Private Exchanges from a Premium Publisher Perspective – What Does It All Mean?

The Sell-SiderThe Sell-Sider” is a column written by the sell-side of the digital media community.

Today’s column is the second of a two-part series and written by Esco Strong, Director, Exchange Marketplace Management at Microsoft. You can read Part I here.

In the first part of this series, I highlighted the common industry definitions of “Private Exchange”. Today, I’ll offer some analysis of these different options, as well as discuss some of the key considerations of each for premium publishers.

One point in common between the three key options we discussed in part I (the siloed marketplace, the tiered auction, and exclusive-access ad slots) is that each model allows a degree of additional control for publishers that may help them overcome their concerns about indirect sales through RTB. Maintaining the viability of direct sales channels while introducing an RTB exchange offering is a complex, challenging, and downright scary proposition for many publishers. If managed incorrectly, this new channel can potentially undercut upstream efforts and eat into a publisher’s largest revenue streams, thus alienating their direct sales force. But when managed appropriately, RTB sales can be a powerful stream of demand to tap into that drives complementary revenue and ad offerings to round out the publisher’s portfolio. However, as we’ll see, each of the “Private Exchange” implementations offers a unique path to doing so – sometime with unintended consequences that may actually negatively impact the publisher’s overall yield or the health of their sales channels.

Use case #1 revisited: “Private Exchange” = a siloed marketplace with controls managed individually by publisher

This implementation makes sense for a publisher who has a good sense for how to set up their own RTB marketplace in a manner that is complementary – and not conflicting – with their other sales channels. Not surprisingly, this requires a high-degree of expertise in marketplace design, policy setting, and pricing, as well as a great deal of human resources and technology to apply against some of these problems. Finally, this model requires a fairly substantive set of inventory to attract demand, or a technology partner that has already created those relationships and is willing to plug that demand into the publisher’s white-labeled “Private Exchange.”

It may be difficult for the publisher to properly merchandise that inventory to the marketplace if there is not sufficient supply mix such that they can maintain a level of opacity over their inventory, such as to protect their brand sales. For example, a publisher with only a handful of placements comprising “Sports” inventory may find it difficult to create a “Sports” category offering through RTB that doesn’t compete directly with (and likely at lower price points than) their direct sold “Sports” offering. This concern may be mitigated by either listing the inventory as part of a larger exchange where other publishers’ sports inventory is present, or by selling the “Sports” inventory blindly without the category label that may otherwise create the risk of channel conflict.

There are also other economies of scale at play when considering a Private Exchange, where it may not make sense for individual publishers to “go it alone” and establish their own marketplaces. Continued fragmentation of the market runs counter to the efficiencies possible through the consolidation of RTB markets into bigger liquidity pools, where proliferated demand meets readily-available supply to ensure the most effective matching of buyers and sellers. Scalable sell-side technologies, such as price floors and custom auction rule sets, require engineering investments that stand as a barrier to entry for many publishers, and these features may perform better when based off of a broad set of different signals coming from a variety of shared sources.

Use case #2 revisited: “Private Exchange” = a tiered auction

Designing the RTB marketplace as a tiered auction seems like a logical way of dealing with some of the core issues of introducing an indirect sales channel, which include concerns with the quality of the demand, pricing, and channel conflict. Giving preferred demand in a higher priority access tier assures that the ad quality and user experience will be more tightly controlled. It also makes it easier to set different rules such as pricing, transparency, etc. depending on the demand segment of the potential buyer. This also syncs naturally with the way that sellers intuitively think about allocating their inventory – first to buyers who can pay more, then to sources with lower bids, and finally to sweeper networks or house ads that can monetize the remaining impressions.

There is however a critical flaw in this strategy: when a downstream, “lower-tier” demand source is actually willing to pay more for an impression than someone in a higher tier. One of the key value propositions of an exchange auction model is the efficiency of matching buyers and sellers at scale, such that the more valuable impressions are delivered to those who can do the most with them. The intended outcomes are then both better ROI for buyers and greater revenue for sellers, resulting in accrual of the value created to the principals involved in the transaction. This efficiency is interrupted when tiering occurs in the marketplace, and can result in both frustrating experiences for buyers as well as suboptimal revenue performances for publishers.

A better solution to this same problem is to implement a “custom exchange” – rather than a private one – where demand is not strictly excluded or tiered, but instead is evaluated at the same time with a custom set of rules for each.

There are two key benefits to this approach: 1) it maximizes the revenue for the publisher by selling each impression to the highest bidder at the maximum clearing price, since it evaluates all potential second-price (i.e. clearing price-setting) bids within the same tier, and 2) it maximizes the visibility into demand and the potential opportunity costs of custom business rules applied to certain bidders, allowing a publisher to regularly evaluate the tradeoffs being made between demand quality/exclusion and potential revenue.

Keep in mind that under this scenario, it is still possible for a publisher to exert total control over the inventory by establishing rules per bidder. Higher quality demand can be preferenced, while a “hurdle rate” may be applied to less-desired ads and while also still gaining visibility into their potential revenue contributions if settings were modified. In a tiered auction, the publisher simply foregoes this ability to assess those subsidy costs due to the cutting off of lower tier demand from bidding on impressions cleared in the upper tiers. It then becomes impossible to make informed decisions in the future as to the effectiveness of those rules, or to determine whether a better set of rules can be implemented to achieve the publisher’s goals.

Use case #3 revisited: “Private Exchange” =  exclusive-access ad slots with predefined transactional terms

This option is an attractive complement to a direct, hand-sold sales channel, but it still enables marketers to leverage the efficiencies of RTB buying, such as global frequency targeting or capping, brand and creative rotation, and proprietary data targeting.

