As pay TV and broadcasting increasingly go OTT, analytics will penetrate to the heart of their strategies and technology platforms to guarantee QoS and follow subscribers across multiple screens for monetization and churn busting.
We are not talking perhaps about satellite, cable and digital terrestrial going away, but becoming just part of an overall delivery infrastructure that will also embrace the unmanaged Internet as well as wireless access over WiFi and 4G/LTE for connected device delivery both in and out of home. There will also be a combination of broadcast/multicast and unicast for linear or live content, depending on its popularity.
This has all raised the bar for TV analytics across both its key dimensions of video quality and service personalization. For video quality, multiscreen raises new problems associated with unmanaged networks and end devices, while for personalization there is the challenge of monitoring subscriber behavior across different screens and exploiting that for recommendations and targeting of both ads and programming.
These are the new frontiers of a new land-grab in supplying technology to what is essentially a brand new industry.
There is great potential for synergy across services that unite broadcasting with online and incorporate social media as well, but few operators have even begun to tap that. To date it is the subscription VoD operators like Netflix that have gone furthest in exploiting intelligent recommendation and as a result have triumphed over pay TV operators, especially in the US, even though at least until recently they have been peddling much the same content. Now they are exploiting their own original programming to great effect as well.
Most pay TV operators now have a good handle on their subscriber activities on the big screen by analyzing return path data from the set top, enabling them to discern what is being watched when, observe channel changing and see how the EPG is being used. In this way it complements the information provided by traditional TV panels with live information, as well as much finer grained statistical analysis that can for example measure viewing of niche channels much more accurately.
But there are significant limitations, such as inability to drill down into the behavior of individuals within a household and also in the case of IPTV when multicast is enabled not being able to detect trick play functions.
It also provides no indication of OTT consumption or quality inside the home or outside. This is where the action is and sucking in vendors from all over the analytics spectrum, including some not until recently in the TV field. It is sparking partnerships to bring together different components and start developing comprehensive analytics packages for multiscreen TV.
When considering such activity it is worth going back to basics and identifying what data we are looking for and where it is coming from. We can sub-divide this into data collected directly from the operator’s delivery infrastructure and what is sometimes called third party data from panels and increasingly now social media networks.
The third party information is valuable for broader planning and content acquisition strategy, although social media feedback can now direct immediate tactical activity such as targeting recommendations at specific demographic groups. The infrastructure data goes to the core of service delivery, personalization, customer relationship management and monetization. This can then be broken down roughly into three components, the Content Management System (CMS), the network and the end device.
Of these the CMS is easiest to address, since it knows exactly who requested what content when and to some extent the end device it was played on. Between the CMS and the network lies the headend, from which a broad higher level view can be obtained across an operator’s different target platforms. This introduces another element, content or revenue security, which itself will become more and more entwined with analytics. Firstly a lot of the analytics data itself is personal and subject to stipulations over privacy and so needs protecting just as much as the content itself, which can most conveniently be done by the same security systems. Secondly and slightly less obviously security, operating on an end to end basis, is itself a potential source of data about content consumption and viewer behavior, if less about QoS. At the head end the security system should be able to view entitlement data to determine target device types at a high level, providing information valuable for service planning and market research, suitably anonymized to prevent identification of individuals.
When it comes to the network the focus is on QoS, monitored via probes deployed at different stages from the core out to the access network. Here Ineoquest has become one of the main players, having cut its teeth in general network monitoring with tools to measure QoS specified by the 8-bit Differentiated Services (Diff Serv) field in the IP header used to classify packets for prioritization. This led Ineoquest to address videoconferencing, Telepresence and other applications where prioritization is essential for QoS over the Internet. But recently Ineoquest has come to specialize almost entirely in video delivery.
The network increasingly includes a wireless domain for the final loop, creating an additional challenge for QoS, but also a significant opportunity for discovery of a lot more about user behavior and consumption patterns on smartphones, tablets and laptops around the home. The QoS aspect is the immediate issue, since many pay TV and broadband operators are now having to field questions from customers about WiFi related problems and train their first line support people accordingly, even though they have no control over that domain. A revealing statistic is that up to 40% of households in the US and UK with WiFi linked to their fixed broadband are deferring to 3G or 4G mobile services to access the Internet, citing slow WiFi speeds and intermittent coverage around the home and garden as the primary reasons. For operators this equates to loss of user engagement for potential monetization, which seems unnecessary because WiFi on paper is faster.
At least some WiFi technology specialists such as Turkey based AirTies are now working on this area. In a recent blog AirTies argued that big data analytics can work at two levels over WiFi. Firstly it can monitor the immediate performance of the WiFi network itself and secondly it can gain valuable information about usage broken down in various ways, such as by user, service or location within the home. Such usage information, said AirTies, can help improve performance by identifying black spots with a poor signal but can also enable operators to improve the overlying broadband or TV service in a number of ways, such as targeting individual users within the home. There is also the potential to identify issues specific to each home that are causing intermittent performance degradation, which can just happen when somebody takes a tablet to the bathroom and starts watching Netflix there.
