The dark art of yield optimization has produced a wide array of techniques for publishers eager to get the most money out of their inventory.
One of the most dependable of these techniques is “waterfalling,” which lets publishers move their inventory from one market to the next to optimize for revenue. But it’s a tactic that has remained stubbornly difficult to implement at times, despite all the promises of efficiency that are often attached to automated selling.
“There’s this assumption in the industry that ‘programmatic’ means easy and that it comes with less overhead, but that’s really not the case,” said The Economist’s director of revenue operations, Daniel Powell-Rees. “It’s a different kind of overhead. Programmatic deals aren’t taking any less time now than direct deals of the same size.”
But despite its implementation challenges, waterfalling is still a go-to strategy for publishers. Here’s a primer.
So WTF is waterfalling?
Waterfalling is a technique publishers use to maximize both the pricing and sell-through rate of their inventory. It’s also often called “daisy chaining.”
How does it work?
Publishers started doing it in an effort to make the most money for the unsold inventory they put on ad networks, which varied in both their specialties and pay rate. Publishers, trying to squeeze as much revenue out of each impression, worked with the networks that offered the highest rates first, before working with those that offered lower rates until they monetized every impression. Hence, “waterfalling.”
But many publishers have moved on from ad networks, right? It’s all about supply-side platforms (SSPs) now.
Yes, but waterfalling is still very much alive. Now, instead of daisy-chaining ad exchanges, publishers are daisy-chaining SSPs. Publishers start by selling impressions with one SSP (say, Google’s AdX) at a high price floor. If the impressions don’t get picked up, publishers push them to second (Rubicon Project) or sometimes third SSPs (PubMatic) at lower price floors until they do.
This sounds complicated. Weren’t SSPs supposed to simplify things?
It is complicated. Optimizing for multiple SSPs often involves a lot of testing, particularly since there are multiple buyers on multiple platforms potentially competing for the same inventory. Complicating it further is the reality that behind every SSP is an ad tech company trying to boost its own marketshare. So they aren’t exactly incentivized to play nicely with each other.
“You very often find that there are some technologies that work with some SSPs and not others,” Powell-Rees said. “You end up with this more complex ad stack that you have to implement so that everyone sees the inventory.”
So why just stick with one SSP?
The simple answer is that if publishers can find a way to maximize revenue, they’re going to experiment with it.
More money is good, but what’s the catch? There’s always a catch.
Poorly implemented waterfalling can cost publishers money. Having multiple SSPs looking at the same inventory to duplicate demand, with buyers bidding on the same inventory via different SSPs.
“You really need to be careful about what you’re doing,” said Darren Sharp, head of programmatic trading at Incisive Media. “If you do it wrong, you can give a network a first-look on an impression that you could otherwise get higher revenue on another platform.”
So what do publishers want?
Publishers say that in an ideal world, SSPs would talk to each other and allow their buyers to bid on each other’s inventory. If everyone sees the same inventory at the same time, the thinking goes, then publishers are getting a better value for the inventory that they’re selling. SSPs also get to see more inventory, and facilitate sales at higher rates.
“But that’s just the ideal,” Powell-Rees admitted.
Photo: Dave Edens/Flickr
More in Media
Legal pressure on AI companies illustrates the myriad challenges for companies that want to use or build generative AI tools.
The news rating service’s new features will track disinformation on websites, social media and video channels.