Cadillac claims highest-ever audience reach during Super Bowl XLVI, but what does this mean?


This news almost had me choking on my Cornflakes this morning. Having churned through our own Super Bowl audience data during the past few days, admittedly with a focus on online behaviour, we had a pretty good idea which automotive campaigns had really ‘resonated’ with the viewing audience.

So when Cadillac claimed it was fairest of them all, we had to take a closer look.

Cadillac’s 30-second preview of the upcoming 2013 ATS compact sport saloon during last Sunday’s Super Bowl XLVI broadcast, was apparently the most-viewed single advertising spot in U.S. television history. (With Super Bowl XVLI reaching a record U.S. television audience of 111.3 million viewers)

The data used by Cadillac was provided by Kantar Media, a well respected audience insight company, using return path data (RPD) sampled passively through the set-top boxes of an opted-in audience. The sample size used was 100,000 (of the 111.3 million audience) and Kantar Media then sample a behaviour they call ‘TuneAway’, which is essentially when a household changes channel to avoid a commercial.

The lower the TuneAway, the more successful the ad or programme. In theory.

A large proportion of investments made in technology companies at the moment are in the field of audience research and response, mostly to optimise online spend or manage customer relationships on 3rd party platforms such as Facebook. Each try to promise a ‘silver bullet’ solution (one dashboard for all needs) and tend to focus on the interpreted data rather than the method (and its assumptions) used in producing the data.

The problem though with all TV audience data is that it measures the behaviour of a household, rather than a person. A household could be a group of mates sitting down to watch the game, a family dictated to by a dominant parent or even a busy social gathering where the TV is switched on in the background.

Additionally, just like many online traffic measurements (such as Alexa or Comscore) the audience panels providing the data must first be opted in, which is often achieved via incentives (free anti-virus software, of gift vouchers). This further skews the data towards those people who look out for free-gift offers.

Cadillac’s spot, called “Green Hell,” aired at 9:37 Eastern Time, during the game’s “two-minute warning” official timeout, so audiences were literally on the edge of their seats. How many of them stayed there, rather than visiting the loo or fetching another beer is anyone’s guess.

Whilst Cadillac may have scored highest (43.46 for their 30 second spot) in the 100,000 household sample, they are currently only 47th in the Viral Video Chart with a little over 536,000 views and 1,989 shares.

This compares with Volkswagen’s 23 million views and 942,943 shares and Honda with 19.3 million and 360,900 shares, so you can see why I almost choked on my breakfast this morning. All of these are of course online viewing figures, but they likewise portray the behavior of individual people, each of whom have ‘chosen’ to click on a video rather than passively viewing it during a live broadcast.

The lesson? Don’t believe everything you read and make sure you understand how behavioural data has been compiled and the context that it refers to.

It’s good to see Cadillac embracing the Nurburgring so strongly during development of their ATS sports saloon, I just hope they’ve also inherited some of their Germanic competitor’s understatement when it comes to the performance of the car.