Partnering with many of the world’s largest and most successful cable and broadcast networks, RSG Media was in a position to deliver the tools required to deliver this kind of insight.
“We had the rights data, the scheduling data and the data on advertising revenues,” says Shiv Sehgal. “We knew if we could marry our data with third-party datasets based on cross-platform viewership, demographic, behavioral, lifestyle and social media data, we could build a coherent 360-degree view of inventory utilization and audience consumption. The missing piece of the puzzle was the underlying technology to manage, house and explore the data—and that’s where IBM came in.”
Harnessing analytics and cloud data services
RSG Media partnered with IBM to develop next generation data and analytics tools to support its new Media Mantra Advanced Audience Signal Intelligence (A2SI) analytics platform.
Shiv Sehgal comments: “We were already using several leading vendors’ cloud services to support our RightsLogic, and Advant solutions, but for the Media Mantra A2SI Platform, we realized that we needed more than just a cloud infrastructure provider. We needed a partner with the skills and technology to help us manage data on an unprecedented scale, and empower both our data scientists and business users to turn that data into insight.
“IBM is one of the few companies that has a complete vision for cloud analytics. They were the only cloud vendor we found who could offer an integrated set of solutions for building advanced analytics applications and coordinating them with all the relevant data services in the cloud.”
From a technical perspective, the Media Mantra Platform is built on the IBM® Cloudant® NoSQL document store and the IBM dashDB™ data warehouse service, orchestrated through the IBM Bluemix® cloud application development platform. Cloudant’s Schema Discovery Process (SDP) is used to ingest and translate semi-structured data from more than 50 sources, and structure that data into a schema that the dashDB relational data warehouse understands. RSG’s data scientists can then perform in-depth statistical analytics and modeling using the open-source R toolset, which is built into dashDB. The results then flow back into Cloudant, and are presented to users via a series of analytical applications, dashboards and reports.
“The combination of Cloudant and dashDB gives us the best of both worlds,” says Shiv Sehgal. “Cloudant gives us the scalability and flexibility of a distributed NoSQL database as the operational data-store, which can easily expand to support new user groups anywhere in the world. Meanwhile dashDB gives us a cloud-based relational data warehouse which is purpose-built for analytics, with embedded predictive models and BI tools that make life easier for our data scientists.”
Bluemix makes it easy to coordinate the Cloudant and dashDB services, and also to develop, run and manage analytical applications based on insights derived from Nielsen, Rentrak, Comscore, Acxiom, Experian, Hulu, Amazon, iTunes, Twitter, Facebook, Snapchat and BuzzFeed—to name just a few of the sources that RSG’s solutions utilize.
For example, RSG’s Cross Platform Reporting application aggregates data from many of these sources, and provides companies with actionable information on content and advertising viewership. Cross Platform Reporting loads, cleanses and normalizes data from hundreds of distribution partners, and creates dynamic dashboards and reporting.
“How can I monitor my overall product usage in real time?” asks Shiv Sehgal. “Which videos do users want to watch? What genres or videos are most popular? Are my top videos short or long form? Which videos are the most engaging? Which platforms or devices are viewers using to watch videos? Was there a production issue with a device app causing low usage numbers? Which platforms should I invest in? Which campaigns result in more engaged users? What is the popularity of my video service in different countries and geographical regions?
“These are fundamental questions if you’re trying to run a successful network in the digital age—and Cross Platform Reporting gives our clients quick, accurate and detailed answers.”
RSG has also launched a Program Schedule Optimization application on the IBM platform. The application features a machine-learning optimizer that helps networks to create schedulers that generate audience growth by encouraging organic audience flow from program to program within dayparts. This creates competitive scheduling that is able to grab and hold audiences from competitors.
To bring the data into the Media Mantra A2SI Platform and enrich it for analysis by applications such as Cross Platform Reporting, RSG is increasingly turning to IBM Analytics for Apache Spark—a fully managed Spark service available via IBM Bluemix.
Shiv Sehgal comments: “With dozens of data-sets to ingest, performance and automation are key. We used to use archaic declarative database tools to transform our data every day, but with the demands of applications like the Cross Platform Reporting Application—where we have to ingest, process and normalize mountains of data on a daily basis—we knew that this would be neither sustainable nor scalable.
“IBM Analytics for Apache Spark and IBM DataWorks now handle this automatically, bringing the data into dashDB or Cloudant and enriching it on-the-fly. This massively reduces the cost of maintaining our applications and data services. For example, one of our apps used to occupy eight people full time on data preparation—now one person can manage the whole app on their own. The ease of use and flexibility of programming with Spark also makes a big difference to the speed of setting up and coordinating new data streams.”
In addition to high-speed data ingestion, RSG is also increasingly using IBM Analytics for Apache Spark to handle algorithmic modeling tasks.
