Ad exchanges are not like the stock market and never will be…thankfully.

Update: Marcus Pratt writes about this topic in AdExchanger.

At the IAB Leadership conference in Miami a couple of months ago, there was a fair bit of discussion of RTB and ad exchanges.  Most of these discussions assume that ad exchanges will follow the evolution of the financial stock exchanges and become the primary method of purchasing online advertising (and one day, the story goes, video, mobile, even TV).  While this is a compelling narrative, there is one problem significant problem with this view – ad exchanges are fundamentally different from financial exchanges.  Consequently, any media strategy which relies solely on “exchange buying” is likely incomplete in very significant ways (Shawn Riegsecker provides some great color on this here.) This does not mean that exchange buying is not an important tool for digital marketing, but it should not be equated to equity trading as an unavoidable and stable end-state for media.

There are 4 ways which ad exchanges are fundamentally different than financial exchanges: “100% perishable”, risk management, infrastructure and transparency.

First and foremost, ad impressions are 100% perishable – most financial instruments are not.  In the financial markets, most purchase decisions are based on a present value calculations using time as an input, assuming there will be a future transaction which is paired to present one (i.e the future “sell” to the current  “buy”).  Financial asset purchasers are not interested in consuming the gold/stock/bond that they are purchasing – they purchase assets with the understanding that a future trade will determine success or failure.  Ad impressions are completely different – when you purchase it, you must immediately use it.  For example, if I win a 300×250 impression on ‘site A’ for ‘user 123’ on ‘Tuesday at 10:27 AM’, it is only available for 200ms.  You can’t sell it tomorrow, package it into a derivative, borrow against it.  Once the ad decision is made, it is gone forever.

Secondly, there are very few ways to collect a risk premium in ad exchanges; alternatively, risk management is a primary function in financial exchanges.  Financial “winners” correctly value assets in the future and optimize risk vs reward (usually using analysis and leverage).  There are a number mechanisms manage risk in financial markets including options, futures, indexes, short sales, etc.  In RTB/Ad Exchange world, the only way to manage risk is bid price – and since nearly all contracts are ‘voidable’, there’s literally no way to collect a significant risk premium by players with better information.  As it turns out, about the only way to do this is to move to the guaranteed marketplace (or ‘traditional buying method’ like up fronts) which resides outside of the exchanges.  There is some validity to life-time-value optimization in ad exchanges to manage risk, but this is not an actual mechanism of the exchange…and difficult to prove out.

The third point is virtually ignored in every discussion I’ve seen or heard on this subject – technical infrastructure.  Primarily, there is the issue of ad server ranking.  Since most ad inventory is controlled by publisher ad servers, there is a priority system at play which supersedes any sort of bidding.  A couple of leading media companies have done considerable work to unify these 2 concepts, but this does not address the implementation issue – publishers will create a priority for their advertisers which is ‘in front of’ the exchange.  That is, advertiser 1’s  $1 bid has preference over advertiser 2’s $1 bid.   One could make the case that this is equivalent to ‘dark pools’ in the equity markets (as noted above), but that does not accurately reflect the structure of the marketplaces.  Secondly, there are major challenges synchronizing data across bidding participants, which is just not an issue in the financial markets.  Essentially this is a reflection that we are bidding on consumers which change in value to marketers depending on their own place in the purchase cycle – i.e. I’m a lot more valuable to an airline when I am vacation planning than the other 99% of my life.  This is not the same as a share of AAPL, which is always a share of AAPL – it may change in value over time, but it does not suddenly become a pharma company for 2 weeks during flu season.  Ultimately, this issue should be solved through innovation, but without data synchronization infrastructure, the concept of aggregating data-driven demand in a marketplace cannot be realized to the extent that it is in the financial markets.

Finally, and probably the most immediate difference, is the issue of transparency.  Admittedly, this issue could be solved easily by the market participants, but the reality is that most inventory is ‘masked’ by the publishers.  Essentially, bidders cannot know exactly what they are bidding on.  The exchange proponents will say that this does not matter if you are using ‘smart enough’ technology, since one can track performance by a unique, anonymous publisher and section ID.  This naively ignores the fact that publishers can easily change which inventory these IDs represent in 10 seconds and there is no way for the market participants to tell that this has occurred until the value has significantly eroded.

