Rentrak Cracks Donovan: Deal Is First To Bring TV Set-Tops To Agency Desk-Tops

In a move that could make planning and buying via digital TV set-top data a reality for many big agencies and their clients, Rentrak has cut a deal to integrate its set-top-based audience estimates into Donovan Data Systems, the company that provides the back office systems and software that power much of what Madison Avenue does, including planning, buying, posting and even paying the media. The deal, which is being announced this morning, is believed to be the first to integrate set-top data directly into the "enterprise" systems that agencies use to conduct business, and could have a profound effect on the way agencies treat the data, especially for highly-targeted niche TV networks that are not currently measured by the Nielsen Co.

The deal grew out of conversations that began last spring as part of a special task force created by MPG's Collaborative Alliance to find ways of utilizing set-top data to buy networks whose audience and/or coverage are currently too small to be measured by Nielsen's national TV ratings sample. The initiative, which successfully developed a method for extrapolating the equivalent of Nielsen ratings from set-top data, initially analyzed three networks - Bloomberg, Sprout and INSP - and at least three agencies - MPG, Carat and OMD - said they were considering using the data as a way to plan, buy and post networks not currently measured by Nielsen.

Industry experts estimate there are upwards of 100 networks whose audiences currently are too small to be measured by Nielsen's national TV ratings panel. Those so-called "long-tail" networks, in aggregate, are estimated to represent about 7% of total national viewing.

While small in total numbers, many of those networks represent the kind of highly targeted audiences that many advertisers and brands actually want to reach, but before the Rentrak/Donovan deal, did not have a way of practically integrating into their planning and buying systems.

Rentrak has been one of the most aggressive of a group of digital set-top data aggregators to bring new audience measurement services to the ad industry. Others include WPP's Kantar unit, TRA Analytics, and TiVo. All say their systems have some superior attributes to Nielsen's methods, not the least of which is the size of their databases; the dynamic, real-time nature of the data; and the fact that they are based on actual "census"-level viewing, as opposed to a proxy sample of the TV universe.

Nielsen's national TV ratings sample currently is about 18,000 households. Rentrak's currently is based on 17 million.

Nielsen supporters say that set-top data systems cannot currently determine who is watching TV inside the homes they gather real-time viewing data from, and cannot generate reliable demographic estimates. Set-top data proponents say who cares, because demographics are crude, and possibly anachronistic ways of clustering actual viewing behavior.

That latter argument is gaining more momentum as big ad agencies begin integrating the way they analyze and plan digital media, especially online video, and some including Havas spin-off Adnetik and Interpublic's Cadreon, have already begun integrating ways of buying TV audiences into their electronic trading systems the way they buy online display and video advertising.

Both Rentrak and Donovan indicate that their deal is not mutually exclusive. Rentrak plans to integrate its data into other big enterprise systems such as Mediabank and Strata. Donovan plans to strike deals that would integrate other digital set-top databases into its systems and software.

To date, Nielsen which recently began trading as a public company again, has paid only lip service to developing a serious digital set-top data-based measurement service, and the company seems to have little incentive in doing so, as long as its panel-based samples are the currency of the $80 billion U.S. TV advertising marketplace.

Ultimately, Rentrak CEO Bill Livek says the real value of set-top data will be in developing new ways of integrating it with other powerful consumer behavior data - such as purchasing behavior or other lifestyle information that could be better and more precise indicators of who advertisers should be targeting than traditional age/sex demographic breaks.

Recommend (1) Print RSS
6 comments about "Rentrak Cracks Donovan: Deal Is First To Bring TV Set-Tops To Agency Desk-Tops".
  1. Tony Jarvis from Olympic Media Consultancy , February 23, 2011 at 9:58 a.m.

    Out-of-Home and the Internet have learned the hard way that only a very sophisticated, integrated audience measurement approach can even start to get close to producing a reasonable estimate for a contact or Eyes-On (OOH) or a page served (internet) respectively. Such a complex approach makes subsequent Reach and Frequency modeling extremely difficult but those models are improving.

    The two major but perfectly imperfect TV measurement currencies, Nielsen and Rentrak, which have major flaws clearly identified by each of their supporters per this article, have an opportunity to embrace a fully integrated approach in order to actually provide an estimate of "TV viewing". Such an approach will inevitably reflect both traditional sampling and set top box measurements but in the new multi-screen world will in addition critically require encoding and/or audio matching explcitly tied to a "viewer panel" via a personal meter in order to achieve the "real" measurement objective. Hopefully CIMM will be encouraging such a technique in every possible way. It would go a long way to eliminating the current practice of calculating "long tail" network audiences based on prototyping - yet another perfectly imperfect approach.

    BTW: Neither Nielsen nor Rentrak measure viewing they both measure set tuning!! When are we as industry fianlly going to admit that?

