Programmatic for sellers: Unit 3 – Understanding how programmatic CTV/OTT is sold

Long-form vs. short-form content

Long-form video content typically lasts 20 minutes or more, while short-form content can run anywhere from 1 second to 20 minutes. Social media videos created by users usually fall into the short-form content category, and full-length TV shows or movies fall into the long-form content category.

In-stream video

Video-first publishers such as Hulu or streaming apps from television networks typically insert video ads directly into their digital content streams. These ads are similar to the commercial breaks used by traditional, linear TV channels. 

Video ads shown before the content or program begins are pre-roll ads. Ad breaks occurring within the content itself are mid-roll ads. Post-roll video ads play after the content is done.

Out-stream video

Outside of streaming video content, video ads fall into a larger catch-all category known as out-stream video. Out-stream video can take on many forms and be challenging to categorize. Examples of out-stream video include:

• Social media feeds
• Websites
• Blogs

There is a much wider variety of out-stream videos than there are in-stream. With the growing demand for video ad space, publishers and resellers are incentivized to fit video ad placements in as many places as possible — and the locations are not always to the users’ or the advertisers’ liking.

The line between video and display has blurred as digital advertising has evolved. The lines between video advertising categorizations are also blurry, so categories are not mutually exclusive. Here are some examples of ad units that walk that line:

In-banner video

Some video ads can be placed within the bounds of a banner ad unit. This typically looks like a small video running within a small box on the side of your web browser. Advertisers typically dislike this type of unit because they do not consider it true video.

Social video

Video ads that run within social media platforms are unequivocally video; however, they are not always clearly defined as in-stream vs. out-stream. For instance, on Instagram Stories or Snapchat, users are swiping through user-generated content and may run into an ad. The ad is sandwiched by content, but not like an in-stream ad would be.

Native video

Native advertising is any advertising that takes on the form and functionality of its location, so native video advertising is sometimes used as a catch-all term. Native video advertising exists within social media feeds, on video streaming sites and as recommended content.

There are several features associated with video ads that differ based on the seller of the ad, including:

• Video length: Standard video ad lengths are 6, 15 and 30 seconds, but these lengths vary widely from seller to seller.
• Autoplay: While some video ads will play automatically, the user must activate others directly.
• Sound: Some ads automatically play sound, while others play without sound unless the user unmutes the ad.

Skippable: Skippable ad units are typically more relevant when it comes to in-stream video ads. For instance, while Hulu ads cannot be skipped, certain YouTube ads can be skipped. Within YouTube, skippability is a feature advertisers can choose whether or not to include in their ads.

The connected TV space has become a unique beast with its own marketplace dynamics. Regarding programmatic, let’s start by understanding the relationship between advertising supply and demand in CTV.

The advertising on CTV apps and channels such as Netflix, YouTube, Amazon Prime, and Hulu is sparse relative to what is seen online. 

From an advertising perspective, this means at least two things:

1. There is a lot less advertising inventory available in CTV, which means supply is tight
2. The supply that is out there is controlled by a select number of publishers and content creators

As demand for CTV continues to accelerate, the supply vs. demand curve for CTV is flipped versus display. As a result, all the publishers and owners of CTV inventory have a lot more leverage and dictate the terms of engagement. These stakeholders can sell significantly more inventory to advertisers on a reserved, direct-sold basis. Since there is greater use of private deals and programmatic guaranteed, the prices for CTV advertising are much higher than display advertising.

Audience targeting on CTV takes a lot of work.

First, many premium, reserved CTV buys don’t allow targeting because the publisher requires the advertiser to book all the inventory. Second, when targeting is enabled, the available data is sparse. 

While basic targetability — geography, device, context, etc. — is fairly doable on CTV, audience targeting is more complex. Two kinds of CTV players usually hold audience data: 

1. CTV publishers can surface their audience data for targeting. Usually, that data revolves around simple demographic information and viewership history. This data can be utilized for targeting within that particular publisher’s app only. 
2. CTV device companies use a persistent ID called the Identifier for Advertising (IDFA) to track basic information and turn that into targetable audience data across apps. Some of these companies also collect automated content recognition (ACR) data, which can give advertisers a precise and robust understanding of viewership on a particular CTV device. 

Typically, a viewer’s primary interaction with CTV is choosing something to watch, so the data elicited about a user from a CTV app or device is not as rich as the data from a user clicking and engaging with web content. This is why an ID graph for CTV is crucial.

Targetability opens up if a CTV device can be tied to a user profile that gives an advertiser visibility into that user’s other devices. The most prevalent way this is done in CTV is by using the viewer’s IP address as the key signal. An IP is readily available when accessing a CTV device, so multiple devices with the same IP address likely mean all those devices reside within the same household. Now, an advertiser can use data from all of the devices to target users on the CTV device.

There are two drawbacks when using IP for targeting, however:

1. IP allows mapping to a household of devices instead of a specific user’s devices. While not being user-specific isn’t ideal, household-level targeting is generally acceptable in CTV since multiple household members can watch TV together.
2. IPs aren’t the most accurate way to connect multiple devices within a household, as sometimes there are multiple networks, the devices travel to and from different networks all the time, IPs change, etc. 

To top it off, IP is being heavily scrutinized as a privacy-unsafe identifier. Eliminating the IP address as an identifier for CTV advertising could significantly impact this ad market. The bottom line is that targetability in CTV is complex and will only get more complicated.

CTV measurement is heavily influenced by linear TV measurement. Like traditional TV, advertisers try to reach certain age and gender demographics and need to measure their success. 

Gross rating points are a metric that allows advertisers to evaluate their ad buy. GRPs are calculated by multiplying the percentage of the desired demographic reached by the number of times the ad aired against that demographic.

In CTV, Nielsen and Comscore are the most prominent companies that provide the data used to calculate GRPs, although some CTV measurement startups are starting to offer attractive measurement alternatives.

With digital channels, marketers can measure more than just reach and frequency. In CTV, advertisers also look at:

• Completion rates
• Brand lift
• Search and website traffic lift
• Direct response metrics like clicks and conversions

CTV advertising is unique in many ways and is going through tremendous change. In the next 5 to 10 years, programmatic sellers are expected to see a massive wave of dollars to shift from traditional TV advertising to CTV.