How long have we been saying this? You cannot blindly take a TV ad and shove it into Facebook newsfeeds or YouTube and expect it to prove effective. People consume these channels differently and respond differently to the content they see in them. Marketers have to work with people’s mindset not against it.

I was reminded of the challenge of the need to tailor video content to the channel in which is shown when I reviewed a case study for Kantar Millward Brown’s new Link for Video. This variant of Link is explicitly designed for marketers who recognize that different channels require different approaches if a video is to be effective. Link for Video also includes some really neat enhancements like measuring implicit associations – the intuitive response to the ad – and behavioral measures like whether the video was viewed to the end.

The case study was not client commissioned but the test was conducted live last year. The test video was one for Smart ForFour titled ‘Reverse Parking’. Watch the ad through and I think you will find it pretty funny, but one of the key findings from the Link for Video study was that most people did not watch the video all the way through when the video was placed in Facebook’s newsfeed.

On Facebook about 20 percent of people do not engage with the video at all and a further 20 percent dropped out in the first 5 seconds. By the end only 24 percent remain watching the video (although this is the norm). Further, when viewing the video naturally 9 out of 10 watched with the video muted, further undermining its effectiveness. Which, of course, is one of the reasons why Facebook is now going to automatically enable sound unless you actively turn it off (guess what I’ll be doing).

By contrast, Reverse Parking performed really well as a YouTube preroll or TV ad. Facial coding reveals a huge spike in smiles just after the scream (even more so on second viewing as people focus in on the most engaging scene) and open ended questions collected using an instant messenger format revealed a positive response with plentiful use of smile, big smile and cheeky emoticons. The top three intuitive associations were cool, fun and comfort. Unfortunately agile was not a strong intuitive association, because the scream scene tends to dominate the scenes devoted to parking.

Overall, the likely impact on short-term sales and longer-term equity were strong for preroll and TV but far less so for Facebook, because the video is simply not well suited for that environment at the time of testing. Obviously that might change when sound is automatically enabled but the video might still need to be recut to deliver the scream scene sooner. Particularly when viewed on mobiles, video works most effectively when attention is earned early and the brand association comes within the first 15 seconds.