September 3, 2018
Content will always be king.
But in the home entertainment business, as in the broader entertainment industry, data is increasingly regarded as the crown, throne and scepter.
Data analytics, quite simply, is predicting future consumer behavior based on as much information about your target audience as you can get. Data comes into play when deciding windows, store allocations, distribution platforms, marketing campaigns, even greenlighting movies.
“Data hasn’t changed the importance of content, but it has helped make informed decisions,” says Jim Wuthrich, president of Warner Bros. Worldwide Home Entertainment and Games. “From deciding which content to make, how to price, where and what to market, data influences everything we do.”
At the studios, data analytics has become integral to doing business. The home entertainment divisions of the big studios say content and data now pretty much go hand-in-hand.
“Content is what we do — how it connects, moves, inspires and makes consumers and audiences feel something is what we are all about,” says Kim Overall, EVP of consumer insights and innovation for Sony Pictures Television and Sony Pictures Home Entertainment. “Understanding what is resonating with people, why and where, is how we get the right content into consumers’ hands.”
“Data and analytics don’t provide all the answers as there are many things that can’t be measured or interpreted — you still need talented, impassioned people to create, market and distribute great content,” adds Wuthrich. “We try to use all our resources to predict outcomes and data is one tool. But there are limitations. For instance, despite our years of experience and endless data, it’s hard to predict how many people will show up on street date, be it measured by box office on opening weekend or disc sales on street date. New tools are coming all the time (such as social sentiment, trailer views on social platforms, etc.) and will continue to improve predictions.”
How has Hollywood’s reliance on data analytics changed in, say, the last three years?
“It depends on which part of the business,” Wuthrich says. “We’ve always seen Warner as a data-driven company, but the tools continue to evolve. For our games business, the fastest-growing category is free-to-play games. Two examples of successful games are Golf Clash and Game of Thrones Conquest. Data and analytics drive the success of these games, from finding an audience to optimizing game play and monetizing.”
“We have invested in data and analytics as a key input to many decisions across the studio (or distribution group),” Overall adds. “As consumers and audiences tastes, preferences and behaviors evolve we have to be in lock step understanding the impact it has on our business.”
Independent content suppliers also are embracing the new world of data analytics. Cinedigm Entertainment Group president Bill Sondheim, speaking in July at the Los Angeles Entertainment Summit in Hollywood, told attendees, “We put a great deal of focus and investment in securing extensive and detailed data, and we have a strong culture of analytics and self-reflection. Cinedigm looks at each piece of content to determine the right demographics. It starts and ends with the consumer.
“And once we’ve determined who will want a particular piece of content, then we determine where that consumer shops. That could mean Walmart or Best Buy, but it can also mean Netflix or Amazon or iTunes or even something more unique, like Crunchyroll.”
Later, in an interview with Media Play News, he elaborated: “These new data analytics are playing an immense role in how we window every film or TV show we represent. First, it is important to note that as a distributor you must be channel agnostic. Our first responsibility to the rights owner is to maximize their revenue potential. This means we must model out numerous windowing scenarios and attach forecasts based on the data to see what is the best fiscal outcome.
“While other considerations like competitive releases scheduled and promotional opportunities and talent support are all taken into consideration with any windowing decisions, at the end of the day financial return on investment is the final arbiter.”
The use of data and predictive analytics has become increasingly sophisticated, Sondheim maintains.
“We have taken a culture of data analytics and applied it to virtually every aspect of our operation,” he says. “From the earliest stages of content development and title selection we have relied extensively on comparative competitive data. We track the performances of individual stars, we segment the genres and we carefully measure the effects of seasonality and consumer traffic. This has provided Cinedigm with a clear and well-defined roadmap of where we want to deploy our working capital as we invest in new content and productions.
“This analytic process then is applied to marketing and package decisions. We determine the target audience and look carefully at where they shop, the prices they respond to, the art work and graphics which seem to perform best and the promotional vehicles and advertising outlets that appeal most to the targeted consumer.
