The Future of Programmatic is Programmatic Futures

ElroyBack in 2007, a company called TRAFFIQ started one of the first programmatic futures exchanges. The idea was simple: publishers committed blocks of premium inventory into the exchange at a stated price (say, a block of 500,000 homepage rectangle units in November at $8 CPM), and advertisers could construct packages of premium inventory at discounted prices by making future commitments. Basically, a better, faster way to buy digital guaranteed.

The idea never really took off. Publishers didn’t really understand how to value their inventory in the future, real-time enablement was just starting to take off, and advertisers and their agencies were deeply stuck in manual inventory procurement run by spreadsheets and fax machines. (TRAFFIQ went on to build some highly innovative workflow automation software, and is now a successful technology- enabled digital agency).

Almost eight years later, we are living in a fully programmatic world—but many of the benefits of programmatic futures have yet to come true. Today’s “programmatic” is still very focused on RTB, inventory pools are still murky, and technology’s ability to value publisher inventory still has a long way to go.  What’s missing?

The problem with today’s programmatic RTB environment is that the “exchanges” aren’t really true exchanges like we have in the financial markets. Although you can liken online inventory to stocks, the comparison is tough to justify. Lacking agreed measurement, value continues to be in the eye of the beholder. More importantly, the procurement process is still driven more by the buyers than the sellers. Private exchanges are starting to make inroads in terms of creating valid counterparty transactions, but the RTB pipes have not been engineered to handle the key aspects of transactional workflow.

The biggest, fundamental problem with RTB is that it values inventory in a singular way. In the open market, a 30-year old male car intender costs the same whether you find him on or Hotmail. Although tweaks in RTB with private transactions can enable premium inventory procurement, it’s not scalable. The right exchange should be able to value audience separately from everything else.

Another issue is the problem of valuing inventory over time. As a publisher with 30 days to go in my quarter, my homepage inventory may be worth $10CPM. But, the day before the quarter closes, that same inventory may be worth only $1CPM if I haven’t sold it yet. Today’s networks and exchanges enable publishers to set a solid floor price, but have trouble managing value dynamically. That’s because future publisher pricing is not being matched with visible demand. Ironically, the real-time nature of today’s exchanges actually limits a publisher’s ability to manage yield, because every impression is always chasing an immediate bid. A real futures exchange would enable a publisher to value inventory dynamically, so it matches the value set not by bids—but by buy orders in the system (real, stated demand for future inventory).

Although the demand side has it pretty good right now, a true programmatic futures exchange could be truly game changing. Yes, today’s exchanges are serious arbitrage machines. Because the buyer has access to the entire market, they have more information available to them to manage their investment. The problem is, in programmatic RTB, they are stuck with a two-tiered system: secure “premium” inventory through private exchanges and/or DealID functionality for branding and demand creation, and drive lower-funnel activity through performance-driven bidded buying in the wilds of the exchange. Ask any trading desk manager—it’s still really hard to get exactly what you want without going to guaranteed buys, cross-channel buying still requires multiple systems, and communicating the value of the “media investments” you are making to clients is near impossible, because everything is bought and measured differently. So, going from “media buying” to true “media investment” necessitates a true programmatic future exchange, akin to NASDAQ or the NYSE.

In such an exchange, publishers would be able to value their inventory by utilizing a combination of their existing rate cards and product catalogs (for selling advance contracts), and data from buy orders in the market itself. Just like in the stock market, prices would fluctuate based on the spread between bid and ask pricing—and the contract date. Publishers could therefore execute any type of guaranteed buy in such a system (direct sales) as well as have the exchange handle direct deals (“programmatic direct” and “private market”). This is because such an exchange would manage matchmaking, not the execution piece. This is critical. Today, we are watching systems built from the ground up to deliver ads try and go in reverse to manage the process of buying them. As we have seen, the rise of “automated guaranteed” platforms suggests that RTB is not quite cut out for the job.

Why would the demand side want a true programmatic futures exchange? First of all, a true futures exchange treats media as a true commodity—and makes it tradable. The beauty of a commodities exchange is that, once you own a future contract for pork bellies, you can sell it. In digital media, once you have bought a bunch of 300x250s in Appnexus, you are stuck with them. Arbitrage is not the same as futures trading in a regulated market. A true programmatic futures exchange for media would actually enable well-heeled buyers to leverage their scale to consolidate positions in media, and resell them in bulk (or in chunks). Think about that. Imagine GroupM buying the entire Q4 consumer electronics inventory in February. What would that be worth to another agency representing Sony as the holidays approached?

The bottom line is that, despite the power of RTB pipes, we are a long way away from seeing the platforms where addressable media will be traded in the future. Eight years ago, my bet was on a programmatic futures exchange, and I am still long.

[This post originally appeared on 11/7/2014 in AdExchanger]



  1. Chris, thank you for a very thoughtful article. I thought of leaving this comment on Ad Exchanger but it might actually be too long for that site. Please let me know if you think it’s worth posting there to stimulate a discussion.

    I think you said it perfectly:
    “Lacking agreed measurement, value continues to be in the eye of the beholder… The biggest fundamental problem with RTB is that it values inventory in a singular way. In the open market, a 30-year-old male car intender costs the same whether you find him on or Hotmail.”
    What could make two otherwise identical impression opportunities (same user, practically at the same time, etc.) have different values because they come from different sites, like Hotmail and Isn’t the answer in the metrics one uses to evaluate the ads/campaign? Unless we start seriously talking about and try measuring the real effect of the campaign in the form of user response and incremental revenue, these two opportunities have the same value no matter what proxy metrics we use. (Certainly, clicks are not the valid metric for user response to display ads, as has been shown by comScore, Dstillery, etc.) But even if we are able to measure the response, we will not be able to measure the effect of each individual ad or the effect of ads shown on a particular site: these measurements are necessarily statistical in nature, and there won’t simply be enough data points. So, we then need to define the features that distinguish one site or page from another, not necessarily in terms of the content (e.g. premium vs. user generated), but in terms of their value for display advertising. What are these page features: contextual category, viewability, sentiment, quality of the text, or something else? Once we have defined such features and are able to measure the dependence of user response on the values of these features, we will be able to discover which features have the greatest impact and then price impressions on each site/page based on the value of these features for a particular site or page. The price of the futures contracts will be determined in a similar way. I suspect that we might actually be in for great surprises as some “non-premium” sites might turn out to have much greater value for display advertising than certain premium sites. Let’s hope this true “quantification” of page quality will come sooner rather than later. It is simply necessary if we want the display campaigns to really work for the marketers.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s