Sunday, September 10, 2017

Is it smart to use data from smart TVs?

Multichannel News has an article (It’s Time to Get the Return-Path Data Together) by Jane Clarke of CIMM on the complementary nature of audience data from STBs (set-top boxes) and smart TVs piped through ACR (automated content recognition). Generally, it’s a good idea, but some significant qualifiers come to mind:
  • VOD can be measured through STB data collection, and more than one data collection platform already supports this, if implemented correctly. However, it is unclear what any measurement on the playback device might add to proper design of the VOD system (such that all play and trickplay, and not just the original order for the programme, is reported to the server farm) and measuring from there. A smart TV, though, is going to capture some OTT traffic from devices connected using the likes of Google Chormecast, Amazon Fire TV and Roku, but unless measurement on the playback device or at the server farm is available, this will be a partial accounting—without any means of determining how partial; the proverbial little knowledge that is dangerous.
  • Likewise, the power state of the monitor, not currently available from an STB, would pertain to an unknown percentage of viewing (unless matched to same-STB data) and not be very usable for capping viewing reported by STBs. It would be far better if the STB-based data collection systems were enhanced to poll the monitor power state over HDMI, which will now be the default connection to UHD/4k as well as HD monitors. Then, as SD diminishes to zero over the coming years (faster in some countries than others), we would have real data in most cases in which currently we must use statistical approximation.
  • All smart TV ACR can reliably provide is the programme identity, and quite likely not in a format relatable to pay-TV operations (as there is little chance of a common reliable identifier; such an animal could exist but licensing policies are a big obstacle). Programmes alone are not sufficient in the present environment, in which media use is still largely organised around channels and rights flow through them as well.
  • Following Vizio’s comeuppance in court in the U.S. for undisclosed snooping, this practice has gained potential to become a slow meme, with even Consumer Reports explaining how to get rid of it. Especially with help from data-protection-sensitive Europe, this might become a common concern a little like the falsehood, often repeated in a certain genre of fiction, that a powered-off cellphone with a charged battery in place can be used to determine the location of its user. How much opt-out from measurement would render smart TVs not worth the trouble?

Internet advertising is being questioned, and some questions have no clear answers

Just as Internet advertising reached spending parity with television, large advertisers started doubting its effectiveness and cutting back, writes Nicole Sinclair in Yahoo Finance (Digital ads aren't working for big consumer brands). She lists two developments last year: a study that claimed the existence of rebates undisclosed to advertisers (kickbacks under another name) from media operators to advertising and media buying agencies (although not specific to Internet media), and Facebook’s admission that it included video views of under three seconds, exaggerating the overall viewing it reported by potentially as much as 80% (whatever that means). The article cites reductions in Internet spending by large consumer goods advertisers of, at most, 1.3% this year. Not much, perhaps, but the direction should give pause.

Individual advertising delivery cannot be measured by sample, but only by a census. This currently can only be self-administered by the carrier, and that has a credibility problem inherently, not just because of abuses. When chief executives of the half-dozen remaining global ad and media buying agencies, like Martin Sorrell of WPP, say that “the player and referee cannot be the same person” and the media operators should not “mark their own homework” (a phrase heard a lot lately), their companies presumably cannot then initiate the spending of large portions of clients’ budgets on such media. Can targeted digital advertising survive the lack of objective verification, never mind transparency of targeting decisions?

Furthermore, even objective reporting and realistic standards (unlike Facebook’s deeming of a video view any exposure longer than three seconds) might not rescue targeted advertising on the Internet. The current crop of targeting algorithms is rather obviously useless, with much irrelevance and ad nauseum repetitiveness of ads (often from an inappropriate competitor) for purchases already made. This is so despite cookies and tracking by the likes of Facebook. On Facebook itself, the problem is different—the complete irrelevance of most advertising messages—but it still means inappropriate expenditure in which advertisers pay high CPMs for targeting but get, at best, scattershot outdoor billboard delivery. Can targeting be seriously improved in a short time?