Getting Started With SSPs, Waterfall & Header Bidding for Programmatic Advertising
For traditional publishers, developing and implementing a programmatic approach to ad sales can feel complex and challenging. There is good news though. If a robust digital ad sales practice is already in place many of the key tools needed are already part of your ad infrastructure; you’ll need to layer programmatic technologies into the ad stack, and apply optimization tactics, but the fundamentals are there. Better yet, publishers with well-established ad buyer relationships can look forward to a bump in yield optimization. Even in the “machine driven” world of programmatic nurturing the human connection pays dividends.
Let’s look at the ecosystem and see how the workflow needs to be tweaked to transition to programmatic trading:
Look at those elements highlighted in red; these will be familiar to you, as your ad server already supports direct sales for digital ad placement.
For a deeper dive on programmatic strategy and header bidding, check out the second blog in this series.
How to Work With Supply-Side Platforms
The vital pieces to add are additional Supply-Side Platforms (SSPs). Your ad server may already come bundled with an SSP, but in order to maximize bidding competition you’ll need to make your ad inventory available programmatically via a number of additional SSPs. Since these SSPs are already connected to ad exchanges, your suite of SSPs will enable you to participate in the most popular form of programmatic, the Open Auction. Open is a great place to start but down the line these SSPs will play a key role in relationship-driven forms of programmatic, such as Private Marketplaces (PMPs), that we will discuss later.
The Waterfall Approach
Let’s look at your initial push into programmatic. You will likely treat your SSPs in a similar manner as ad networks or even direct advertisers and schedule them into your ad server. By configuring delivery priorities and price floors in each SSP, you will manually dictate the competition for each of your programmatic ad impressions. This will set-up a “waterfall” where you get to dictate the order in which each SSP gets the opportunity to fulfill an impression by site, platform, audience or any other delivery criteria available within your ad server. The optimization tactics analyze past performance of each SSP, and establish the priority within each waterfall accordingly.
The obvious shortcoming of the waterfall approach is the manual nature by which the competition is set. Why can’t you have every possible programmatic demand source provided by your SSPs bid on every impression that you deliver? This “super auction” would always find the winner across your entire suite of available demand as opposed to a possibly lower CPM winner from an SSP that you have manually set with a high priority. Additionally, with programmatic advertising typically running second-priced auctions, you want to maximize the CPM of the second-best bid.
This “super auction” exists. It’s called header bidding. You choose one partner, typically an SSP not built into your ad server, and this partner gathers the highest bids from other SSPs to which it is connected. For “client-side” header bidding, every mediated SSP also places a tag on the pages. As you can imagine, the yield increase from the super auction is very attractive to a publisher. However the glut of tags does introduce performance concerns with content and ad loading. As an alternative, the more recent implementations of “server-side” header bidding mediation, only requires a tag from the primary partner. This addresses the latency issue but does put the non-primary partners at a disadvantage when it comes to matching an ID to the user, a critical need in determining a bid amount.
Managing a programmatic waterfall and then upgrading to a super auction environment will require a adding a few new tools and skills -- such as yield optimization specialists to your team. The revenue improvements will be worth the additional costs. But to take it to the next level, and achieve another leap in yield, it’s your sales team and their old-school relationships with buyers that will be the key drivers of success.
The CPM lift that you will achieve by optimizing the Open Auctions is extremely valuable, but moving programmatic supply up the food chain to people-managed areas such as PMPs will garner even higher returns. You’ll need to harness your Data Management Platform (DMP) to the infrastructure you already have in place for the Open Auction. Almost all of targeting activity in the Open Auctions is based on buyers “bringing their own data” and then bidding on audiences that match their targets. To entice buyers to pay more per impression for your Private Auctions or one-to-one versions of programmatic (such as Preferred Deals and Automated Guaranteed), you are going to have to offer ad products not available in the Open Auction. Since the Open Auction is typically filled with your run-of-network (RON) and universal audience inventory, creating audience segments in your DMP that are both attractive and based on proprietary data points will translate into unique ad products. If you are not going to offer any unique ad products in your PMPs, though, it is likely not worth participating.
There are no technical “silver bullets” when it comes to programmatic yield optimization. But, with a few upgrades in technology and team, a publisher will be able to leverage existing expertise and relationships to create a very successful – and lucrative - programmatic strategy.