increase viewing?
These insights, however, are limited. Owned and operated (O&O) properties—a lucky few—can easily go deep into exact viewing patterns if they have the right technology to do so. But many more media companies don’t have this opportunity because they are distributing their content in another company’s free ad-supported television (FAST) or another environment.
This article explores the topic of data rights versus technology capabilities as it impacts media companies’ ability to leverage accrued data and analytics to target relevant audiences, expand their reach, and monetize their content offerings. If the intersection of data rights ownership and technology capabilities access were captured in a Venn diagram, the space in between would be very small for some media companies.
When we discuss analytics and KPIs in the media and entertainment space, what exactly are we talking about? Naveen Narayanan, head of sports and data products at Quickplay, an OTT solutions provider, says that his company captures data in four categories: acquisition, engagement, monetization, and churn.
Naveen Narayanan, head of sports and data products for Quickplay
“We’re able to collect signals from each and every stage of a customer’s lifecycle,” he notes. This includes tracking marketing campaigns, which leads to customer acquisition, free trial/paid subscription conversion, engagement measurement, upsell opportunities, and retention and churn prediction with a customer’s O&O services. “We help our customers understand the subscribers and how to act on it by personalizing the content and providing recommendations and a personalized UI/UX experience,” Narayanan says.
Damian Pelliccione, CEO and co-founder of Revry, says that the company defines KPIs as keeping people inspired. The LGBT+-first content distributor operates on 58 platforms, and it delivers to 85%–90% of the LGBT+ population in the U.S. and is expanding market share worldwide. As a distributor, Revry is in the fortunate position of selling 90% of the advertising spots within its content on all FAST platforms. The major KPI, Pelliccione says, “for all of the departments within Revry is watch time. The analogy here is that if people are watching for long periods of time, we’re increasing our ad load and our revenue.”
Damian Pelliccione, CEO and co-founder of Revry
“Streaming intelligence” specialist Conviva has a Streaming Performance Index (SPI) that provides a combination of all of its most important KPIs: rebuffering ratio, video startup time, and bitrates. “That’s really going to give an overall picture,” says Melissa Yurash, Conviva’s senior customer success engineer. “There’s a direct correlation between that SPI and engagement. That’s the other really important metric that we’re looking at: the average minutes per viewer.”
According to Jeff Annison, co-founder and president of Legion M, “The KPI when you produce a movie is, how much revenue does that movie make?” Legion M is a crowdfunded entertainment company comprising 35,000 shareholders. “What makes us unique is that we’re the first one that has been designed from Day 1 to be owned by fans. We’re using equity crowdfunding to effectively finance content development.”
Advertising Business Model
With 5 million monthly active viewers, Revry has relationships with all of the environments it distributes to, but doesn’t own the user (or user-related) data in those environments. “We have a basic understanding via each of the different dashboards that we have access to,” Pelliccione says. Revry can see watch time and if a title is doing well. “I think this is an industry-wide issue that has not been solved. In a direct-to-consumer app, we can get all the data that we need because we build or have partners that build it with us. But in these other FAST environments, we don’t necessarily get all of that data and analytics,” Pelliccione notes.
In any given quarter, Revry has 50–75 different advertisers trying to influence the $3 trillion a year in disposable income from its market. “That is definitely a very big addressable market, and right now, Revry commands the greatest amount of video ad inventory for the LGBTQ market in the United States and even globally,” Pelliccione says.
Revry gets data from its ad servers and its header-bidder solution. According to Pelliccione, “We’re spending our money on the DSPs [buying advertisements to run on its content] to market ourselves and then, via that, matching it with the data that we do have access to as a distributor. We can now see some correlation and some patterning on how to better program and super serve our audience.”
Revry started as a subscription service and then in 2018 transitioned to FAST and AVOD. In early days, the startup was more gut reaction-oriented before becoming more data-focused. “We became really smart and particular about how and where we were collecting that data to understand the programming. In the FAST model, it’s just always about the programming,” Pelliccione says.
