Cable Ops Lag as Google, Others Mine Set-Top Data for Advertising

Michael Manzo, CMO, Openet

Michael Manzo, CMO, Openet

February 22, 2010 – If the ability to tap the digital set-top box as a data mining resource is a measure of where the balance of power will lie in the emerging advanced advertising arena, the holders of the keys to the kingdom may have some catching up to do.

Over the past year two giants of the Internet world, Google and Microsoft, have parleyed expanding access to set-top data into a key selling point for their TV advertising platforms. Microsoft, leveraging its acquisition of set-top data aggregator Navic, is facilitating NBC Universal’s advertising operations in the Los Angeles market in what officials say is the first of many such deals now in the offing. Google, in affiliations with Dish Network, TiVo, Nielsen and Visible World, is making use of data generated moment-by-moment across millions of set-tops both for metrics on which to base valuation of the ad avails it auctions through Google TV Ads and for new ad targeting initiatives.

But it remains to be seen how aggressively the cable companies will act to make use of new data-gathering platforms on offer from industry vendors that could strengthen their positions as direct ad sales players, both locally and at the national level. Theoretically, through the auspices of Canoe Ventures, cable operators could leverage highly detailed information gleaned from tens of millions of digital set-tops to provide a far better measure of viewing than currently exists and to establish targeting profiles for addressable advertising.

“The cable industry has been talking about audience measurement capabilities for a long time, but it hasn’t adopted an end-to-end solution or a migration strategy on how to get there,” says Michael Manzo, CMO of Openet, Ltd., a supplier of advanced event processing and transactional management systems for network service providers. “Now there’s a real threat from the likes of Google who have the potential to dismediate operators and programmers in this space.”

Google now claims it has the ability to compile detailed viewing records on all types of programming across four million Dish subscriber set-tops and 1.6 million TiVo set-tops, which not only represent a huge sampling base for rating program performance but also provide advertisers detail on what types of viewers are watching specific programs based on geographic and demographic profiles. To enhance ad targeting capabilities to a larger national viewing population the company is also tapping Nielsen’s Claritas Prizm data base, which aggregates Census and other data to define every U.S. household in terms of 66 demographically and behaviorally distinct segments.

“2009 has been a breakthrough year for our advertisers, inventory partners and the TV Ads platform,” says Mike Steib, director of Google TV Ads and emerging platforms, in a recent posting on the Google site. “We have worked with thousands of advertisers to target TV campaigns more effectively, and our network has grown to include 17 inventory partners. Google TV Ads exceeded 100 billion served impressions in 2009, as we continue to engineer solutions for better targeted and more relevant ads.”

The company is applying the analytics capabilities of its Web-based ad network to provide extremely detailed views of viewing performance, Steib notes. For example, Google, by tracking how many seconds viewers watch before changing a channel or hitting fast forward on their DVRs, is able to assure advertisers they won’t be charged if viewers watch less than five seconds. And Google has taken steps to incorporate tracking of viewing of its programming affiliates’ content on the Web as well as on TV channels.

To provide its advertisers the flexibility to leverage all its data for purposes of targeting ads, Google has licensed access to Visible World’s software, which allows advertisers to tweak their ads to fit specific user profiles and to switch out ads on the fly based on how ads are performing. Visible World recently enhanced its targeting capabilities by partnering with Acxiom, Experian and Nielsen to create a data exchange that pulls together all the available set-top data these companies aggregate through their various affiliations with service providers and other entities.

Of course, Google can only auction avails from the 17 cable networks it has cut deals with, which include a handful of majors such as CNBC, MSNBC Bravo, MTV, USA and many lesser niche channels. And Google has not cracked through to the broadcast networks.

Similarly, Microsoft, while it has access to data compiled by Navic from 35 million set-tops, has yet to announce program affiliations beyond its NBC Universal deal in Los Angeles. But, as these players seek to gain traction, the pressure is building on cable operators to come up with a set-top-based data base that will enable audience measurement and ad targeting on a massive scale, far beyond what these competitors can accomplish individually.

