So Much For Conventional Wisdom, Study Finds Consumers Connecting Longer On Long-Form Connected TV Ads

by Joe Mandese , June 26, 2019

In a finding suggesting the shift to shorter-form video ad units may not apply to connected TV (CTV), a new quarterly tracking study finds consumers watch more of longer-format commercials via connected TV sets.

The average time spent with ads on connected TVs (CTV) has moved up slightly for 30-second units, but is flat for 15-second units, according to an analysis of first-quarter 2019 data from ad-tech firm Extreme Reach.

The analysis found that the average CTV viewers spent an average of 26 seconds, or 87%, of 30-second units. That's up one second -- or 83% -- of the average 30-second unit -- in the first quarter of 2018.

That compares with only 12 seconds, or 80%, of the average 15-second unit in the first quarter of his year, which is the same duration for :15s in the first quarter of 2018.

The average time spent viewing findings correlate with video completion rates for CTV ad units of various lengths, with :30s showing the greatest year-over-year improvement, up to 91% in the first quarter of 2019 from 86% in the first quarter of 2018.

"While this may seem counter intuitive, the popularity of 30-second ads on CTV, a TV-like environment where viewers do not have the option to skip, is responsible for driving that number up," the report notes, adding that video completion rates for 15-second ads also increased slightly, while completion rates for 6-second units are trending downward.

advertising, connected tv, digital, research, television

5 comments about "So Much For Conventional Wisdom, Study Finds Consumers Connecting Longer On Long-Form Connected TV Ads".

Check to receive email when comments are posted.

Ed Papazian from Media Dynamics Inc, June 26, 2019 at 12:10 p.m.
Joe, I went to their website and tried to find an explanation of how they determined whether a consumer was "watching" any of these commercials---but I couldn't find it. Maybe it's there, somewhere. I assume that they are doing the usual thing, namely equating the ad being on-screen with "watching", in which case, the findings really need to be taken with several large grains of salt.
Reply

Joe Mandese from MediaPost Inc., June 26, 2019 at 12:32 p.m.

@Ed Papazian: That's probably my bad. Extreme Reach used the term "time spent" for duration, and I don't think they were measuring cognition, just the duration of time the ads were present on the TV screen.

In terms of video completion rates, they did used the words "viewers" and "watched," but again I don't think they were measuring cognition, just that the ads completed on the screen or not.

Here's there own language:

"We believe the reason CTV consistently outperforms the other devices in this metric is that viewers tend to be committed to the content they’ve chosen to watch, don’t seem to be interested in clicking around to bide time while waiting for the program to return and typically don’t have the option to skip the ads."

John Grono from GAP Research, June 26, 2019 at 6:30 p.m.

I suspect that a few things are at play in the data - leaving aside Ed's correct question about what the measurement actually is.

The first is that Q1 appears to tend to have higher 'duration percentages' - which could be one of the many patterns seen in TV data such as higher viewing levels during winter and lower in summer when you are out-and-about more. So this may not be a trend but seasonality at play.

The second is that the measurement appears to be rounded to an integer. So moving from 25 seconds to 26 seconds may not be the +4% in duration that it appears to be. It might have been 25.4 seconds to 25.6 seconds - a +0.8% increase. Conversely it might have been 24.6 seconds to 26.4 seconds - a +7.3% increase.

The third is that this may just be sampling error. Did Extreme Reach sample within their CTV pool or use a census based approach. If it was a sample was it a random sample (i.e. pot luck) or was it a longitudinal sample (i.e. same homes over time).

It is very hard to analyse and draw any conclusions without knowing how the metric is recorded (thank you for the follow-up Joe) or the precision of the reported data.

jen hahs from essence, June 27, 2019 at 1:47 p.m.

Thank you! Evaluating whether to flag this study for colleagues (it is making the rounds) and hoped that Joe and commentators had taken a look.

Alex Kanner from Sublime Inc. replied, June 29, 2019 at 4:47 a.m.

Completely agree with Ed Papazian here. It would be greatly appreciated if that was in the article.