Link to home

News from the open internet


Surprise, surprise: Ad tech is overcomplicating SPO. Here’s why it’s simpler than it seems.

A hand pulls a twisty line into a straight path with people walking across it.

Illustration by Holly Warfield / Getty / The Current

Supply path optimization, or SPO, is simultaneously the most covered, and least understood, topic in ad tech — with nearly every constituent in the supply chain using the term to describe some form of choice or optimization. The lack of precision creates unneeded complexity because it widens the surface area of the discussion beyond its scope, making it near impossible to have an informed debate. I would argue the coverage of SPO dabbles more in palace intrigue and superficial competitive dynamics than the core opportunity — creating market efficiencies that make programmatic perform better.

To put it another way, this is not about winning and it’s not about beef between demand-side platforms (DSPs) and supply-side platforms (SSPs). It’s about ripping out structural inefficiencies in the market, root and stem. If consolidation creates a market with fewer players but with the same core problems, SPO will have solved nothing — inarguably leaving our industry worse off than before. Be wary of framing that does not center the conversation on efficiencies gained from economies of scale and superior unit economics.

As a starting point, let’s go back to basics. The intent of SPO is to clean up the supply chain and make it more efficient, competitive, and transparent. This means ensuring that everyone who plays a role in serving an ad impression is adding more value than they are extracting. And the stakes are big. In terms of volume, advertising’s inventory marketplace will exceed that of Wall Street and the global financial markets. In terms of total addressable market, it is one of the few trillion-dollar industries.

Header bidding, and its near-ubiquitous adoption, is the genesis for SPO. Even in non-web environments, where there is no page to run an auction, the principles of header bidding are a forcing function toward a unified auction environment everywhere, elevating price-based auctions as the primary determinant of yield.

Let’s look at the two primary types of SPO in today’s marketplace: commercial and performance. Commercial SPO — by far the most popular type today — involves a reduced take rate in exchange for scale. And while there is nothing inherently wrong with volume-based rate cards, I worry it conflates take rate with performance. A lower take rate on its own may not help win rates because the cheapest path is not always the best one. If an advertiser looks at take rate independent of other variables, they might create more problems than they started with. A lot of key questions need to be considered in addition to take rate: Did it make my bids more competitive? Did I win more impressions at a certain price than before? How did it impact my overall cost per mille (CPM)? Am I now paying more in CPM than the savings I gained in a reduced take rate?

"With performance SPO, in time, the advertiser will be able to winnow down the most efficient paths and deprioritize those exchanges that aren’t delivering value."

Just as problematic, a focus on take rate assumes all paths operate at parity. Dear readers, all paths do not operate at parity. Not even close. There are wide variances between paths. Not every path is ads.txt or sellers.json compliant. Not all paths send all avails — in fact, some paths duplicate avails to a troubling extent, some shape traffic down. Not all paths send the correct metadata in the bidstream. Not all paths have identity. Just as important, not all paths are fast, and not all paths run auctions the same. I would argue the above has more impact on win rate than take rate. When information used to bid correctly is not standardized, but gamed, it makes price discovery impossible and enables value extraction in the middle and dilutes marketers’ budgets.

This is why The Trade Desk’s focus is on performance-based SPO because it allows us to measure success across a multitude of factors in the marketplace. Performance SPO achieves the original mission of supply-path optimization — enabling true price discovery, while providing the correct metadata and signals to enable buy-side decisioning. For instance, if an advertiser bids one dollar on the same opportunity on the same publisher across different paths, they will see different variations between those paths across the multiple exchanges they use. With performance SPO, in time, the advertiser will be able to winnow down the most efficient paths and deprioritize those exchanges that aren’t delivering value.

End state also resolves the take rate issue at scale. Performance SPO creates incentives for the most efficient paths to standardize around the most important variables and ensures each path is not gaming the auction. Over time, as the paths mature, competition will create parity between paths, until the last remaining variable is take rate. The competition for marketers’ budgets will then force downward pressure on margin and lower the transactional cost of each impression.

Folks, we can have our cake and eat it, too. We can create the right market dynamics for scaled, efficient paths and create market competition that will lower the cost of those paths all at once — but we must move past superficial take rate conversations.

If SSPs are upset about anything, it’s the way performance SPO dynamics are creating tons of pressure on their bottom lines. But the upside for the SSPs that succeed in this market are much, much greater than status quo. The cost of running any exchange is not the amount of impressions served but the amount of auctions run. Even the best SSPs monetize only a single-digit percentage of the auctions they run. If the average publisher works with more than 25 SSPs, and each SSP only monetizes a small percentage of the impressions, the amount of unmonetized auctions is as staggering as it is expensive.

In short, performance SPO enables the market to squeeze out inefficiencies. That means some SSPs that are not delivering efficiency and quality will be challenged in this marketplace. But the most important outcome here is that it creates positive economic incentives for supply to have a clean, scaled supply chain because it rewards good behavior.

It is in this vein we decided to launch OpenPath. We wanted to get a direct look at publishers’ inventory without any gamed auctions or manipulated bid requests. We have no other ambitions. We just want to find the most scaled, clean look into supply. OpenPath is not a marketplace. It can’t reward The Trade Desk any special privileges or benefits. It doesn’t shape traffic, obfuscate, or alter floors to artificially raise CPMs. It gives us a clean benchmark in which we can make the best possible decisions on each opportunity. Our goal is to provide advertisers with direct access to premium digital advertising impressions, while giving publishers the ability to maximize revenue from those impressions. In this way, OpenPath aimed to remove inefficiencies often present in the programmatic supply chain by forcing the market to compete on value, not mechanics. It has allowed us to better understand what a clean, unadulterated supply chain can look like.

And so, one of the things that OpenPath has given us is truth. How effective is a dollar for each one of these paths? There’s a beauty in that kind of simplicity.