“We have a really big flagship show called Drag Latina—imagine if RuPaul’s Drag Race and Eurovision had a baby. It’s a competitive drag show all in Spanish language,” Pelliccione explains. Revry is programming much like any TV cable programmer by focusing on which day and time to deliver content. “We always do Sundays in prime time, and we use day-parting to understand when we can garner the biggest audience,” Pelliccione says. Drag Latina had 1 million concurrent live viewers on both Spanish- and English-language channels. “It was the biggest ever opening show,” according to Pelliccione. The viewing pattern then valleyed for the next several shows, but was expected to peak for the finale.
Would this have changed if Revry had more data about viewing patterns? Revry could have reminded viewers to tune in, recommended other content, understood the exact locations of viewers, or understood on what devices the viewing occurred or the overall consumption time.
Should distributors like Revry expect more? Realistically, they get data and—more importantly—revenue from ad sales. They also are able to buy their own inventory, which provides some additional viewing stats, so the end result is they can extrapolate more than they would otherwise have access to.
Legion M has a different relationship with data. The company measures engagement with its shareholders or wider audiences to estimate how engaged they are. “We’d love to know when somebody buys our movies on iTunes so that we can market against that and have that as a conversion event,” says Annison. Since this is not the case, Legion M has developed a few different approaches to gain data insights.
“The whole thesis of Legion M is this idea of equity crowdfunding a company that is literally owned by a legion of fans,” Annison explains. “This has competitive advantages when you’re bringing your projects out to market—you’ve got a built-in audience to go to opening night, talk with their friends on social media, and help do all the things that allow you to rise above the noise, get discovered, and get eyeballs.”
Jeff Annison, co-founder and president of Legion M
The company goes to film festivals to acquire and partner with potential film properties. “We work really hard to figure out ways to make it fun to collect data from [fans] directly,” says Annison. In an effort to generate better engagement, instead of creating surveys, Legion M developed a game called Film Scout, which “allows our community from around the world to help us evaluate films [at festivals] so we can get an idea of what people are looking for,” Annison notes. Players are scored on how effectively they predict the films’ performances.
“We release a leaderboard at the end of the festival that tells you how you stood up against everybody else that’s playing, including the professional scouts,” explains Annison. “The top 5% get invited to our Elite Scout program.” These elite amateur scouts are identified as a subset of Legion M’s investors/viewers that has a demonstrated ability to predict what a larger audience will think. “One advantage we have is harnessing the wisdom of the crowd to make decisions,” says Annison. This gives executives a much more curated choice of data from which to make initial selections on what to fund next.
As a Vendor, What Questions Come Up About Analytics?
Conviva’s Yurash says that clients have “very specific” questions they hope analytics will answer. Most of them are related to the “existential question of how you define an experience,” she explains. “What we’re trying to do is measure something that is qualitative and make it quantitative. How do you measure that?”
One quantifiable drag on user satisfaction since the dawn of streaming is time spent waiting out network rebuffering. Yurash notes that for the last couple of years, “we’ve been focused on the networking and infrastructure, where the industry benchmark on rebuffing was around 2% [in 2020]. Now it’s at 0.2%”—a significant improvement. Still, she says, the question remains: “How do you define engagement, and how do you define quality of experience? For VOD, the biggest question is, do you have bingers, and are they auto-playing? Also, is your recommendation working effectively enough that your autoplay is capturing at least 60% of viewing and they’ll continue to binge to the next episode?”
Melissa Yurash, senior customer success engineer at Conviva
Beyond the reliable metric of continuous viewing of bingeable programming, other questions come up around churn. Generally, a new subscriber enters into the system, watches a platform’s marquee content, maybe engages in the first week, and then completely drops their engagement in the next week. That scenario “is something that our machine learning models catch when we’re trying to predict churn,” Quickplay’s Narayanan says. “For a lot of customers, engagement in the last 4 weeks is a very strong indicator that the subscriber is likely to churn. There are certain thresholds, depending on the type of content—long-form, short-form—that we’ve identified. It could be as simple as total number of playback minutes or a total number of playback sessions that you initiated in the last 4 weeks. If it’s below a threshold, we clearly understand that that customer is likely going to churn.”
Although FAST is rising quickly, the medium’s churn numbers remain striking. “We are talking about 35%–40% churn for some of these services compared to the low-single-digit churn that we are used to for pay TV. It is definitely a major issue,” says Narayanan.