To see how this undermines the financial market metaphor, imagine that you are purchasing shares of ‘mobile company 123’ for 6 months and it is performing great.  You bid at higher and higher rates for these shares based on this information.  Then without your knowledge, the broker switches out the underlying stock from AAPL to RIMM.  For a while, you continue to bid high for ‘mobile company 123’, your performance will (most likely) rapidly deteriorate while the broker makes huge profits.  Financial markets have disclosure rules to prevent this sort of behavior.    Only wishful thinking prevents this in ad exchanges (and maybe a verification service, but that’s another discussion).

In the end, we need to stop thinking of ad exchanges as some nascent stock exchange and that the ad whole industry will look like Wall Street in 10 years (thankfully).  The reality is that ad placements and financial instruments serve entirely different purposes.  To think otherwise is to engage in abstractions which will never be implemented.  More importantly, this metaphor hurts the development and acceptance of ad exchanges as we wait for them to develop into something they are not.

Ad Exchanges are a fundamental innovation in the digital marketing industry.  For marketers, they are a tool to help unlock the power of data and transform some channels of media buying.  For the industry, they providing a platform for technical innovation and entrepreneurship, which one could argue our financial markets are begging to prevent.  Isn’t that good enough?


About Jason Lynn

Product guy in NYC, enjoying the digital revolution.
This entry was posted in Ad Tech and tagged , . Bookmark the permalink.

7 Responses to Ad exchanges are not like the stock market and never will be…thankfully.

  1. Pingback: Google’s Mohan Makes More Ad Predictions; Bizo Opens B2B Exchange

  2. Phantom says:

    I don’t think the metaphor of the stock market was ever intended to be taken as literally as you explained here. It’s only ever used to educate those unfamiliar with programmatic buying.

    • Jason Lynn says:


      I wish I could agree but I have had too many people declare that non-exchange inventory purchasing cannot compete with programmatic buying. There is no durable proof of this – it is 100% conceptual based on the notion of market efficiency. There are definite efficiencies to programmatic buying, but it is important to understand the limitations.

  3. Phantom says:

    Agreed, and you make valid points in your article. My perspective is from the video side of things but I imagine the display exchanges by nature of their DR focus have more relevance to this metaphor. Great article, Jason!

  4. Seb says:

    I am far from being a specialist of ad exchanges but I liked this article very much. Interesting parallel with Stock Exchange.
    In you third point three you write: “Ultimately, this issue should be solved through innovation…” I’d say this apply although to you point two. Couldn’t we innovate and come up with derivatives of some sort to help manage Risk?
    As for your point four, don’t you think that some regulation could help? I am more of Libertarian, but some regulation could help in some extent…

  5. Jason Lynn says:


    Thanks for the comments.

    There are folks which have tried to create a derivatives-like market, but fundamental structure of the market makes these so frail that I can’t see them getting any traction. The main issue is that the market participants don’t think in these terms – sellers want to create premiums and scarcity through segmentation and customization, buyers want something that is indirectly attributable (brand awareness / product sales) or to arbitrage (ad networks, buying houses). Aside from the challenges of time shifting valuation by buyers (iow, advertisers don’t know the results of the campaign spend for several weeks after the fact), derivative products are moving against the interests and MO of the market participants. Essentially, the market players who control the buy and sell side would not allow a derivative maker to profit. Ad networks effectively do the closest thing to this through inventory aggregation, but they are far from having the characteristics of financial instruments and are actively watched by sellers as a potential threat.

    I could be wrong and would love to see an opinion where derivative markets exist _and_ account for these factors.

    As for regulations to enforce transparency, there are groups which do create standards, even ones around transparency, notably IASH in the UK. Assuming that this catches on (which IMO is highly doubtful), this could move us to situation where transparent ‘trading’ becomes more prevalent…but we’re a long way from this today.

    • Seb says:

      Thank you very much for such a clear and interesting answer. Last question: What about going the other way around and deal leads (i.e. use big data to profile buyers and sell the leads)?
      Have a good day,

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 )

Google+ photo

You are commenting using your Google+ 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 )


Connecting to %s