“From there, online or in-store placement becomes the focus and again we look at past sales performance metrics to determine how we can best maximize the value of any given title.”
How is Cinedigm management able to view all this rich and detailed data?
“We rely on internal data collection as well as outside subscription services and, on occasion, customized consumer studies,” Sondheim says. “We maintain an elaborate data warehouse that allows our sales and marketing teams to delve deep into granular data that comes from our sales history.
“We also subscribe to numerous outside data collection services that gives us competitive performance data looking at each major channel of distribution, from cable buy rates to VOD transactions to units scanned at stores across the United States and Canada. When you combine all these factors and make them a crucial part of the decision process it can have a profound impact on your ability to accurately predict future performance. We spend significant dollars and human resources and we see an excellent return on those investments.”
The drive toward relying on data and predictive analytics has accelerated in recent years with past experience “a gentle footnote in the process,” Sondheim maintains.
“Increasingly, we are going to have to develop dynamic real time and predictive ways to understand what is resonating with consumers,” Overall adds. “Equally, the other opportunity is how data and analytics can help us be more efficient across the distribution business.”
The Quest for Efficiency
The quest for efficiency is what drove John Daly, then an SVP at Sony Pictures Entertainment in charge of supply chain for home entertainment, to call on Algomus three and a half years ago. Algomus at the time was handling the studio’s vendor managed inventory.
“I realized, oh my gosh, there are a bunch of smart guys here, a bunch of guys from MIT,” he recalls. “The meeting goes from an hour-long meeting to like a four-hour session.”
Daly’s problem was the eternal one for supply chain executives, how to get the right physical product in the right amounts to the right stores at the right time. Sony was shipping to 25,000-plus stores, which kept Daly up at night.
“I needed to understand how the stores were behaving differently,” he says. “Although we had the data, I just couldn’t see it. I could get to it in an excel file, but it would take me forever to play around with it, and by the time a week and a half went by, it would be old data.”
That’s where Algomus stepped in, creating tools to speed up that process and evaluate the growing mound of data coming out of the retail pipeline. The tools use predictive analytics and machine learning — or using statistical techniques to give computer systems the ability to “learn” with data, without being explicitly programmed — to streamline the process.
After being a client, just this April, Daly joined Algomus as president.
“What our tools do is it takes all that data, and it simplifies your ability to look at it,” he says.
All stores don’t behave the same, he notes.
“You don’t want too much inventory; you don’t want too little inventory, and you want to have the right assortment for a customer that tends to shop in those stores,” he says. “And the only way we know we can do this is by using big data and analytics to figure this out.”
Algomus tools answer such questions as why a store has a low in stock rate, why certain stores have high return rates, how a particular new release is selling for the first eight weeks of its life cycle, how corrugated placement is selling, how promotions are selling and how certain genres sell in certain stores.
“It also gives us the ability for what I’ll call anomaly detection,” Daly says. “For instance, we had stores that were selling well and in the last two weeks the sales just slowed down, and now it allows us to go back and look at that. We can see all that data down to that store and SKU level very clearly, so that’s where the excitement comes in.”
Algomus tools also help studios allocate limited edition product, such as steel books and gift sets, more accurately.
“You can’t go back and remake them,” Daly notes. “They have to be ordered six, seven, eight months in advance.”
Algomus clients include the majority of studios, but the data analytics are appreciated by retailers as well, he says.
“They are very much encouraged that the studios are taking on this opportunity to drive the supply chain,” Daly says.
The home entertainment industry is in the vanguard in using data analytics, he notes.
“Very few industries are leaning in as much as home entertainment,” he says. “And that’s a pretty good story, that’s a positive story for the home entertainment physical part of the business.”
Exploring the Digital Frontier
While physical distribution has built the business and continues to be a big part of the home entertainment industry, digital delivery of content is the future — and data about that market is key.
“At one point the data focused primarily on physical formats, but as more data subscription services were launched covering transactional VOD, cable ratings, and SVOD performance, we have seen this become a dominant management tool,” notes Cinedigm’s Sondheim. Cinedigm has its own subscription, over-the-top services, Docurama and the Dove Channel.