“Everybody realizes it’s a problem,” says a Time Warner Cable executive, asking not to be named. “It’s widely recognized that the ability to use this data on a national level would be game changing for the industry.”

But, he adds, the industry must come up with a set of best practices respecting the gathering and reporting on this data that operates on a continuum, which means there has to be a mediation process that everyone agrees on that can work with all the incompatible vendor-specific billing and set-top data collection systems in operation across thousands of cable systems. “The reality is Time Warner Cable and the other Canoe partners are working together to help the industry define a best set of practices, but it’s not easy,” the executive says.

While MSOs have access to the raw data on a scale unmatched by anyone else, they face big hurdles trying to put that data to practical use without help from new advanced software systems, says Alan Breznick, a senior analyst at Heavy Reading. “For one thing,” Breznick says, “set-top box and video server measurements tend to be operator-specific or even system-specific, leading to inconsistent data formats. So pooling set-top data from several systems that are run by the same cable operator can be a tremendously complicated task, and this grows more complex across systems managed by different operators.”

But, he adds, there are now solutions in the market that allow operators to systematize data gathering into a useful mode of measuring viewership click by click, second by second on all TVs in the home and to define audience profiles in ways that are highly useful to advertisers. “The set-top data is there to use, the key is to do it in a better way,” he says.

The advantages are obvious. Conventional ratings reports from Nielsen rely on reporting from just 18,000 “people meters” plus another 25,000 or so polling boxes placed in households nationwide. “Not only is the current sampling base too small to catch all the nuances of viewing on the local level; Nielsen doesn’t even measure many of the less viewed channels, which means it’s missing about a third of the prime time viewing audience,” Breznick says.

Nor, he adds, is there any sure way to determine whether people are actually watching when the TV is on or what the viewing patterns are on all the TVs in the house. Studies have found that diary errors and pushing wrong buttons on people meters may be distorting national ratings by up to eight percent, he says.

In contrast, the new software technologies collect real-time, second-by-second click streams from three basic sources – real-time viewing (linear broadcast and switched digital video), on-demand viewing (VOD and DVRs) and interactive engagements (interactive TV programming, tru2way applications and Embedded TV Binary Interchange Format, or EBIF, features). This means, for example, that an SDV manager can collect the linear broadcast and switched digital tuning data while an EBIF/application server gathers the critical set-top box data and a VOD server would collect all on-demand purchases.

A vital element to the new capabilities is the aggregation of all this real-time click stream data into data bases where it can be stored over long periods of time and accessed for use in whatever types of reports match operator requirements. For example, operators can send transactional data, such as VOD purchases, to their billing systems for processing, as well as collect and store ad placement data from the ad servers in the data warehouses for reporting and analytics, providing key visibility into the viewership of specific ads.

A big first step toward facilitating the national effort to build a measurement and reporting platform for use through Canoe would be action on the part of individual MSOs to implement these new data analysis systems in support of a holistic, highly detailed view of content consumption across their footprints. “We’re seeing a need for evolution not revolution,” says Openet’s Michael Manzo.

Openet offers the new type of audience measurement platform described by Breznick. The solution enables the correlation of real-time television viewing with demographic and psychographic profiling information to deliver a highly granular tabulation of data that operators can leverage many different ways. “This has been deployed and is a proven solution,” Manzo says, adding that the platform performs similar functions with regard to high-speed data and broadband mobile usage.

The key to moving forward, Manzo suggests, is to identify immediate benefits to existing spot sales operations that will justify putting a platform like Openet’s in place, rather than trying to rationalize implementation on the basis of long-range goals alone. “The first use of this data will be by operators for internal media sales use,” he says.

