There Is No Magic Co-Viewing Factor On Connected TV

"On TV and Video" is a column exploring opportunities and challenges in advanced TV and video.

Today’s column is written by Maggie Zhang, executive vice president of video research and insights at Dentsu Aegis Network.

One of the value propositions of over-the-top (OTT) and connected TV (CTV) advertising is its TV-like co-viewing activity. Watching video content and ads associated with it on the big screen, regardless of how it’s delivered, can be inherently a communal experience.

IAB research has proven that TV is the most social viewing device with 96% of viewers watching linear TV with others, and 93% co-viewing on connected TV.

But is there an OTT co-viewing factor?

Co-viewing on CTV inevitably raises an important question of CTV impression counting and measurement. CTV measurement is digital in nature and at the impression level. That's different from Nielsen’s linear TV measurement, which is at the person level and inherently inclusive of the co-viewing effect. A conversion is required to translate served CTV impressions to audience impressions while accounting for co-viewing.

For the past few years, Nielsen has worked with Roku and Hulu to develop its OTT measurement methodology to enable audience assignment, including co-viewing, within their respective footprints at the total audience (P2+) and target demo levels. Since then, some advertisers have started to embrace the benefits of audience extension and engagement that co-viewing facilitates, and they’ve agreed to transact on co-viewing inclusive CTV impressions with Roku and Hulu.

However, other OTT publishers also started to tout a co-viewing factor of 1.2 and propose that this co-viewing factor be applied to all served CTV impressions.

Is such a number valid? Is it in the advertiser’s best interest to apply such a factor to all CTV impressions? How can advertisers protect themselves to ensure fair transactions?

Some investigation of the magic factor

We did some digging into the magic 1.2 co-viewing factor by leveraging Nielsen’s OTT Digital Ad Ratings. As it turns out, that magic number appears to be nothing but magic at this point. It is an average proxy of CTV co-viewing based on total audience (P2+). Marketers should not currently accept the approach of applying a static co-viewing factor from measurable assets to other unmeasurable assets for a few reasons:

1. Co-viewing calculation is dynamic, not static. Although the general concept of co-viewing is intuitive, the extent to which co-viewing boosts audience reach can widely vary by target audience, content genre, time of day or day of the week. Streaming an episode of romantic comedy in the middle of the day may not have the same co-viewing level of watching a feature film after dinner. The demo efficiency rate (co-viewing factor) is a dynamic metric and shouldn’t be reduced to a static number. For multiple campaigns we examined, the same publishers have different co-viewing demo efficiency rates.

2. Co-viewing incrementality of the total audience does not linearly translate to the target audience. Keeping the static issue aside, this magic co-viewing factor is largely at the total audience level (P2+). In most cases, marketers need precise measurement of their target demo audience to facilitate transactions. The analysis suggests that the demo efficiency rate between total audience and target audience varies across publishers and does not necessarily have a linear relationship. That means one publisher can drive high incrementality at the total audience level but can be insufficient in driving target demo audience extension.

3. Co-viewing effect within unmeasurable assets is unsubstantiated. And to throw another wrench, Nielsen’s OTT DAR measurement is only integrated with Hulu, Roku and Amazon DSP as of today. For an OTT publisher or network’s full episode player, a considerable amount of CTV impressions on other streaming devices and platforms are not measurable or verifiable for co-viewing with empirical and quantifiable evidence.

Path forward

With the rapid growth of CTV penetration and usage in US households, we are going to see more stable and consistent viewing behavior on CTV, including co-viewing. As marketers continue to embrace CTV advertising and glean the benefits of co-viewing, they also need to apply critical thinking and avoid any potential misunderstanding of a complex issue at their expense due to current measurement limitations.

Marketers need to take a discriminant approach in CTV measurement and transactions. Co-viewing should be measured and validated by third-party measurement vendors, with the increased audience impressions accounted for. Otherwise, marketers should continue treating CTV impressions as one-to-one digital impressions without applying proxies or assumptions.

Fundamentally, co-viewing measurement on CTV is just one aspect of audience validation that marketers urgently need. The analysis was for demo audiences, but similar consideration should be extended to advanced audiences. More empirical research and industry norms are needed to better understand the relationship between co-viewing and audiences by publisher and platform.

Many promising initiatives underway are leveraging innovative data sets and machine learning techniques to find the right solution. CTV measurement will continue to evolve where all CTV impressions can be dynamically and fairly converted to audience impressions, inclusive of co-viewing, in the foreseeable future.

Follow Dentsu Aegis Network (@dentsuaegis) and AdExchanger (@adexchanger) on Twitter.

Enjoying this content?

Sign up to be an AdExchanger Member today and get unlimited access to articles like this, plus proprietary data and research, conference discounts, on-demand access to event content, and more!

Join Today!

1 Comment

  1. I am curious if you could elaborate further on the specific Machine Learning techniques being leveraged, along with the datasets. Any references to recently published papers are greatly appreciated. Also, just to be clear on terminology, what does "demo" convey? [demo audience, demo efficiency rate, etc]? (I am assuming demographic, but would like to be clear).

    Reply

Add a comment

XHTML: You can use these tags: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>