Cutting to the Chase on Set Top Box Data - Bill Harvey - MediaBizBloggers

It seems everybody is talking about Set Top Box (STB) data nowadays. Last week Nielsen put out not one but two articles focusing on STB, both in MediaPost.

Manish Bhatia, who runs Nielsen's STB program, pointed to the fact that if STB data were used today as the measure of audience size, cable networks would inherit an extra $3.1 billion a year in ad revenues, while broadcast networks would lose $1.1 billion (he did not spell out where the other $2 billion going to cable would come from).

Ed DeNicola, whom MediaPostindicates gives "STB 101 presentations to media companies and advertisers," talks about how STB research "is difficult" and "hard to calculate" and calls for a "more open dialog." (Those of us who have been chopping the first path bringing ARF standards to STB for over ten years are tempted to feel the sin of pride at that. It sure was hard at first! Now when somebody talks to me about 'more open dialog' I hear the words 'let me pick your brains.')

A researcher with broad and deep industry experience responded by sending me a note: "There is something for your next blog---how to keep the industry from progressing! If the industry can't work their way through these issues then they may as well fold up and go home…this is like not moving from LPs to DVDs because you have to buy new equipment – or not moving from Sami Warehouse Withdrawal data to store level POS because it is not census…these types of mindsets make me crazy!!"

Frankly there is nothing in the two Nielsen articles that I have not been saying myself for over ten years. It is perhaps the way these things get said that cause confusion in the marketplace. So I decided to address the Nielsen articles in my blog, not by way of refutation or complaint but simply to keep minds clear as we all approach a future that will be increasingly heavily based on STB.

In a 2002 ESOMAR/ARF paper with Tony Jarvis and Russ Booth entitled "Better Television Audience Measurement Through Cable and Satellite Set Top Boxes," we wrote under a subhead called "Universe" in a section devoted to the challenges of STB: "All TV audience measurement is imperfect and the real question is which set of methods is least imperfect. The authors suggest that the use of set top box data with the appropriate edit rules, adjustment factors, and demographic questionnaires, in conjunction with a method for attaching people data, offers the least imperfect method available, largely because of the sample size and second-by-second measurement aspects."

Nielsen itself acknowledges that the same flaws it continuously cites regarding STB can be dealt with by means of processing adjustments; as Manish says in his article, "We view [set-top box] data as an extremely valuable source of information that, once adjusted for inaccuracies and gaps, can provide deeper insights on TV viewing, advertising delivery and effectiveness." (Italics added.)

Manish's dramatic point about the shifting of billions of dollars stems from the fact that cable (including AT&T and Verizon) and satellite digital set top boxes (DSTBs) are the only means by which Nielsen, TNS and Rentrak get their STB data. DSTB homes also, by definition, all get hundreds of channels, most of which are cable channels. The broadcast shares in these homes are lower than broadcast shares in basic cable homes and much lower than broadcast shares in over-the-air (OTA) homes. Hence if these cable/satellite DSTB homes were to be used as the audience size measure, audiences and dollars would shift from broadcast to cable just as Manish says.

By contrast, TRA and TiVo differ from Nielsen, TNS and Rentrak in that TRA and TiVo measure all types of homes, including those that do not have cable/satellite DSTBs and therefore do not receive hundreds of channels. In addition to measuring hundreds of thousands of DSTB homes (the ones that receive hundreds of channels), TRA for example measures hundreds of thousands of basic cable homes (that generally receive fewer than 100 channels), and has more OTA homes (that receive a handful of channels) than the entire Nielsen national sample size. But that is not the main point, which is as follows.

The articles ignore the differences between TRA and all other companies in the ways in which STB data are used. This would perhaps not be worth mentioning except that Joe Mandese identifies TRA at the end of Manish's article along with TNS as two companies that have already learned how to process STB and are commercializing STB data in the marketplace.

There is a difference between the way TRA uses STB and the way everyone else has been using it. It is important to recognize that profound difference because nothing said in either Nielsen article has any bearing whatsoever on the way TRA uses STB data today.

TRA does not use STB data as a measure of audience size (although our system certainly allows the user to do so if desired). The shifts of vast amounts of money to cable away from broadcast that Manish says would occur if STB were the size currency do not occur with TRA. TRA assumes that its users (see my prior blog posting for example) are using Nielsen for audience size measures. What TRA measures is the probability that the audience reached by a given program is the right viewer (a.k.a. purchaser) set that the brand ought to be targeting for maximum ROI. We are measuring composition within the audience, not the absolute audience size. We consider this a valuable use of STB data that can be made today with none of the downside consequences Nielsen points out for leveraging STB for audience sizes.

Peeling the onion a little farther for those who have not read the prior blog posting: Today's planning/buying practice is to cross-analyze two variables: cost and the number of average minute viewers in a sex/age group. TRA's recommendation is to add a third number that gets weighted into the value rankings of TV programs for a given brand. The third number is the likelihood that these viewers are actually the kinds of viewers (a.k.a. purchasers) the brand wants to reach.

Because networks price inventory based on number of 18-49 or 25-54 viewers, the index of value equation comes out where all TV programs are very close together in value (Cost Per Thousand viewers). TRA's purchaser likelihood weighted CPM suddenly shows huge differences in real value to the brand across TV program options, and opens up the ability to increase the ROI-producing power of the same TV budget.

For example, let's say there are two programs, Program A and Program B, and they each have an average audience of 1 million viewers who are W25-54. Let's say one costs a little more than the other so that Program A has a CPM W25-54 of $20 and program B's corresponding CPM is $22. The practices of the last half century say pick Program A.

Now what if you had a way of analyzing what these audiences buy, and found that the type of purchasers you want to reach are 30% of Program B's audience and 20% of Program A's audience? Then weighted by this highly relevant measure of value and predictor of ROI, Program B would have a purchaser-weighted CPM of $73.33 and for Program A the corresponding CPM would be $100.00. Now we can see that Program B is the better choice.

That is the way its customers are using TRA today, plus post evaluation of ROI by subsets of total TV that offer planning strategy/tactical options (different creative, brand integration of various types, program types, frequency levels, recency related analytics, and much more).

Since it is not aimed at being a measure of audience size, TRA steps aside from the fray churned up by the Nielsen articles. That is a debate among those using STB to replace Nielsen's current methods today.

As a final note we would be remiss if we did not mention that the overall correlation between TRA STB data and Nielsen is close to .9. This is a result of TRA weighting/projection, edit rules, and the overall processing algorithm regime we have learned over the past ten years (and exclusively licensed to TRA), to address the challenges that others are just beginning to face right now.

Bill Harvey has spent over 35 years leading the way in the area of media research with special emphasis on the New Media. Bill can be contacted at bill@traglobal.com.

Read all Bill’s MediaBizBloggers commentaries at Bill Harvey - MediaBizBloggers.