Perhaps the most direct use of big data and predictive analytics is in the OTT/subscription streaming sector of home entertainment. Netflix, Amazon and Hulu are known for their use of data.
As Thomas H. Davenport and Jeanne G. Harris note in their book Competing on Analytics: The Science of Winning (2017, Harvard Business Review Press), “Netflix employs analytics in two important ways, both driven by customer behavior and buying patterns. The first is a movie-recommendation ‘engine’ called Cinematch that’s based on proprietary, algorithmically driven software. Netflix hired mathematicians with programming experience to write the algorithms and code to define clusters of movies, connect customer movie rankings to the clusters, evaluate thousands of ratings per second, and factor in current website behavior — all to ensure a personalized web page for each visiting customer.”
Netflix also created a $1 million prize for quantitative analysts outside the company who could improve the Cinematch algorithm by at least 10%, according to Davenport and Harris. “It was an innovative approach to crowdsourcing analytics, even if the winning algorithm was too complex to fully adopt,” the authors wrote. “But no doubt Netflix’s data scientists learned from the work and improved the company’s own algorithms. CEO Reed Hastings notes, ‘If the Starbucks secret is a smile when you get your latte, ours is that the website adapts to the individual’s taste.’”
Now that Netflix is churning out original content — the company has said ultimately it wants 50% of its total programming to be original shows — the company uses analytics “to predict whether a TV show will be a hit with audiences before it is produced,” according to Davenport and Harris’ book. “Netflix … has used attribute analysis, which it developed for its movie recommendation system, to predict whether customers would like a series, and has identified as many as 70,000 of movies and TV shows, some of which it drew on for the decision whether to create it.”
Just last month, during the Television Critics Association media tour in Beverly Hills, Calif., Netflix VP of original series Cindy Holland pulled back the curtain a bit on its data analytics. She told the audience that Netflix doesn’t look at demographics per se but at “taste communities,” groups of subscribers who gravitate toward the same shows. She called the connections somewhat “unituitive,” noting that fans of Dave Chapelle’s stand-up also paradoxically like the film The Theory of Everything, a biopic of scientist Stephen Hawking.
Good information, no doubt. But since Netflix and other OTT leaders are proprietary and carefully guard their data “silos,” competitive analysis can be difficult. Calculating an overall measure of the SVOD market is a more complicated endeavor that third-party companies, such as Parrot Analytics, are tackling.
“We are becoming the source of media information that is beyond the traditional,” says Alejandro Rojas, regional director, Parrot Analytics.
Parrot — which provides global demand data on specific content to such clients as CBS Studios International — uses various data sources to measure the overall demand for SVOD programs. For instance, Parrot offers a weekly top 10 of the most popular digital original TV series in the United States, based on the firm’s proprietary metric called Demand Expressions. Demand Expressions measures global demand for TV content through a wide variety of data sources, including video streaming, social media activity, photo sharing, blogging, commenting on fan and critic rating platforms, and downloading and streaming via peer-to-peer protocols and file-sharing sites.
“Parrot Analytics measures ‘popularity’ by holistically capturing digital footprints left by TV consumption journeys across multiple viewing and engagement platforms,” Rojas says.
The data monitored is everything from a “like” on social media to a download from pirate site Bittorent, a rare instance in which piracy may actually help inform the industry at large. It’s a different metric than is provided by services such as Nielsen – which explains why Nielsen and Parrot Analytics charts are not often in sync. For example, for the week ended July 30, Nielsen ranked these five shows as the most popular: No. 1, “America’s Got Talent”; No. 2, “60 Minutes”; No. 3, “NFL Preseason Game”; No. 4, “NFL Preseason Kickoff”; and No. 5, “Big Bang Theory”. Parrot’s top 5 was quite different: No. 1, “Spongebob Squarepants”; No. 2, “Steven Universe”; No. 3, “The Walking Dead”; No. 4, “America’s Got Talent”; and No. 5, “Flash”.
