An Introduction to Video Header Bidding

Post on March 12, 2019 by Mike Chowla

Mike Chowla
Mike Chowla Senior Director of Product Management, Header Bidding

Header bidding, regardless of the ad format, operates in a similar manner. It is being used for various formats, currently, with video being in its nascent stages. Before jumping into video header bidding, let’s lay a foundation for how the process typically operates.

First, multiple demand partners are simultaneously given the opportunity to bid on ad inventory. A demand partner is simply an ad exchange, SSP (supply-side platform) or ad network. Header bidding differs from the older method of waterfalling demand partners where a single demand partner was given an opportunity to bid. If the demand partner returns a bid above the floor price, that bid fills the impression. If not, another demand partner is called.

Waterfalls are conceptually simple but problematic from a yield perspective since the order of waterfall is typically decided on historical data. Let’s say demand partner A had the highest average yield yesterday, thus it was put at the top of the waterfall but for any given impression, another demand partner may bid higher.

How Video Header Bidding Works

With video ads, similar to display, the header-bidding wrapper (or software that manages all bidders) is ahead of the impression being sent to the ad server. The wrapper invokes all the configured bidders, waits for bids to be provided, or the maximum wait time is reached, then selects the highest bid.

The highest bid is then fed to the publisher ad server by adding targeting keys with the bid information. Those targeting keys create a wrapper line item on the ad server. The ad server then looks at all eligible line items and selects one.

In general, ad servers select based on the configured priority of the line item and then among line items at the same priority, by price. If the wrapper line item is selected, the corresponding creative from the wrapper’s bid will serve.

Yield Improvements from Video Header Bidding

Header bidding improves publisher yield—no matter the format.  However, there are two unique reasons why.

First, let’s start with an intuitive approach. Say you have a valuable painting you want to sell at an auction. Instinctively we know that to get the best price, we want all interested buyers at the auction. If every interested party is bidding at the auction, then whomever wants the painting the most, has to outbid all the other buyers. However, should half the buyers not show up, odds are the winner wouldn’t have to pay as much.

Additionally, rather than an auction, if we asked interested buyers one-by-one, how much they would pay and then sold the painting to the first buyer who was willing to pay over our minimum price, our intuition tells us we’d get a much lower selling price.

Having all competing bidders means we get the true market price. In the sequential scenario, the first bidder has an unfair advantage since they can pay less than the market price by being the first bidder.

The other way to understand why heading bidding results in better yield is to look at a concrete example. Let’s say we have an impression with three demand partners bidding and our floor is $2.00 CPM.  Here are the bids:

  • Bidder A: $2.50
  • Bidder B: $4.00
  • Bidder C: $6.00

If the order of the waterfall is A then B then C, A win the impression with a $2.50 bid. But with a header bidding setup, C wins the impression with the $6.00 bid. My example assumes a first-price auction but with header bidding, most auctions are first-price since the bid has to go on to compete at the ad server with direct demand and any other wrappers the publisher is running.

Our first waterfall winner was bidder A because on average, A has the highest yield. However, for any given impression, A may or may not be the highest bid. By getting the bids from all bidders simultaneously, yield goes up because whoever is willing to pay the most for the impression, wins.

Implementation Options

The main differences between display and video header bidding are in how the integration is done. With display, the mechanics of how the wrapper sends the bid to the ad server and serving the ad are relatively simple. For video, the situation is a bit more complex when using a client-side wrapper as there is generally JavaScript needed on the page to connect the wrapper to the video player.

The complexity here comes from the many different video players in use. Additionally, if mid-roll or post-roll ads are being served, the integration code needs call the player’s APIs to insert the ad in the correct place. In reality, every player handles this differently. The alternative to the client-side integration is a server-side wrapper that runs behind a VAST tag. With a server-side wrapper running behind a VAST tag, implementation is simplified.

Want to Learn More?

Client-side and server-side integrations each offer a distinct set of benefits and challenges. For new implementations in particular, we see strong CPM and fill rates when starting with Prebid client side (CS) integrations. As we have seen here, multiple demand partners bidding simultaneously improves video inventory yield. That’s why we ultimately recommend building toward a multi-integration header bidding strategy that incorporates both a client-side and server-side approach. Adopting more than one header-bidding integration allows for more control and better user experiences. This, in turn, results in increased competition and more publisher revenue.

PubMatic’s supports all major client-side and server-side video header bidding integrations, including OpenWrap, our server-side header bidding solution that operates via VAST tag to deliver additional monetization opportunities. To learn more about our flexible solutions, visit our product pages or contact us.