In this method, a publisher slices off a predefined segment of inventory and delivers it to an advertiser through RTB-like plumbing – typically via a DSP. They are then able to apply the same real-time logic to these buys: global frequency management, brand optimization, DCO, etc. The catch here is that such an offering needs to be priced high enough to be consistent with the publisher’s direct sales, while reflecting the increased efficiencies delivered to the buyers through a premium on that pricing. This way, both buyer and seller realize gains through this method, and share the benefits gained from more efficient matching of inventory to ad.

One potential pitfall for publishers to avoid when implementing this flavor of “Private Exchange” is around guarantees and passbacks. Similar to my assessment of tiered auctions, siloing off inventory for delivery to a private ad slot is akin to that same concept of putting up walls around your demand and introducing opportunity costs of not selling to downstream demand that may be willing to pay more for some of the impressions. Therefore, it is critically important that publishers make this method work for them by securing a guarantee of consumption at a set price in exchange for the privilege of reserving this inventory.

This method is quite similar to guaranteed direct sales, where the average pricing is favorable to both parties considering the volume being consumed. Many buyers will press for this configuration, but without the guarantee of consumption – sometimes called a “passback” or “cherry picking” of the inventory available through the ad slot. This is a risky proposition for publishers, as they are essentially executing an extreme version of a tiered auction (where only one buyer has right of first refusal and can select all of the best inventory while leaving much lower quality inventory behind). In this scenario it may be difficult for the publisher to determine an appropriate pricing structure for such an offering, since they will no longer be able to assess other buyers’ willingness to pay for that inventory, nor the quality of the inventory being consumed relative to that being “passed back” to the publisher to try to monetize elsewhere.

In Summary

Navigating the world of RTB exchanges and their “Private Exchange” variants can indeed be a minefield of confusing and complex choices. Premium publishers should carefully consider their options for an exchange design to ensure it is as complementary as possible to their offerings.

Follow AdExchanger (@adexchanger) on Twitter.

Enjoying this content?

Sign up to be an AdExchanger Member today and get unlimited access to articles like this, plus proprietary data and research, conference discounts, on-demand access to event content, and more!

Join Today!


  1. Hi Esco,

    I completely agree, as a broad stroke, its easy to see these private exchange scenarios as the tool that brings more premium inventory into the RTB ecosystem. Of course MSFT, can claim a First in that regard.

    Aside from Appnexus, can you cite a few other companies SUCCESSFULLY offering private exchange tools to Premium Publishers? I imagine it will come from a combination of SSP’s, Exchanges, DSP’s and DMP’s – any opinion on which vendor type will be more successful? Or any trends on who will be successful here?

  2. Ben Trenda (@btrenda)

    Hey Louis- Rubicon Project has supported all three of the use cases outlined here for over a year, at least. I believe some of the other SSP’s do now as well.

    For Esco- your comments on use case #3, i.e. private exchange, are spot on. These deals, while exciting in theory, don’t seem to have generated much traction yet, other than the conference panel airtime they’ve been given. A couple other concerns you didn’t cite:

    1. if you add the margin captured by the SSP/ exchange and the DSP, you’ve got quite a bit of buying power erosion. Let’s say a publisher gets a $10CPM direct, and the buyer wants to buy it at $10CPM via private exchange (a pipe dream today) and is willing to make commitments (unlikely), the publisher will net somewhere between $6 and $7 once all the ad-tech platforms have taken their cut (not including any additional fees charged by the Trading Desk). Does this mean the publisher should charge $14-$15 or more for private exchange inventory?

    2. All this assumes the RTB platform is getting access to the best inventory, i.e. the stuff you would get if you bought from the publisher directly, through the sales team. Let’s face it– this is not the case with any RTB platform today. It’s still ALL remnant. The direct sold ads run first, then when the adserver runs out of direct-sold ads to show, then the RTB platform gets a chance at the impression. This is always the case, and needs to be– why would a publisher risk a big advertiser relationship to squeeze an extra nickel out of RTB by running an RTB platform before a direct sold campaign? It doesn’t happen. Ever.

    Therefore, there are really just two options for a publisher approaching a private exchange/ RTB strategy:

    1. place your premium inventory in a private exchange and charge $14 as opposed to the $10 you get direct; hope that buyers are willing to pay 40% more for the privilege of using their DSP; hope that buyers will make volume commitments; figure out how to predict exactly which impressions the buyer will want, or let them cherry pick.

    2. use RTB only for remnant and don’t worry so much about pricing; hope for the best when it comes to channel conflict

  3. Esco Strong

    Thanks for the comments guys. I wanted to add a little more color to the discussion here by clarifying a couple of points. First, I believe we’re all doing a disservice to the industry by continuing to use the “private exchange” term so broadly with multiple meanings. I’d love to see the term possibly go away completely and instead be replaced with more descriptive and precise new terminology.

    The other point I want to make is that I’m generally of the belief that the “private” part of private exchange is not such a good idea. It might make sense in use case #1 if it’s “private” with respect to who is selling, i.e. a seller leveraging their own brand equity to create their own marketplace. But to the extent that the “private” term refers to tiering auctions or limiting demand, I think the concept runs counter to some of the primary benefits of running an RTB auction in the first place – voluminous supply and demand and the efficiencies of matching at scale. It is much better for everyone involved if premium publishers instead implement a “custom exchange” where there is a single auction per impression with granular sell-side rules in place to control who they are selling to and how.

    Use case #3 isn’t really an exchange at all, and hence I don’t think it’s a proper reference to call it a “private exchange” – great example of where I’d like to see different terminology take hold in our everyday conversations.

  4. John Garber

    Hi Esco

    How about an example of the “custom exchange” in some more detail? What types of rules might someone impose on buyers that would result in your vision of this model?