“As opposed to traditional ratings, Parrot Analytics does not rely on panels or depend on existing linear programming schedules,” Rojas says. “Its Demand Expressions is an empirical measurement that gauges consumer interest on thousands of TV shows, independent from their airing status or platform consumption. Its top list reflects the fact that not all views are created equal. TV shows with an active and vibrant fan base tend to outperform those that do not establish strong and sticky followings.”
Nielsen show measurement only includes shows that are currently being broadcast, he notes, while the set of shows being evaluated by Parrot Analytics not only includes shows on-air but also shows that are not currently being broadcast.
“Demand Expressions also reflects how passionate a fan base is while traditional ratings just look at views,” Rojas says. “In the end, it gives you an understanding of the emotional connection [to a show].”
Content and the Machine
The future of data analytics is what some term greater “machine learning” and others categorize as “artificial intelligence” — basically computers beginning to solve problems intelligently. The new analytics schemes can do everything from anticipating your business questions to telling you what star to cast and what genre to produce for a particular territory.
“We’re just hitting the first phases of artificial intelligence,” says Algomus’s Daly. “Where I really think this will go is — and we’re starting to get there now — the tool will answer questions before we ask them and create actions so you can take your hands off the steering wheel. You’ll come in in the morning and the tool will be able to tell you, ‘Hey, you better start shipping product to these stores. Something just happened last night. And I’ve created that order. Here’s what we’re going to do.’”
The new learning algorithms are starting to take some of the guesswork out of what content to produce and with whom.
“We have met with several companies in the last few months that are building robust AI predictive models that deal with casting and script development when in the past we only used these tools after content was already in the can,” says Cinedigm’s Sondheim. “The utilization of data analytics has evolved from a way to check certain decisions on limited aspects of the content acquisition and distribution process to the central driving force. And we see the tools becoming more sophisticated in handling multiple data sources, which allows them to be more accurate and predictive.”
Once such company looking to help make decisions in greenlighting a project is Cinelytic, which launched its analytics program for the film industry about a year ago.
“Our DNA is a combination of entertainment, science and finance,” says the company’s CEO and co-founder Tobias Queisser, noting that his partner, co-founder and CTO, Dev Sen, is a former NASA rocket scientist and that key team members include entertainment industry producers and an MIT data scientist.
“I myself was in finance for 10 years in merger and acquisitions and a hedge fun, before producing independent films,” he says.
Cinelytic’s cloud-based platform is designed to help the industry better inform key decision making across the film lifecycle by offering predictive financial forecasts, key talent analytics and distribution strategies.
“We have the industry’s leading predictive forecasting model, in which you can input key parameters, including the film’s budget, genre, key talent, if it’s adapted from other media and if it’s a sequel or franchise. Based on these inputs, our system forecasts revenues for a range of scenarios across release windows, including digital and physical home video, free and pay TV,” he says.
Cinelytic also allows clients to assess the economic value of actors, directors, producers and writers.
“You can view an actor’s profile and see their top genres, top countries, and top release windows to understand where an actor is most valuable,” Queisser says.
Cinelytic is working toward forecasting audience size and audience demographics, as well as predicting home video revenue for films that are not released theatrically.
“We developed machine learning algorithms which are basically going towards AI,” Queisser says. “Our limit is only the data.”
Predictive Talent Scouting?
The music industry has long depended on talent scouts, and human ears, to discover future superstars.
But even that time-honored process appears to be changing.
Warner Music Group in May announced it has acquired Sodatone, a data analytics platform designed to unearth emerging talent. Sodatone tracks streaming, social media, touring and playlisting data and their analytics provide managers with enough information to track their current roster and search for emerging talent, help concert promoters to find acts that will draw a crowd and assist A&Rs in finding the “next big thing.” CEO Stephen Cooper said in May 2018, “A&R expertise has always been informed by different types of data, but today, tech tools are bringing deeper insights to our decision making.”