The New In-Stream/Out-Stream Divide: Explained

Tim Cross 15 May, 2023 

When an advertiser is offered a digital video ad impression via programmatic trading tools, they get passed a lot of information to help them understand what exactly it is they’re being sold. Some of this information is about the seller: who they are, and what content the impression appears alongside. Some is about the individual viewing the ad: any personal data, their location, the device they’re using, and so on. And some is technical information which the buyer needs to know in order to send the right piece of creative – what the dimensions of the ad are, what the maximum duration of the ad is, whether the ad is skippable or not.

But advertisers also want to know information about the type of ad being offered besides the technical specifications; is it a pre-roll ad on a piece of video content which the user specifically wants to watch, an out-stream ad in the middle of a news article, or something in-between.

IAB Tech Lab, the IAB’s body for developing industry standards and specifications, said last year it was rewriting the way this latter category of information is passed within programmatic buys, as the previous system has created some bad incentives, and made it hard for buyers to really know what they were buying. Thus early this year new guidelines were introduced, rewriting the definitions for in-stream and out-stream video ads.

The Basics

IAB Tech Lab’s guidelines previously defined five different types of video ad placements: in-stream, in-banner, in-article, in-feed, and interstitial/slide/floating, and while Tech Lab has updated its guidelines, these categories are still commonly used. Sellers label video ad impressions as one of these five categories, and that information is sent through to buyers in programmatic transactions. Buyers then use that information to inform their bidding strategies.

However after industry consultations, Tech Lab found that these categories don’t really give buyers the information they are most interested in. The in-article, in-feed, and in-banner categories each have distinct definitions, but from a user’s (and advertiser’s) point of view, there’s not a whole lot of difference between them – each refers to a video ad appearing next to non-video content. The ‘instream’ category meanwhile is too broad, covering lots of different types of content, and incentivising questionable behaviour on the sell-side.

Thus, Tech Lab has reworked its guidelines to four new categories for video ads: instream, accompanying content, interstitial, and no content/standalone. The idea is that these new categories will provide more useful information to buyers, giving buyers more clarity on what they’re bidding for.

IAB is now encouraging publishers to label their video impressions with one of these four terms. However publishers may want to label the impression using both the old categories and the new ones, to account for any buyers or tech partners which haven’t implemented the new terms.

The Technical Details

The primary contention with the old definitions was that they create an inaccurate divide between in-stream and out-stream ads. In-banner, in-article, in-feed and interstitial/slider/floating ad slots are all considered outstream ads. Meanwhile the in-stream ads category covers any and all ads which plays before, during, or after “the streaming video content that the consumer has requested” according to the old definition.

This wording is vague, grouping together quite different ad experiences. A pre-roll ad on a YouTube video or a mid-roll during a TV show on a streaming service would all be labelled in-stream. But if a publisher runs a piece of video content in a small floating player next to an article, and that video has some sort of relevance to the article the user had chosen to visit, they can also argue that ads which appear on that player are in-stream ads. And they’re incentivised to do so, since in-stream ads are considered to be more premium, and so command higher prices.

These floating players have become more popular with publishers for this reason. But there’s a world of difference between a large ad playing in front of a video which a user has clicked on, and an ad on a small muted video player which the user might barely pay attention to.

The new definitions, which aim to create better distinctions between these different formats, are as follows:

  • Instream: Pre-roll, mid-roll, and post-roll ads that are played before, during or after the streaming video content that the consumer has requested. Instream video must be set to “sound on” by default at player start, or have explicitly clear user intent to watch the video content. While there may be other content surrounding the player, the video content must be the focus of the user’s visit. It should remain the primary content on the page and the only video player in-view capable of audio when playing. If the player converts to floating/sticky subsequent ad calls should accurately convey the updated player size.
  • Accompanying content: Pre-roll, mid-roll, and post-roll ads that are played before, during, or after streaming video content. The video player loads and plays before, between, or after paragraphs of text or graphical content, and starts playing only when it enters the viewport. Accompanying content should only start playback upon entering the viewport. It may convert to a floating/sticky player as it scrolls off the page.
  • Interstitial: Video ads that are played without video content. During playback, it must be the primary focus of the page and take up the majority of the viewport and cannot be scrolled out of view. This can be in placements like in-app video or slideshows.
  • No content/standalone: Video ads that are played without streaming video content. This can be in placements like slideshows, native feeds, in-content or sticky/floating.

Tech Lab has created a new attribute called ‘plcmt’ in Object:Video, where the value given is either 1, 2, 3, or 4, depending on whether the impression in question is in-stream, accompanying content, interstitial or no content/standalone respectively. This is a separate attribute from the ‘placement’ attribute which is used for the original categories. Thus, sellers can pass on both a ‘plcmt’ value and a ‘placement’ value, labelling the type of impression using both the old definitions and the new definitions.

As an example, a streaming service might send the values “placement”: “1” and “plcmt”: “1”, to indicate that the impression is both an in-stream ad under the old definitions and an ‘instream’ ad under the new definitions (Tech Lab has dropped the hyphen for the new definition to make the distinction cleared).

Meanwhile a publisher which plays sound-off content on a floating player would label any video ad impressions offered within that unit as “placement”: “1” and “plcmt”: “2”, to show that it is an in-stream ad under the old definitions, and an accompanying content ad under the new definitions.

The Pros and Cons

As intended, these definitions create a distinction between ads which play alongside video content which is the primary content on the screen (e.g. a YouTube video or TV show), and ads which play alongside videos which sit next to an article or other piece of content (e.g. a floating player on a news article).

They also account for some specific edge cases. On YouTube for example, a user can choose to keep the video playing in a small floating unit while they continue browsing the website. This is a somewhat different ad experience from an ad playing when the video player takes up most of the screen. The new definitions account for this,  specifying that if a video player converts to a floating player, subsequent ad calls should be updated.

However, like with any definitions, there will still be some specific cases where it’s unclear which category is the best fit. For example, some news websites will put a large video at the top of a news article, covering the same topic as the article itself. Some users may just watch the video and avoid the text, for others it might be the other way round. But whether the video is “the focus of the user’s visit” is a little unclear.

Also, since the definitions are self-labelled, they don’t in and of themselves prevent any bad behaviour. Dishonest publishers could incorrectly label their content using these parameters, though it will at least be harder under these new definitions to make this look like an honest mistake.

Lastly, as is always the case with standards and definitions, these new categories rely on industry uptake in order to work – and clearly not all parties are incentivised to adopt these new definitions. There has been some progress here, with The Trade Desk last week announcing it will adopt the new definitions. But there is no hard deadline, which creates a difficult situation for publishers. Those who adopt these definitions earlier could be punished with lower CPMs, as video inventory which was previously labelled as in-stream is now labelled as accompanying content. And at a time when publishers’ ad revenues are already very squeezed, few will be keen to take a voluntary hit to their CPMs.

Follow VideoWeek on Twitter and LinkedIn.


About the Author:

Tim Cross is Assistant Editor at VideoWeek.
Go to Top