Shiv Sehgal says: “Leveraging traditional data management technologies, the more dimensions you have in your data greater the infrastructure struggles—not just from a storage and processing perspective, but also in regards to data modeling. With the enormous complexity of our first- and third-party media-related datasets, we were finding that some modeling jobs were taking 12 to 48 hours to completely execute. Spark helps us get over this hurdle and analyze extremely high volumes of data at high speed.
“Real-time rights analysis is going to be vital for our business, because consumers aren’t prepared to wait. If a subscriber wants to watch the latest episode of a show on their tablet via the TVEverywhere Application at the local Starbucks, our clients need to know instantly whether the subscriber has the right to access such content and the appropriate viewing experience—which means checking the fast forward restrictions, device concurrency rules and dynamic ad insertion rules, to name but a few. Spark can help us deliver these automated high-performance and high-value solutions.”
He adds: “Spark is such an exciting technology because it acts as a general-purpose platform for all kinds of analytics. It’s easier to work across all kinds of business problems: we don’t need to use a different technology to solve each different issue, Spark can handle almost anything. This aligns perfectly with our vision for the Media Mantra A2SI Platform—it’s all about eliminating silos and using a common platform for all analytics. And increasingly, Spark is the engine that powers this platform.”
While Spark is already being used for data ingestion and algorithmic processing at RSG, the company is also interested in its potential as a platform for ad hoc big data analysis and rapid development.
Shiv Sehgal comments: “For example, the IBM Data Scientist Workbench is a great concept—it gives data scientists access to Jupyter and Zeppelin Notebooks and statistics tools like R and Python in a learning and experimentation environment, which supports the hackathon-style culture of rapid, collaborative development. Time-to-market is certainly a competitive advantage for many organizations, but the ability to prototype good ideas—backed by the powerful Spark engine—has truly changed the dynamic in which we now work with our clients.
Going forward, RSG Media plans to integrate IBM Watson Analytics™ into its applications, to enrich the user experience and help business users build their own reports and dashboards.
“We’re excited about Watson Analytics, not just to make it easier for our users to create beautiful reports, but as a tool that understands what the user is trying to do and helps them ask intelligent questions.
“For example, at the moment we have tools that allow users to set business rules for the schedule optimization process, but we want to use Watson Analytics to literally prompt and nudge programmers: ‘You have a dip in audience viewership Thursdays at 5:30 pm due to viewers tuning into competitive programming—please try stacking the following set of shows to prop up this weak time slot.’ That’s the kind of power and intelligence that we can see Watson bringing to the table.”
Transforming scheduling and marketing
RSG Media’s Media Mantra A2SI analytics platform, powered by IBM Analytics and Cloud Data Services, is transforming how media companies manage their media rights investments and build hyper-targeted strategies for programming, advertising and marketing teams.
For example, one of RSG’s new applications revealed a “sweet spot” in a client’s schedule. Although the network’s 8 pm slot generates the highest ad revenue (USD 4 million), it is also the most expensive time to air a show (nearly USD 1.5 million). By contrast, airing a show at 5pm costs about a third as much, but that time-slot still generates USD 2.5 million in ad revenue—making it nearly twice as profitable for the network.
For another client, RSG’s schedule optimization algorithms helped to line up a series of shows that would build the biggest possible audience over the several hours leading up to primetime, ensuring the maximum audience size for the biggest primetime shows. This delivered a four percent boost in overall viewership, which equated to a lift in revenue of USD 6 million.
Yet another client was concerned about viewership for a major live broadcast awards ceremony. Leveraging behavioral-based “act alike” algorithms, the client was able to evaluate the best times, channels and platforms to air promos for the event. The results were impressive: compared to the previous year, the awards show gained 40 percent more viewers (from 28.3 million to 40 million), even though the network used 17 percent fewer promos.
Shiv Sehgal comments: “The stats themselves are quite striking, but it’s truly remarkable to have the technology in place to monitor cross-platform viewing patterns, measure changes and unearth strategies out of a massive pile of data to deliver custom audience segments with precision.”
He concludes: “Our partnership with IBM is enabling us to build an analytics platform that can handle the full scope and complexity of the media industry, empowering our clients to build optimized programming schedules and deliver hyper-targeted audiences for advertising and marketing campaigns like never before. And the ability to leverage exciting new sources of data such as Twitter, Snapchat, BuzzFeed, Instagram and The Weather Company offers great potential for us to expand our analytics services in future.
“The apps we are building help business users feel like data scientists—all the data they need is at their fingertips, and they can make smarter decisions faster than ever before. With IBM at our side, we’re excited to expand our range of cloud data services and analytics applications to help our clients get even further ahead of the competition.”