Header Bidding for Video Makes Sense: Q&A with Eric Hoffert, AppNexus

Header bidding adoption has ramped up quickly over the last few years, but has yet to spread significantly to formats other than web display. AppNexus is one of the companies pushing to change that, having recently released a guide for publishers on using header bidding for one of their highest-performing formats: video. In this Q&A, ExchangeWire asks Eric Hoffert (pictured below), SVP video technology, AppNexus, for more insights.

ExchangeWire: Why does header bidding make sense for video?

Eric Hoffert: Advertisers know that video is a powerful way to deliver a message, which is why the CPMs for video ads tend to be so much higher than for display. And we only expect more ad dollars to flow into digital video as the cord-cutting phenomenon continues. So, if header bidding can increase publishers’ ad revenue by upping the competition between buyers, why wouldn’t they apply that to inventory as valuable as video?

Header bidding can also improve latency, which is especially important for online video. If a pre-roll video ad takes too long to load, users will simply stop watching – studies show that every second of delay causes 6% of users to stop viewing a stream. Everyone – the user, the publisher, and the advertiser – loses in that scenario. That can happen frequently when publishers use a traditional waterfall setup to source video ads. Header bidding is an alternative technique that lowers latency by letting a publisher send ad calls asynchronously to all their video demand partners in parallel.

Are there any myths around header bidding for video you’d like to address?

Eric Hoffert, SVP Video Technology, AppNexus

Yes, I think there are three key misconceptions some publishers have that are holding them back from trying header bidding for video.

1. Myth: Header bidding is sure to add latency for videos. This concern is borne out of a pre-header bidding understanding of video advertising. Many publishers use a waterfall system for video now and pass impressions along a chain of as many as 30 demand partners. These publishers see header bidding as another complicating factor that will slow things down even more, when in fact it can directly address what’s most to blame for latency in the first place – inefficient waterfall setups.

2. Myth: There’s no header in a video player. It’s true that the header is part of the web page. But the video player itself is actually just another piece of Javascript within the page. If the publisher implements video header bidding properly, the header can easily pass a winning bid into the video player.

3. Myth: Video header bidding means data leakage. Some publishers are concerned that if they use header bidding for video, their demand partners will be able to see how much programmatic video inventory they have available. But, while on-page header bidding adapters may get information about the total number of impression requests, they’ll have no idea what proportion of these impressions are already sold out in the ad server versus how many are available to programmatic demand.

But how would a publisher implement header bidding for video?

Implementation with video is more demanding than display header bidding because the process varies depending on the header bidding wrapper, video player, and video ad server you use. That said, there are four key steps you need to carry out as a publisher to get started with header bidding for video:

1. Create video ad units on any pages where you have video content you’d like to monetise.

2. Add those video ad units to your header bidding wrapper in order to request video demand, including video demand adapters you need in order to access demand.

3. Integrate your wrapper and your video player so that your wrapper can pass the video creative of winning ads into your video player.

4. Create new line items for video ads in your ad server.

We explain each of these steps in greater depth in our video header bidding guide, but these are the key steps to do to get started.

What else do publishers need to know to get started?

There are a few best practices we recommend publishers follow to get the best results.

1. Source video demand before it’s needed. Set up your page to auction off video inventory as soon as the page loads, rather than when the user clicks the play button. That way, you can run the ad as soon as the user clicks play rather than waiting. We call this, “moving the auction away from the play button”.

2. Set up video deals. Targeting those advertisers that are well matched for your site content can lead to higher CPMs, compared to placing your inventory on the open exchange only. In certain cases with premium video publishers, deals and private marketplaces may be the only approach that is viable with header bidding in order to ensure that CPM goals are met when inventory is priced in to the $15 to $50 range.

3. Manage your video playlists. If you have a playlist, and make an ad request from the page level, you’ll get ads aligned with the metadata/content of the page, but not necessarily every video in the playlist. Instead, be sure to use the metadata associated with each video in the playlist, so that ads can align with the content in each clip; this should increase relevance and increase engagement.

4. Take the time and establish the team to do video right. Give your developers the time and resources to set everything up. Implementing header bidding for video isn’t the world’s toughest project, but make sure they have the bandwidth and expertise to do it correctly.

5. Work with multiple video demand partners. Remember, one of the biggest reasons you ought to use header bidding for video inventory is to increase bidding competition and drive video CPMs even higher. The more demand partners you have access to, the more competition you create, and the more you can monetise.

What about client versus server-side video header bidding? How do I choose the best approach?

There’s been a recent debate over which is a superior approach – client or server-side header bidding. It’s been stated that client side provides greater transparency for buyers combined with better user sync, while server side provides greater latency reductions while simplifying publisher page implementation.

Similar to display, AppNexus believe that a server-to-server video header bidding solution can work best in a hybrid capacity with a traditional header bidding wrapper like prebid.js. The combination of capabilities offers publishers the flexibility to maximise cookie matching with header partners and to increase video bid density via server partners, while managing latency and optimising revenue. We see this approach as an important next step in the evolution of video header bidding.

To wrap up, what are the big picture industry goals for video header bidding?

We are only at the earliest stages of video header bidding. We recommend an approach that is designed to be independent of underlying video infrastructure, with the following key objectives in mind:

1. Video player independence

2. Video demand source independence

3. Video ad server independence

The goal is to make implementation of video header bidding as easy and as flexible as possible so that video publishers of all shapes and sizes can deploy quickly. To achieve this goal optimally, it will be necessary to secure support and collaboration with actors in the video supply chain including video ad server providers, video SSPs/DSPs, and video player providers.

The broader goal is to build an open and transparent video advertising marketplace that promotes meaningful connections between video buyers and sellers, fair auction processes, and clear video pricing, while driving down latency and increasing user engagement with great content.