|Starting March 31, 2021, docs.xandr.com is the place to be!|
For more information, see Building a Better Customer Experience.
How It All Works: The Big Picture
An Ad Call Is Received by the Impression Bus
The AdNexus Impression Bus is the heart of the AdNexus platform. It is a server cluster that processes ad requests, feeds data to partners, conducts auctions, returns ads to the publishers, keeps track of billing and usage, returns auction-result data, and enforces quality standards.
At the ad call, the Impression Bus receives a unique AdNexus user ID and fundamental page information from the user's AdNexus cookie and the AdNexus TinyTag or PriceCheck tag. These two type of tags is how a partner will manage all campaigns with ease and how AdNexus will be able to create a universal way to identify inventory. And since TinyTag and PriceCheck tags are stored server-side, once in place, you will never have to re-tag a publisher page again.
The Impression Bus will also send your bidder a Pixel Request. Pixel Request creates a simple way to synchronize your user IDs and metadata with the AppNexus user IDs and data.
With TinyTag and the Pixel Request, the Impression Bus will send your bidder the appropriate user and page information. Your bidder will receive this information either from your preexisting campaign or from another bidder's preexisting campaign.
The Impression Bus Holds an Auction
When an ad call from one of the AppNexus tags hits the Impression Bus, we review the content and send a bid request to the various bidders on the AppNexus cloud. The bid request will include the content categorization, page information, user information, and in certain cases a reserve price and reserve creative.
Bidding on Impressions at Auction
Data providers integrated on our cloud recognize the user ID and can input any information they may have on that specific ID and will pass the user data to bidders that have the rights to that information. The bidders then evaluate the ad call on behalf of their advertisers and return a bid value to the Imp Bus.
Serving the Impression
The impression is served in one of these ways:
- Find Highest Bid and Serve Immediately: The AppNexus Imp Bus determines the highest bidder and will serve the winning bidder's creative as dictated in the bidder response.
- Highest Bid Wins if It Beats Reserve Price: If a reserve price is passed via the AppNexus ad tag, the Imp Bus passes this price as well as the other ad call information to all bidders. Bidders respond with a bid value and creative ID and the highest bid that beats the reserve price will serve its creative. If no bid beats the reserve price, the Imp Bus will pass the ad call on to a third-party system.
- PriceCheck - Send Highest Bid to Third Party: The ad call is directed to the AppNexus Imp Bus to get first glance in order to perform a price check. The Imp Bus holds an auction as usual among its bidders and passes the highest bid on to a third-party system to decide how to fill the ad call. Only if the highest AppNexus bid is chosen as the winner will the ad call be passed back to AppNexus, where the Imp Bus will serve the winning bidder's creative.
One of the main problems with working in a closed advertising ecosystem is that it is difficult for bidders to understand why they won or lost an impression. If you bid $2.00 for a car buyer on nytimes.com/autos and consistently lose, what action should you take? Are you being outbid by $0.05 or $5.00? If you are selling inventory, which user segments are driving your CPMs? This crucial information will be passed to your bidder via Notify Request.
Couple this with with full reporting and API support, and you have all the tools you need to make informed and powerful decisions for your clients.
- Optimized ad delivery by leveraging the low-latency cloud computing environment
- Brand control and quality
- Guaranteed range of CPMs for sellers
- Independent platform allowing deep integration with technologies, optimization engines, and bidders
- Increase the number of high-value users for your clients while decreasing the number of low-value users