Sofa-driven insights: Unlocking the joy of co-viewing audiences on CTV

Picture the typical living room. Front and centre will be the TV, surrounded by a nest of furniture where family and friends can gather round. The communal viewing experience has shaped how we lay out our homes, determined programming schedules, and fuelled the evergreen “home cinema” market. However, settling onto the sofa with loved ones to watch the latest big show is more than a cosy way to spend an evening; it’s the site of an industry concern that keeps media planners and platform owners up at night.

Co-viewing - where multiple individuals watch the same content together on the same device - has been both a desired and contested metric in TV advertising for decades. Today, there are an average of 1.37 people in front of the TV during broadcast, rising to 1.49 in the evenings. Fail to account for these figures, and we get a skewed and inaccurate picture of what’s happening on the other side of the screen, undermining the value of advertising for buyers and sellers alike.
 

Why is co-viewing worth measuring?

The need for co-viewing measurement is clear for media owners, broadcasters, and platforms: if multiple people are exposed to an ad, they don’t want to be paid for just one impression. If the value of TV is its communal experience, the sell-side wants to see this value reflected in the price of its inventory.

Advertisers, meanwhile, want to be able to target ads based on who is behind the screen, what they’re watching, and their viewing environment. A single person who has a TV quiz show on as background noise while they cook, is going to have a very different ad experience to a family squashed on a sofa to watch the latest Disney+ movie, or a group of friends gathered to watch live sports.

The co-viewing factor for linear TV is traditionally applied as a flat rate, with an average applied across all inventory to account for the added value of multiple impressions in a single view. But a flat rate is less appropriate in today’s data-driven ecosystem where buyers expect far more precision, and don’t want to spend on impressions that weren’t delivered.
 

Is there demand for co-viewing measurement?

Certain genres, shows, or channels naturally attract more co-viewing audiences. While binge-watching and catch-up services dominate most modern viewing habits, major events such as sporting finals, global ceremonies, and must-see TV shows still command live, communal viewing. These events not only draw massive audiences but also attract premium ad spend, with just 30 seconds of ad time at the Super Bowl earlier this year costing a pretty $7 million.

2024 is a huge year for tentpole TV, with major events like the UEFA European Football Championship and the Paris Olympics on the horizon. Advertisers are eager to ensure their campaigns reach the right audiences and capture the wider range of demographics drawn to cultural moments.

Demand for knowing who is on the other side of the screen will only increase as more ad budgets are funnelled towards the rapidly digitising TV space, and as video campaigns become increasingly omnichannel in their approach, necessitating accurate and comparable per-channel measurement.

This demand will bleed from sports finals and prestige TV to day-to-day TV watching, where co-viewing is also prevalent. The TV is often a focal point for the home, with families gathering and dispersing, creating different configurations of viewers throughout the day. So, how can we keep track of who is actually watching?
 

What are the challenges to calculating the co-viewing factor?     
                                     
Even with the wealth of data at the fingertips of the CTV ecosystem, accurately measuring co-viewing remains a challenge. As much data as there may be, access to it has become increasingly restricted due to privacy regulations limiting what personally identifiable data can be used, how it can be used, and the consent required from the user to do so.

This data is also scattered across various silos, as the proliferation of channels and platforms fragments audiences and the consistency of the buying experience as different sellers might have different methodologies for measurement. This audience fragmentation makes it harder to track who is watching and where, posing a dilemma not just for buyers but also for broadcasters and media companies who rely on this data for content development and scheduling.

The definition of TV has also broadened, and we can’t assume that co-viewing exclusively occurs in front of a big screen. A couple watching their favourite show in bed on their iPad before going to sleep would be counted as a single viewer under the one device, one user logic. Log-ins aren’t a reliance proxy either, due to account sharing and the likelihood the person watching isn’t the one paying the bill.
 

What approaches to co-viewing measurement are being explored today?

Traditionally, TV audiences have been measured through audience panels such as BARB in the UK and Nielsen in the US. However, with the evolution of the TV landscape, measuring audiences accurately has proved a frustration for marketers who are accustomed to the numerous measurement tools of digital platforms, and for publishers being underpaid for the impressions they’re delivering.

Efforts have been made to bridge this gap, such as Barb’s hybrid approach integrating panel data with online viewing data - which has been adopted by streaming giants such as Disney+ and Netflix. Barb is also testing CFlight, a unified measurement tool for the industry. Google has revealed its own plans for co-viewing measurement on YouTube, although it has faced pushback from agencies dissatisfied with the lack of transparent methodology.
 

How can on-device measurement support precise co-viewing estimates?

As the industry scrambles to piece together data to reveal co-viewing habits, there is one method that stands above the rest when it comes to efficacy: on-device measurement. The devices used to stream TV - from set-top boxes to smart TVs, consoles, and tablets - generate insights on user behaviour and content preferences that can reveal a great deal about who is watching, what they are watching, and how.

All processing is carried out on the device itself, ensuring that all buyers see are anonymised profiles and attributes that can’t be linked to a single person - or people, in the case of co-viewing. 

By calculating how various on-device profiles align with co-viewing factors as determined by panels and third-party socio-demographic classification bodies, a more accurate, real-time estimate for co-viewing can be inserted into the bidstream. 

At a time when we need to be more cautious about what data we collect and how we use it, on-device measurement provides a data source that has enormous privacy advantages. By smartly layering its insights with the wider world of consenting panels and third-party sources, we can deliver digital-like accuracy and speed on CTV even for the most challenging metrics such as co-viewing. 

For the sell side, this provides the opportunity to monetise inventory based on its true value and provide better cross-platform targeting. Buyers, meanwhile, get access to high-quality insights, which makes measurement more accurate, and gives them the ability to tailor ad placements to specific, highly-engaged audiences.