It's Time For The Futures Exchange

Networking"Networking" is written by members of the online advertising network community.

Today's column is written by Andy Atherton, COO and co-Founder,

My former boss, Wenda Harris Millard, had one of the most quotable, and quoted, lines in the online media industry in 2008 when she said, "We must not trade our advertising inventory like pork bellies", referring to frenzied interest in online media exchanges – interest which for good reason shows no signs of abating.  Her remark was intended as a warning to publishers to maintain differentiation and manage indirect sales channels carefully.  But consciously or unconsciously, Wenda also surfaced a fundamental distinction that, oddly enough, doesn’t get talked about in the context of online media:  the difference between a Spot market and a Futures market (media contracts are not yet exchange traded, so technically this would be a  “Forward” market, but I will ignore this subtlety in the interest of simplicity).

Since Futures market buying (e.g., the TV “upfronts”) is the standard practice offline for Brand advertisers, and Brand advertising is (rightly) seen as a huge growth opportunity for online, it behooves those of us interested in catalyzing that growth to identify where current market capabilities and technology infrastructure are lacking.  To do that, let’s take a deep dive into a specific example, moving  beyond the fashionable, but vague, notion of “Madison Avenue meets Wall Street” to the technical and operational details required to make this actually happen.  So let’s (1) walk through an example of a Spot media transaction, (2) replay the same example in the context of a Futures transaction, and (3) evaluate the key capabilities required to enable Futures transactions for online media.

To keep it simple and to adhere to our financial market metaphor, we will assume that both transactions take place on Google’s DoubleClick Ad Exchange 2.0 (“AdX”).  To further simplify, we will express the example in terms of a single impression, even though an actual transaction typically includes millions of impressions.

(1) A Spot Transaction: The buyer (typically an agency) selects targeting criteria and expresses a willingness to pay for a particular impression.  The targeting criteria can be “objective” (e.g., demographic, geographic), or “subjective” (e.g., past purchase behavior, purchase intent) and willingness to pay can be expressed in terms of CPM, CPC or CPA.  This “bid” is transmitted directly or through an intermediary (network or DSP) to AdX, either through the UI or a real-time bidding (RTB) interface.  AdX only accepts bids with objective targeting and CPM pricing through the UI.  All other bids require some sort of translation (e.g., deciding what constitutes “purchase intent” for a user, predicting how expected CPA translates into a CPM bid, etc.), most often involving the use of a vendor-specific demand-side cookie, either by the buyer or the intermediary.  AdX compares all bids for each impression; the winner is served a re-direct allowing its ad server to deliver an ad to the requesting page.  AdX records the buyer, the seller and the price of the transaction, then moves on to the next impression.  AdX provides up to date reporting on the outcomes (pricing, delivery) of this impression-by-impression auction, as well as centralized payment processing and some limited protection from counterparty credit risk.

(2) A Futures Transaction: For simplicity, we will assume the identical scenario above, except that this bid is for guaranteed delivery of a specific impression 3 months from now.  This is the type of contract large brand advertisers routinely sign directly with content providers offline, and a handful of online publishers with both the size and sophistication to enable confident future transactions.  However, this simple difference in timing has dramatic implications.  (a) The seller isn’t soliciting bids for an existing impression – it must be sure it will have the desired impression available in 3 months.  (b) With no current ad serving opportunity (including cookie, etc.) the parties must agree on some other way to describe this future impression exactly and to determine what it will be worth to them.  (c) Each party must be confident that the other will be willing and able to honor today’s agreement 3 months from now.

These distinctions between agreeing now to transact now (spot) and agreeing now to transact in the future (futures) drive fundamental differences in the capabilities, and thus the technological infrastructure, required to make an Online Media Futures Market work.

(3) Key capabilities required to enable Futures transactions for online media:

  1. Forecasting: For the spot market, forecasting is helpful.  For the futures market, it is critical.  Without accurate inventory and price/cost forecasts, future delivery guarantees with sustainable economics are impossible; forecasting is the difference between a profitable business and out of business.
  2. Standardization: As I mentioned in the example above, the vast majority of the spot market business today relies on the use of a “subjective” demand-side cookie to value an impression and generate a bid.  However, it impossible for a seller to predict future instances of a buyer’s cookie with enough accuracy to provide a future delivery guarantee at any scale.  Furthermore, performance metrics (e.g., CPA/CPM) that are fundamental to value for spot market buyers change dramatically over time and in ways not visible to the seller (different offers, product prices, creative, etc).  Even more problematic is cookie churn which means that most of the cookies that could theoretically be used to specify a transaction today won’t even exist in 3 months.  For these reasons, a stable, non-vendor-specific (i.e., objective) set of targeting criteria is required for the futures market (e.g., Context, Demo, Geo, Quality).  While these objective descriptors may seem “old fashioned” compared to some of the more futuristic online targeting capabilities used in spot market buying today, they map well (indeed, identically) to the targeting criteria large Brand advertisers use and have used for a century across their offline media.  Objective targeting criteria are so widely used because data consistently shows they work.  Operationally, they can be efficiently and predictably deployed across different media types and providers.  Most importantly, though, they move cases of product profitably offline, where >95% of retail commerce still takes place.  Marketing mix models and panel-based studies confirm that effective management of objective targeting criteria and frequency is fundamental to driving high ROI offline sales. 
  3. Clearing: A lot can happen in 3 months.  For a futures transaction to work, each counterparty must be confident, operationally and financially, that the other will honor the terms of the deal when it is time to deliver.  Operational confidence means believing that the counterparty is capable of honoring its commitment - primarily, the buyer must believe the seller can actually deliver inventory exactly meeting the specification in the contract during the specified delivery window.  Financial confidence is more of a two-way street.  Buyer and seller each must believe that the other will choose to honor the contract and have the financial resources to do so, understanding the high likelihood that the spot price on the delivery date will represent a better deal for one of them than the contracted futures price.  Anyone questioning the importance of this financial confidence need look no further than the huge spot price swings over the last year.  If a buyer agreed in January of 2009 to buy impressions for April 2009 delivery at “50”, could it be confident that its counterparty seller would fulfill its delivery obligation in April 2009, when the spot market was at “75”?   In working futures markets, the financial integrity of transactions is generally ensured by a clearinghouse that (among other things) holds collateral from both sides until they have met their obligations under the futures contract. 

