Commentary

Moving Beyond Branding: Holding TV Accountable For ROI

by Boaz Cohen , Op-Ed Contributor, October 13, 2015

Last week, Facebook announced a product called TRP Buying, a way for TV buyers to coordinate TV efforts with video ads for more effective measurement. Why? Because branding, traditionally the purpose of TV commercials, simply doesn’t cut it in a world full of ad optimization and digital data.

Advertisers want and need results that can be tied directly to ROI and sales. As video budgets continue to shift from TV to online, it’s about time we hold TV accountable for actual, standalone performance. But how?

In 2005, Mark Lieberman and Bill Harvey started TRA, now TiVo Research, with the promise of deterministic data based on Set Top Box (STB) information for ROI measurement. Rentrak took it to the next level with data on almost 25% of TV households. Both ventures enabled better targeting and measurement on TV using data.

However, the disconnect between the data providers and the TV buying systems coupled with the complexity of the TV delivery system – which still requires manual work, excel sheets, and fax orders – limited the ability of utilizing data for TV advertising to big brands that have the money to spend heavily on TV.

Ten years later, ROI for TV campaigns is still clearly lagging its digital video counterpart. TV advertisers that want to go beyond GRPs and measure the true ROI of their campaigns are limited to test-and-control markets, or one-off projects limited by match rates with STB data providers. This, in turn, prevents real-time campaign optimization—light years behind digital campaigns, which can render thousands of versions of videos to drive results.

Thankfully, the TV industry has started to innovate in an effort to maintain its piece of the advertising pie. Two competing technologies have emerged that can help bring TV ad buying in line with digital video buying capabilities: addressable TV and programmatic TV. Michael Kubin, EVP of Invidi, expertly described the differences between the two. While each has its own limitations, they both get TV closer to digital video buying.

While allowing for the most granular – household level – targeting and measurement, addressable TV suffers from limited coverage of about 25% of US HHs and still requires a manual buying process. The efficiency of programmatic TV led agencies to adopt programmatic buying, and they will do so in TV.

Worried that programmatic will drive prices to the bottom, the seller side put limitations on programmatic TV that prevent it from reaching its full potential.

Both technologies will improve in the next couple of years: The scale of addressable TV will increase to cover most MSOs, and the buying process will become more programmatic. Simultaneously, programmatic TV will integrate more data and improve targeting resolution, and TV sellers will learn to take advantage – similar to what happened in the digital world.

ROI from TV ad spend can be scaled by combining all of the TV advertising solutions: addressable, programmatic, and connected, with data and measurement solutions. This combination can help advertisers get beyond reach and frequency and start seeing real return on their investment—it’s about time!

advertising, roi, tv

5 comments about "Moving Beyond Branding: Holding TV Accountable For ROI".

Check to receive email when comments are posted.

Ed Papazian from Media Dynamics, October 13, 2015 at 8:40 a.m.
Reading this, it would seem that the only way that TV advertisers evaluate the impact of their campaigns is via reach and frequency estimates and, by GRPs. Actaully, this is not the case, as such "metrics" merely define the opportunity for ads to be seen and how often this might happen---not the effectiveness of the campaigns. Most TV advertisers routinely pre-test the communicative power of their commercials---though new measurement techniques in this area may be called for. They also track the recall and sales moitivational effects of their commercials once a canpaign begins as well as the over-all ad awareness and sales point registration. Finally, many monitor how their sales are responding on an ongoing basis.

We should also bear in mind that direct cause-and-effect sales responses are not always the prime objective of TV ad campaigns. Many seek to sell an "image" for a brand or a corporate entity. Others are mainly supportive of local advertising by dealers, franchises, store chains, etc.

While it sounds simple enough, deducing the "ROI" of TV ad spending is a very complex matter, with many interacting influences and dynamics entering into the equation. It's not as easy as sending an ad to a "targeted" viewer---though current "addressabe" systems don't even know if said viewer is watching---then checking to see if the viewer, or his household, bought the product on the next visit to a store---assuming that such data is available.
Reply

John Grono from GAP Research, October 13, 2015 at 6:52 p.m.