  2. John Grono from GAP Research , February 23, 2011 at 3:53 p.m.

    Tony I concur fully on OOH and online. Both industries here in Australia are using or looking at hybrid techniques as the 'one-size-fits-all' approach is no longer appropriate in a fragmented media world. Clearly TV also needs to go down this path - and rapidly.

    But I do have one question. A Nielsen home with only a set-meter is only measuring tuning. But a Nielsen home with a people-meter ... why do you say that is only measuring tuning and not viewing? Happily here in Australia we have only ever used people-meters for TV viewing ever since we went electronic in 1991.

  3. Joe Mandese from MediaPost , February 23, 2011 at 5:50 p.m.

    John,

    I don't want to speak for Tony, but I believe it's because people meters -- in the U.S., anyway -- really only measure a person's presence in the room (at the time they pushed their people meter button), and not that they were actually viewing television.

    Joe

  4. John Grono from GAP Research , February 23, 2011 at 6:15 p.m.

    Thanks Joe.

    I have worked in this area for more years than I care to admit to, and have seen many, many coincidental compliance studies involving thousands of panellists and spanning more than a decade and a half.

    With the compliance studies we telephone the panellists unannounced and at random and ask who is in the home, which TVs are on, who is watching them, and what channel/programme they are watching. We regularly find an overall compliance rate (i.e. pushing the buttons correctly) of 92%-94%. The most surprising thing is of that 6%-8%, MOST of the discrepancies are because the respondent nominates the right programme on the wrong channel - i.e. watching a spill-in broadcast signal or watching the programme on affiliate or Pay TV service.

    While it is an oft-quoted issue, in practical terms it is a non-sequitur at least here in Australia.

  5. Tony Jarvis from Olympic Media Consultancy , February 24, 2011 at 9:10 a.m.

    John:
    I think we are on the same page - sort of! If only meter compliance rates were really that good! They vary widely by "type" of home, composition and daypart/program type. Any water pressure study will confirm this notably during commercials. As I believe we agree, in this world of significant and increasing DVR usage, fast forwarding during commercials and multiple out of home video screens, both the buyers and sellers desparately need something far far better than either a "simple" survey of people meter homes - per Joe, merely presence in the room at a moment in time (what am I doing and where am I since I last pushed the button?!) or some agglomeration of set top box data.

    Sorry but it's not viewing! And that is not a non-sequitur at least here in the US!

  6. John Grono from GAP Research , February 24, 2011 at 6:23 p.m.

    Gees ... we Aussies must be a compliant bunch!

    I agree that 'average compliance' disguises the fact that some demos (the young) are less compliant than others and that heavy viewers are less compliant (the more you watch the greater chance you get it wrong) and that some household types (lots of people and lots of viewing options) are less compliant again because of probabilities.

    Still, for the first half of 2010 we were able to check compliance in almost 80% of the panel (maximum call attempts to avoid annoying the home) and found 91.8% compliance. Of the remaining non-compliant we found slightly more claiming to be a viewer but not logged in than claimed non-viewers but logged in. (Again the caveats that this is an All People average).

    The age-old argument of whether you are present in the room with the TV constitutes "viewing" was de-facto tested when Australia switched between 'actively watching' (i.e. panellist perception) and 'presence in the room'. You couldn't even see a blip on the ratings radar.

    This argument applies to all media as well in some form. As you would be well aware of, OOH uses the 200ms threshold for a fixation on a billboard. Readership mainly relies on recall of a front cover. Radio relies on recall captured by a diary. Online relies on the assumption that just because a page or ad was served then someone was at the other end viewing it (and ignores multiple browsers, tabs in focus, below the fold, auto-refresh, person away from computer etc). Comparatively TV is not in a bad place!

    Regarding fast-forwarding through commercials or programmes we use audio-matching here and it works neatly. Any 'trick-play' (FFWD, pause, Rewind etc) has no audio, therefore no match, therefore no ratings credit (and of course a viewer can still take in an ad message during fast forward).

    Despite all the above, TV ratings are clearly under pressure. This pressure comes mainly from the plethora of viewing options and distribution channels rather than capturing viewing behaviour.

    Also clearly a hybrid of back-channel data to establish more accurately the quantum of tuning, melded with behavioural data from a (smaller) panel will allow us to robustly equate tuning to viewing without having to attempt to enumerate the entire population. Our subscription TV services are well down the track with that project here in Australia, but it is still early days.

    I'd go so far as to say that hybrid modelling will enter the realm of all media:

    * back-channel data with a panel for TV
    * server-log data with a panel for online
    * melding circulation with sample-based readership studies a la JICREG UK for print/press
    * traffic data, footfall data, bus and train counts with sample based data for duplication and eye-tracking in OOH a la MOVE in Australia
    * ticket sales with sample data for cinema
    * ... but radio appears to still be more of a challenge

    Tony and Joe ... a great debate ... more of it I say!