Indeed, as Breznick notes, there’s a real payoff for operators who can use a platform like Openet’s to measure viewing on channels not surveyed by Nielsen. “It’s an opportunity to turn avails on those networks into real paying spots,” he says.

Best estimates from recent analysis suggest adoption of set-top based techniques would allow cable operators to increase their own spot sales revenue by as much as seven percent each year, Breznick says. Operators can also use the new audience tracking software tools to recover revenue currently lost to unbilled subscriptions. Moreover, the new measurement system can provide a full accounting of viewing on time-shifted programming, in contrast to the three-day limit imposed by Nielsen’s new 3C system, which, in any event, has yet to be widely adopted.

Daunting as the challenge might be for legacy data collection and reporting systems, the comprehensive corporate-wide capabilities extending across all vendor products that Openet offers is right in stride with the “transactional intelligence” solutions its customers already have in operation worldwide, Manzo notes. “Right now 70 percent of all U.S. voice and data traffic is processed through Openet systems,” he says, citing AT&T, Sprint and Verizon Wireless as major customers in the billing reconciliation and revenue assurance space.

Globally, he adds, 84 entities serving 500 million customers are using Openet, resulting in the processing of over 20 billion events and transactions per day. “All of this requires really fast processing of data from complex networks, which we do at sub-100 millisecond latency rates,” he says.

Along with speed and scaling of software processing capabilities, the volume of data to be managed in the cable audience measurement arena requires massive storage capabilities. “You need to store pedabytes of data over 12 to 18 months,” Manzo notes. “You’re not gathering up FTP files every night. Your software has to manage a steady inbound volume that adds up to billions of daily events and records.”

This includes detail on ad campaigns collected from all the video elements – VOD servers, DVRs, ad servers and set-tops – as well as everything off the high-speed data and wireless platforms, once operators are ready to weave those outlets into a converged ad view. And all of this has to be enriched with reference data collected from billing and other sources.

Once all this information is aggregated it has to be analyzed and distributed in reports that are easily accessed by all stakeholders on user-friendly dashboards. “There’s a tremendous amount of slicing and dicing going on to run reports very fast against such massive amounts of data,” Manzo says.

Many issues have contributed to cable operators’ hesitation to put such capabilities to use, especially at the national level. One is their reluctance to share information. Another is the resistance of the programming community to accepting precise measurements that could result in damaging adjustments to their standings in the Nielsen-generated ratings competition. And, of course, the word “targeting” has become so sensitive owing to the privacy uproar that industry executives have stopped using it in public forums.

But with addressable advertising strategies that employ privacy-protecting anonymity techniques now on offer from the likes of Google, Microsoft and Visible World, it would appear the cable industry will have plenty of cover on the issue. The real risk is that targeting could gain mainstream momentum before cable operators get to the starting gate.

As for the battle over ratings systems, here again the demands of Madison Avenue in a weak TV advertising market are becoming too great to ignore. “It will be a long, long time before Nielsen is replaced holistically,” Manzo says, “but there’s every chance Nielsen may refine its model and work with Canoe rather than looking on Canoe as a competitor.”

Openet is only in a competitive position with Nielsen by virtue of Nielsen’s adherence to an outmoded model, he adds. “If Nielsen is willing to improve the quality of their data-gathering methods, we’re more than willing to talk with them.”

Ultimately the economics of the benefits attending the type of measurement capabilities Canoe can bring to the table will prevail, he asserts. Measuring that 30 percent of prime time viewership that watches channels not measured by Nielsen will generate new advertising revenues, he says – for example, $6 billion annually on a ten percent lift. And the ability to reach highly targeted audiences will bring in a vast population of advertisers who have been unable to make use of TV.

As for whether cable operators will finally cooperate on setting a national template for data aggregation, clearly, time is running out on the old modus operandi. Whoever ends up leveraging set-top data to the full potential demanded by the advertising community will be in the driver’s seat when it comes to negotiating the terms in the emerging advanced advertising arena.