How can FAST providers mitigate churn? “By segmenting the users, we are able to understand what drives them and give them what they want,” Narayanan explains. “A fanatic [viewer] will really need a UI/UX experience that’s more tailored toward their needs. They want more interactivity, polls, trivia, and leaderboards.”
The Quickplay platform is collecting data, but only 50% of companies are using data to try to upsell viewers, Narayanan says. Plus, companies that don’t own the platforms that distribute their content can’t benefit from collecting it. So, the more important question might not be if we can get specific data, but do we have rights to it at all?
New Data Rules of Engagement
When data is siloed and captured in different formats, content owners and services have no way to cross-connect the information from a marketing campaign to engagement behavior. “There was a Forrester study that was done on Conviva 2 years ago,” says Yurash. The results showed that “it’s really important to have data that’s reliable, fast, in real time, and can be used across multiple departments and not be siloed.”
“Many systems—from playback to marketing to CRM—are built to be standalone, serving one purpose,” says Narayanan. “Nobody’s really looking at how systems talk to each other or use a unique identifier to identify subscribers. How do you understand that a user coming into the front end is the same user who’s actually generating a $30,000 lifetime value?” He adds that Quickplay is currently tracking viewers who use multiple playback environments. “We understand at each and every step who is that unique ID that comes through the
different stages in the funnel.”
Both Narayanan and Yurash agree on creating a unified dataset to drive unique insights. “We need to be able to map effectiveness,” Narayanan explains. “Where are you spending your dollars? What is your return on advertising spend? And how does that translate into lifetime value? Whenever we capture signals from the radius modules of the streaming platform, we are trying to normalize as much as we can, clean up, and also enrich the data using third-party data sources.”
The key to circumventing fragmentation and data silos, according to Yurash, “is democratization of data, making data accessible. How do you do that? Ten years ago, I was writing code, writing my own reports, and now I’m able to use these dashboards in real time. I’m helping other people to use these dashboards themselves and use data themselves, even ingesting APIs into Datadog or BigPanda.” From there, her users can transform their data into actionable insights. In this way, a company’s own data can be used within various groups in a media company, whether analyst, marketing, product, or operations.
The path to insight isn’t as direct for Revry. “We have access to dashboards with all of our partners; some of the partners will send us Excel sheets,” Pelliccione says. “We have full-time employees who have to translate that, so we can make sure that we’re paying our royalties correctly. Then marketing can try to correlate the engagement that they’re seeing on their spend, so programming can understand that no one is watching this title and decide to take a title off of FAST and keep it on VOD.”
The issue for companies like Revry that don’t own their data is that the platforms they are distributed on will not always share everything. The result is that these media companies don’t get the types of insights that vendors like Quickplay or Conviva can provide. “I think we’ve hit the tipping point. This is an issue that only the government will be able to solve—specifically the FCC,” Pelliccione says.
“I don’t think that this is going to get solved until there is lobbying or at least a government official that stands up and says, ‘Data needs to be owned, shared, or purchased.’”
Changing State of Business
We’re all familiar with the stereotypical media executive who, convinced of their unerring sense of the market’s tastes and direction, makes major content decisions based on hunches rather than research and analysis. Does that still ring true, or does the abundance of accessible data make things different now?
“I think it’s remarkable how many content decisions made in Hollywood are still based on the gut feelings of a small group of executives,” says Legion M’s Annison. “The first year we were at Sundance, it was all about the guy in the Uber saying, ‘I’m hearing a lot of people talking about this film.’ There’s got to be a better way. For us, data is essential.
“When we can harness the power of the crowd, that data helps not to make the decision like it’s a democracy, but to provide the insight and that extra lens that allows us to make good decisions,” according to Annison. Through its specific business model, Legion M is focused on the idea of bridging the gap between the content creator and the consumer. This approach has allowed it to have more control over its data than a more traditional media investor might have.
If we look at the data haves and have-nots, it’s obvious that Legion M and Revry (both have-nots) have found a way to run their businesses without the in-depth data they could get if they were able to take advantage of all that Quickplay and Conviva can offer pureplay, O&O media. By making creative uses of the data they do have, both of these media companies have prospered, and that seems like the best outcome.
The big challenge here is that the technology for data capture and analysis has evolved to a point we never would have anticipated years ago. The next step, we can hope, is for media companies to gain more control over the data originating out of their businesses.