Measuring TV Attention: Presence, Impressions and Person-Level

Prior to joining TVision as Chief Product Officer, I spent the past decade working to establish digital attention standards at Google, IBM and Integral Ad Science. It took hard work and years of tinkering, but as an industry, we eventually arrived at a standardized measurement that allows advertisers and publishers to mark whether the consumer had the opportunity to see a given ad.

In television, this work is only just getting started.

For years, the industry operated under the assumption that TV is 100% viewable — a pretense that is simply not true. After all, viewers don’t have the opportunity to see an ad if they’re in another room when it airs, as is the case for a full 29% of TV commercials. These non-viewable ads are costing billions in wasted spend each year.

Fortunately, the TV industry, led by the MRC, is now paying attention to this under-measured aspect of TV. Brand advertisers, agencies, media sellers, technology companies and industry bodies have started mobilizing to address this problem. Beyond reducing waste and increasing return on ad spend, stakeholders are motivated by shifting media consumption habits and the rise of over-the-top (OTT) television. As these trends continue to blur the lines between linear TV and digital video, it is only becoming more urgent that we develop unified, person-based TV attention standard that measures when ads are shown to people, and when they are not.

While it’s encouraging that some in our space have begun addressing this crucial issue, for the metrics to deliver real value to advertisers, it’s important that they help advertisers establish to what degree audiences really have the opportunity to see their ads.

At TVision, we believe that a successful TV Attention standard must be able to measure three key elements:

  • The presence of a viewer in the room while the ad airs.
  • The individual impression that’s being assessed.
  • Who is seeing (or not seeing) the ad, at the person-level.

And as it just so turns out, we’ve recently developed a measurement method that does all three.

Why presence, impressions and person-level matters.

Before getting into the exact mechanics of our measurement solution, it’s useful to explain the values and priorities that led us to it. For us, the TV attention standard needs to include three attributes:

  • Presence: We need to measure whether there’s anyone present in the room when the ad airs. In digital, this information is inferred — because consumers are already in front of their PC or mobile device when ads are on or off the screen. In contrast, TV has content filling the screen, but viewers are in and out of the room - regularly leaving the TV on while they’re in another room to grab a snack or talk on the phone. 
  • Impression-level measurement: Advertisers need to know the attention of their individual ads, not the average attention of all ads that run during a given broadcast. After all, attention ebbs and flows over the course of every program. For instance, people are much more likely to stick around through the ad breaks at the beginning of a football game than they are toward the end of a blowout.
  • Person-based measurement: Household attention measurement is insufficient — advertisers need to know which viewers are actually watching their ads. If you’re looking to reach 18-35 year-old women, you’re not achieving your campaign goals if the ad airs while only an older man is in the room to see it. Further, attention varies widely from person to person.

While other solutions may include some of these attributes, our product focus is creating a best-in-class solution that includes all of them. While there are other valuable TV data sets for marketers to leverage for various use cases, including smart TV, STB and metered data, no one else has a complete solution to solve these problems for marketers.

This market gap is why TVision has built a new method for measuring this problem.

How TVision measures attention.

TVision uses an opt-in panel of real TV viewers to passively measure how they watch TV. Person-by-person, second-by-second. We don’t use a lab, we don’t require panelists self-report what they’re watching, and they do not have to push any buttons. 

Through computer vision technology and automatic content recognition, we can tell which person is present in the room for every piece of content and advertising that airs in our panelists' homes. Because we also measure on a second-by-second basis, we’re able to assess whether each ad is viewable to each viewer in a household on an impression level. And since our measurement is completely passive, it requires no change in panelist behavior and advertisers never have to worry about compliance.

Our methodology begins with defining an impression. While digital advertising was built on the impression, TV was not. TV settled on average commercial ratings, representing the average minute of audience size. But as digital and TV continue to merge, advertisers want to know the audience size and engagement for their ads, not the average of all ads that ran during a program. Before we can analyze how many viewers are in the room when an ad runs, we must first build the denominator - person-based impressions. Historically this could not be measured, but now we have the technology.

To start measuring impressions, we created rules with the intent of capturing “intentional viewership”. The industry needs to differentiate between people who leave a room momentarily while watching a program and those who leave and don’t come back. This includes ignoring channel surfing or people who pass through a room while someone else is watching TV. These are all new scenarios for us to define a TV Impression. Once we accurately build the denominator, then we can identify how many commercials are viewable and driving results for advertisers.

In order for us to measure an ad as viewable, the impression must run while a viewer is present with the sound on for two consecutive seconds AND the impression must meet the following additional criteria:

  • The TV is tuned to the same program for at least five minutes
  • A person is present for at least two minutes during the program
  • A person is present for at least one second during a rolling 21-minute window

These guidelines largely mirror the rules applied by TV meters, making it easy to layer our data on top of more traditional metrics. In fact, our guidelines are substantially stronger than those used by TV meters, as our passive measurement technology means we never have to worry that viewers will forget to check in or check out.

It’s time to follow digital’s roadmap.

Now that we finally have a thorough solution for measuring TV attention, there will certainly be an adjustment period for our industry — similar to the one we watched play out in digital over the past decade.

At first, there were growing pains, as stakeholders were skeptical of a new item to measure that challenged the way things had always been done. We see this happening now for TV measurement. But over time, everyone in digital came to acknowledge that an unseen ad has no chance of being effective. Eventually, the industry adopted a unified standard, and attention rates effectively doubled over a five-year span, which was better for advertisers and publishers alike. Today, attention is a consideration in virtually every digital ad buy.

From my perspective, the TV industry is presently where the digital space was five years ago. We recognize the problem, we have a great solution to build with and it’s up to us to hammer out the details and create a unified standard. The only difference, of course, is that there’s more money in TV than there’s ever been in digital. All the more reason for us to confront attention quickly and head-on.
TV and CTV Attention Report
The TV and CTV Attention Report