Just a thought. 25% of HH is a pretty big whomp of the US households.

I wonder if anyone has done a comparison of the HH ratings when all 25% are used, against maybe 10%, 2.5%, 1% etc. Sample sizes have to grow exponentially in order to reduce sample error. At some stage there is really very little point of utilising more data if you end up producing what is substantially the same ratings.

Ed Papazian from Media Dynamics Inc, October 13, 2015 at 7:39 p.m.

John, in times long past, when TV household diaries were the common local market TV rating measurement while meters were what was used nationally, we had many opportunities to compare the aggregate ratings from all local markets for network shows with what the meters were saying about the entire country. At the time, there was no cable, no Internet, no Netflix, etc. and it was felt that the diaries were quite good indicators of TV set usage, which was dominated by the three TV networks and/or their affiliates. The results on a show by show basis, in all dayparts, were almost universally the same with comparative results aggregated for well over 100,000 local diaries comapred to a mere 1,000, metered homes nationally. In other words an increase of 100 to 1 in sample size did not have much of an impact on rating "accuracy".

It must be noted that things are quite differernt today, with so many channels to choose from and so much viewing being done with a single solitary viewer in attendance. Indeed, the typical TV show's avertage telecast rating---counting cable---is probably one twentieth the size of the old norms---maybe smaller.

I believe that the main advantage of the "big data" panels---assuming that the right adjustments are made to make the findings more representative of all TV homes, not just cable homes, is not their overall accuracy. I expect that you will get almost the same ratings for most shows on a national basis with millions of homes supplying data compared to 20,000-25,000. However, the big data panels will be able to supply statistically reliable data for almost any show, even those with tiny audiences, plus they can give you "granular" data by small, selective demographic segments--providing that these are properly identified by the cable companies who allow their set-top-box information to be used. This is certainly true of geographic subsets, which are the cornerstone of most "addressable TV " systems. Whether such systems can distinguish between set usage and "viewing" is another matter.

John Grono from GAP Research, October 14, 2015 at 3:46 p.m.

I agree with your analysis, apart from the 100 to 1, as the diaries were for a single week whereas the meters are continuous longitudinal data capture and reporting, but it is probably 10x or thereabouts.

When Australia switched from diaries to meters in 1991 we saw an average uplift of just under 10%, so I agree that diaries did a better job than most people gave them credit for.

Here in Australia we have around 5,500 metered homes (population 24m). But we have also had an interesting development with our subscription TV homes (penetration around 30% mainly via Foxtel with around 200 channels).

Their sales operation (MCN) developed a system called MultiView as a sales tool. From the outset I must stress that this is NOT a currency and does not claim to be one - it is an adjunct. They get return-path-data from 110,000 subscribing homes (universe of around 2.5m homes) and generate daily 'HH ratings' for all channels, The advantage of the large sample is less 'bounce' from day to day with those estimates. They then apply the panel based people-data from the OzTAM currency to these 'small channel' estimates to give them some base to show the likely audiences. Of course, aggregating these smaller channels not only produces a larger 'pot' of the audience, but also is good news for incremental reach (albeit in small increments, but with a smaller price tag). I see great hope in this approach as distribution and audiences continue to fragment.

But the interesting thing is that this service started with 10,000 RPD homes to which an incremental 100,000 was added a couple of years ago. The gains in stability and accuracy were made with the initial service, rather than than the 11x sample size.

Ed Papazian from Media Dynamics Inc, October 14, 2015 at 3:57 p.m.

Interesting, John. By the way, the meters invariably find more tuning to lower rated shows, rerun fare, movies and the like ---much of it being not loyally viewed content, hence less easily remembered by a diary keeper. When it comes to more popular shows, which viewers watch with greater regularity, the differences between diaries and meters, where household tune-in ratings are concerned, are the smallest. This has implications regarding viewer involvement and "engagement", suggesting that the added audiences noted by the meters, may not be as valuable in terms of attentiveness, as those found by both methodologies. Just a thought.