Supply path optimization, better known as SPO, might seem like another buzzy tech phrase looped into the programmatic landscape, but it’s one of the industry’s most effective ways to cut the clutter between media buyers and publishers hungry to sell their inventory.
Though different players in the ad-tech ecosystem need different approaches, the general idea is the same: getting the best price with the least amount of manpower.
Before header bidding gained steam in 2017, Adam Soroca, who heads Rubicon Project’s global buyer team, said that roughly each programmatic exchange represented a different supply route, and competed against each other based on their “queries per second.” The more pings an ad placement got, the better the demand side platform fared—but that gave programmatic buyers new headaches.
“You had these symptoms bubbling up everywhere—queries that DSPs didn’t have the bandwidth to handle, the introduction of fraud through multiple resellers, prices being driven up,” said Soroca, adding DSPs started taking action as symptoms piled up, leading to the first wave of SPO.
Though some companies have been public about using SPO to skim off extra players in the marketplace, it’s still a word that’s taboo in some corners of the industry. “Publishers want demand from as many sources as they can, and exchanges want demand from as many DSPs as they can,” said one Xandr representative. Having a buyer choose one path goes against that ethos.
As Soroca explained, it’s still a pretty new idea, and that newness might make ad-tech execs skittish. It could also be a hard sell to the publishers they work with, who have long learned that more resellers means more money in their pockets.
Those suspicions, though, are unfounded, he said. “SPO brings more transparency to the buy side, and gives the sell side more tools to manage their yield,” said Soroca. “It’s a win-win on both ends.”
Ad Tech’s SPO strategies
The ad-tech company focuses on the number of “hops” between an advertiser and publisher’s inventory. Soroca added that not all paths to a publisher are created “equally;” some are direct, some are through a reseller and some offer different viewability numbers. For its DSP, Agency and Brand customers, the answer is to trim certain exchanges they work with from networks with dozens of resellers to a series of sleeker omnichannel exchanges. Rubicon Project spends time consulting with these customers to find the best paths. Soroca said that removing inefficient paths means better outcomes.
Soroca added the purely direct inventory that advertisers buy is pricier, but ultimately worth it. While multiple hops in inventory makes for cheaper buys, it cause ads to lag, load improperly or appear in less-than-ideal locations. Meanwhile, advertisers who pay premium prices typically land premium placements on the page, more eyes seeing their ad and more money in their pocket.
Its SPO algorithm automatically shuts out supply-side platforms (SSP) that favor “aggressive auction tactics.” As former CEO Brian O’Kelley explained, header bidding was initially set up to reward those that are most cut-throat in auction—and will earn a publisher the highest yield. Favoring less aggressive ad-tech players is better for both sides in the long run, he said, by encouraging competition among the smaller players, and with time, CPMs drop.
The ad-tech platform’s strategy is an answer to aggressive auctioning, according to its vp of global marketing, Michael Rasmussen. He explain SSPs used header bidding to run blatantly unfair—or undisclosed auctions—to try to increase win rates, forcing buyers to spend more cash.
Supply-side platforms aren’t beholden to publishers or agencies when it comes to running a particular auction. Some can run a “fair” second-price auction, while others might modify them in some way. Some SSPs run first-price auctions. Meanwhile, each has its own logic for submitting the winning bid to a particular publisher.
To cut through that thicket, Iponweb introduced “optimal price discovery,” which leveraged historical data in the bidstream to predict what kind of auction a particular SSP was running. Too often, he said, assuming a “fair second-price auction” results in DSPs bidding higher than they need.