“Data-Driven Thinking" is written by members of the media community and contains fresh ideas on the digital revolution in media.
Today’s column is written by Ben Epstein, engagement manager at Jounce Media.
The near-universal deployment of publisher ads.txt and app-ads.txt files and the equally wide deployment of exchange sellers.json files brings a much needed dose of transparency to the programmatic supply chain.
However as buyers peel back the layers of complexity now visible as a result of these initiatives, new questions arise in pursuit of a sound supply-path optimization (SPO) strategy.
Identifying the most efficient and highest performing paths using the data available may actually lead a marketer to take a misguided approach that could be prevented with additional information not publicly available today. The deeper marketers go, the clearer it is how much is still left unanswered regarding the different ways publishers monetize and how bids are being transmitted.
Here are the three most commonly asked questions we hear from the buy side.
1. What are the technical connectors that handle my bid?
The RTB SupplyChain object and corresponding entries in exchange sellers.json files disclose all companies that handle payment between the marketer and the publisher, but there are also nonfinancial intermediaries – wrappers, mediation layers and ad servers.
Savvy marketers understand that these technical connectors have a critical impact on the efficiency of a supply path, yet there is currently no visibility into these technical hops. For example, a supply path that is integrated with the publisher via a code-on-page Prebid wrapper vs. a server-side Amazon TAM solution are currently indistinguishable to most buyers. Neither Prebid nor TAM handles payment, so their presence is not disclosed by the publisher's exchange partners despite these supply paths carrying significant trade-offs for match rate fidelity, latency risk, bid stream filtering and, most importantly, the total economics to the buyer and seller.
The solution today is for marketers to establish data-sharing agreements with exchanges. Among other things, these data-sharing agreements disclose the technical intermediaries that transmit exchange auction clearing prices to the publisher's ad server. These data-sharing agreements can take the form of log files for marketers who are equipped to work with multibillion-row data sets, or they can take a more aggregated metadata form that is consumable by most business analysts.
The long-term solution to technical transparency was anticipated by the IAB working group that defined the SupplyChain specification, which contains an optional hp ("handles payment") variable. Today's implementations of the RTB SupplyChain object only disclose financial intermediaries (hp=1), but future implementations will also hopefully declare the presence of technical intermediaries (hp=0). This would expose technical intermediaries to all buyers, not just those who have the resources and leverage to negotiate custom data-sharing agreements.
2. What is the fully loaded cost of each supply path?
Disclosing take rates is a sensitive topic, but most exchanges seem to be warming to the idea of providing fee transparency. Marketers and publishers have highly aligned incentives to transact through cost-efficient supply paths, and the technology companies that power these cost-efficient paths are highly motivated to showcase their low fees. It's worth noting here that savvy marketers don't simply optimize to the path with the lowest fee, but fees are certainly an important component of a complete SPO strategy.
Today, marketers obtain take-rate data via the same data-sharing agreements previously discussed. Fee disclosures are most commonly provided by exchanges and are limited only to publishers that have authorized take-rate transparency. And like transparency into technical connectors, marketers can gain transparency into take rates through either log files or aggregated metadata.
Looking ahead, there are multiple initiatives underway to increase marketer access to take-rate data. Companies such as Amino and Lucidity are building technology that may create an auditable end-to-end record of each programmatic transaction. DSPs, most publicly MediaMath SOURCE, are developing preferred transaction pathways for which exchanges declare take rates in bid requests. There are also proposed solutions for more directional signaling of the publisher's preferred supply paths via ads.txt files. Publishers could, for example, order their ads.txt files to list preferred partners first. They could alternatively mark low-fee paths with a special comment.
I don't expect to see widely broadcast details on the contractual terms of each exchange and publisher relationship, but I do see many possible solutions to expand the pool of marketers that are able to identify cost-efficient supply paths, which ultimately leads to more money in the publishers’ pockets.
3. What inventory does each supply path sell?
When a supply path has exclusive right of sale for an ad placement, there is no SPO opportunity. The publisher gives the marketer only one option. We call this "single-path supply," and it's critical that marketers do not unintentionally disable bidding into single-path supply as part of a broader SPO effort.
But it turns out to be surprisingly challenging to identify single-path supply. The root cause is that each exchange assigns a different ID to each publisher placement. Rubicon may have an ad slot that it calls 123, for example, and Index may have an ad slot that it calls ABC. Are these different placements or just different IDs for the same placement? It's really hard to get that answer today.
One solution is to somehow synchronize these IDs so that we know, for example, that Rubicon placement ID 123 and Index placement ID ABC represent the same ad slot. DSPs are best positioned to build this mapping by observing similarities in the bid stream, such as whether bid requests share a user ID, timestamp, publisher URL and creative spec. But this seems like a backward way of approaching the problem. We ought to be able to do better than inferring the inventory that lives behind each supply path.
A much preferred solution is to have a common publisher-generated ID for each placement, which would then be signaled in bid requests by all exchanges. There are efforts under way to pilot these solutions today, but we've got a long road ahead for a unified placement ID to have broad industry adoption. An even more granular solution would be to transmit a common publisher-generated impression ID through all supply paths, giving DSPs a signal to (a) deduplicate the bid stream and (b) quantify each supply path's access to each publisher placement, but this is unlikely to be deployed any time soon.
In the absence of industrywide initiatives to address these questions, the most resourced marketers achieve an information advantage that allows them to outperform their less-resourced peers. This information asymmetry is great for some, but it's probably not great for the long-term success of the programmatic advertising market.
So let's start by addressing these three questions, and by the time they're solved, odds are high that the savviest marketers will have new questions for the industry to resolve.