The other interesting aspect of WiFi Big Data is its ability to drill right down into household activity, not just by user or device but also room. It might be that movies are viewed mostly in the bedroom and sport in the kitchen and something else in the toilet for example, which might at least be of interest for analysts.
Then we come to the device itself, which is the only place where QoS can be observed directly from the user’s perspective. This has become a major focus in the multiscreen era where premium content is being consumed on connected devices no longer issued and managed by the operator in almost all cases. This is bringing new vendors into the fray from a variety of angles. We are seeing some vendors such as Sweden based Birdstep offering smart analytics packages to mobile and fixed broadband operators combining data from the device and the wireless domains, both WiFi and cellular. At that level data can be gathered not just about what the user is watching, what the network is and how well the video is being displayed but also pertaining to the device itself. Operators can see what impact an app is having on battery life for a smart phone or tablet and then take action to improve that, for example by implementing intelligent WiFi power management.
Pay TV security systems are also in a position to gain valuable data from the device since they deliver keys or licenses there relating to specific assets for each subscriber. As Verimatrix claimed in its recently published e-Book entitled Getting a “Verspective” on Secure Multi-screen Analytics, this can give the operator a detailed view of content consumption for each household, with the ability to drill down further in some cases to individuals.
Bringing together all these components is a challenge not just for service providers but also vendors, sparking a spate of partnership and takeover activity. Established TV analytics vendors are at the center of partnership and takeover activity involving both other parts of the TV ecosystem and encroaching on players from the IT and general Big Data sectors.
Content security players seem to be playing a pivotal role, with Verimatrix especially active on both the partnership and takeover fronts, having in September 2015 acquired Concurrent’s multiscreen video analytics business for $3.5 million. This brings Verimatrix analytics skills plus access to Concurrent customers in this frame, including a few Tier 1 operators like Kabel Deutschland. Verimatrix is also working with a number of vendors across the analytics spectrum, including Genius Digital for actions based on viewer data, Fourthwall Media, which specializes in return path data from cable set-top boxes, consumer viewership information provider Rentrak, along with Ineoquest for network monitoring (See separate piece this week).
In most cases Verimatrix is helping secure the analytics ecosystem and at the same time sourcing some of the critical data. It is helping Ineoquest for example identify encrypted content during transmission over the network. As its e-Book explains, Verimatrix comes in by decrypting the content that it is protecting, so that Ineoquest can analyze the flows with its probes and provide valuable information about QoS.
Irdeto is taking a slightly different tack, at least as far as embracing online analytics is concerned. It argues that it is futile and a waste of resources to retread ground that Google has already trampled all over with its analytics. Google has all the tools and most probably has already gained valuable insights into any given operator’s customers’ online behavior. Therefore operators might as well use it and overcome their wariness of Google as a potential competitor. Irdeto therefore urges operators to incorporate Google analytics into their multiscreen platforms. This will help them monetize multiscreen more effectively, for example by linking a given customer’s content viewing statistics and website ecommerce behavior, which can be valuable for cross platform ad targeting. Such information is already driving online advertising and can now be applied across multiscreen TV viewing.
Apart from security vendors from within TV, analytics is also drawing in the big beasts of IT, such as IBM, HP and Accenture. Of these HP was ready to weigh in with intelligent analytics on the back of its $7 billion acquisition of Cambridge, UK, based Autonomy for $11.7 billion in October 2011. Apart from other things Autonomy had some sophisticated software called IDOL (Intelligent Data Operating Layer), capable of divining trends and patterns from large structured and unstructured data, including images, audio and video. But this takeover quickly turned sour amid claims and counter claims over accounting irregularities that were alleged to have caused HP to overvalue Autonomy. All this has taken eyes off the ball and HP has yet to make any serious impact in video analytics.
IBM is very much in the business, having developed a cloud based analytics platform used by Television Broadcasts (TVB) of Hong Kong. TVB is using this to provide more accurate ratings forecasts about audience sizes and demographies to advertisers, combining information from its own data warehouse of ratings data and intelligence from social media, all analyzed by IBM.
Accenture has moved furthest into TV analytics of the big IT players and has just consolidated its hold by acquiring S3 TV Technology, the analytics arm of Dublin-headquartered Silicon Software Systems (S3), for an undisclosed sum. This brings Accenture all S3’s TV technology spanning consultancy, automated testing, service monitoring and diagnostics, along with access to customers such as Sky, Time Warner Cable, BT and UPC Cablecom.