OK, the spot market and the futures market are fundamentally different.  So what? These concrete, fundamental differences between spot and futures are important to understand for a number of reasons.

First and foremost, we want the online Brand media market to become large and sustainable.  Markets of any type do not become large and sustainable without a high level of efficiency, standardization and predictability.  Large corporations with complex operations cannot and will not tolerate supply disruptions or quality problems with key supply chain inputs whether they be corn, oil, steel – or media.  Most large corporations view futures as important tools to reduce volatility in the cost of their commodity raw material inputs.  As the same procurement teams that rely on corporate commodity hedging programs get ever more involved in media purchasing, this “futures” mindset will transfer to the raw material of media as well.

This is not pure supposition, the futures market is how large corporations already manage the majority of their brand media buying offline (where 94% of US Brand advertising spending occurs).  Like it or not, large corporations operate this way because it makes sense for their businesses.  So as online becomes a more material portion of the media mix (and/or the distinction between online and offline blurs), we should naturally expect to see the futures model become much more important.  Increased involvement of procurement teams in online media purchasing will only accelerate that.

Despite the fact that futures is the logical way to transact online Brand media, the online media market to-date has largely ignored the distinction between spot and futures, even as it wonders “where is all that Brand money?”  It’s time for that to change.  Developing an online media futures exchange beside/atop the spot infrastructure that exists is the critical next step for online media.

Follow (@brand_net) and (@adexchanger) on Twitter.


  1. Nice piece, showing where the "Madison Avenue meets Wall Street" depiction starts to fall apart.

    A futures market driven primarily by brand advertisers would certainly be interesting; one can imagine a brand buying impressions on a futures basis, then upon delivery of the ads, launching an immediate follow-up campaign via the spot markets. The follow-up campaign would be optimized, by focusing on the audiences in which the futures campaign saw greatest traction (as measured by engagement with the ad, a visit to the brand's site after viewing the ad, etc). Assuming they could determine the most interested audiences, could that be attractive for a brand advertiser?

  2. Andy, great article. Quite a bit has been written about whether financials analogies hold up to our marketplace but with what you've written above, the "futures" market is within reach.

    A few things:

    1. I do think you can have brand dollars within a spot market. Not everything you need to do for brands needs to be a futures market.

    2. I see a futures market around non-standardized creative units (i.e. takeovers, dominations, etc)

    3. In this ecosystem, who plays Moody's?


  3. A futures market seems to map well for the large publishers of content who have enough inventory and volume to provide forecasting and the economic incentive to lock in inventory acquisition months ahead of time. Certainly the publisher yield systems need to offer more automated ways of dealing with futures buying as a complement to the Exchange inventory.

    As you pointed out the difficulty will be excluding the transient and client specific data which makes the RTB environment so valuable. Often this data is time sensitive and unique to the Brand's intelligence (think re-marketing).

    As advertisers price this correctly it should increase the yield for publishers across the board. If we look to the Google example it is a completely live and dynamic market that has dramatically risen in CPM pricing over the years as advertisers learned the value of the inventory. If publishers see the same yield increase from the live environment it could hamper futures acquisition.

    The other issue I'm not sure how to deal with is the long tail. How can the brand buyers lock down inventory that is happening across the long tail of sites? Even if they can take care of the major publishers the aggregate reach and impression volume of the long tail might be as high as 80% of total volume. Is this something that would have to be guaranteed by the exchange itself?

  4. Thanks for the thoughts and comments.

    I think what you’re getting at here is the classic marketing funnel. So yes, brands absolutely want and need to work the full funnel online. Today they just have more tools for working the middle to the bottom vs. the top to the middle.

    @Darren Herman
    1. Agree that not everything for brands needs to be futures. Brands can and will do some spot buying, but I believe the majority of spend will still be planned in advance.

    2. There could also be forward market in non-standard executions - there already is really - although I would remind everyone of my disclaimer in the first paragraph that non-standard, non-exchange traded forward contracts are not technically “futures” - one of the trackbacks clearly missed that.

    3. This is a great question (although given the strong ratings Moody’s gave many soon-to-be toxic assets this arguably isn’t a critical piece of infrastructure). There will need to be automated assurance that inventory provided in fulfillment of a contract meets the agreed specifications, but I think that’s a bit different than what Moody’s claimed to provide. I think this objective spec-checking service could be provided by the exchange itself or by a third party, but I think assessments of “value” will be hard to do from a third-party perspective as they vary based on the counterparty (the “subjective” discussion in the article).

    @ Michael Andrew
    I agree that large publishers are more likely to be early primary sellers, but (to your question about how smaller publishers can participate) I also think there’s a market for intermediaries who have the meta-forecasting and delivery management capabilities to sell futures, delivering an aggregation of inventory from smaller publishers (or of smaller blocks from large publishers). Also want to make the point that both spot and futures markets are critical. Definitely not making the argument that the spot market isn’t tremendously valuable, just that it’s an incomplete solution.


Add a comment

XHTML: You can use these tags: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>