I've been hearing a new battle cry recently in online ad circles. The call for dis-intermediation, and the efficiency it will produce. The online ad network market provides great examples of intermediation as a flawed, but necessary part of the market ecology.
I have to confess, I used to think that dis-intermediation was the only way to achieve true efficiency in the online ad marketplace. I'm revising that stance. It's not that we need to dis-intermediate so much as we need to make the intermediaries more efficient and accountable.
I recently posted the thoughts below in a forum where a discussion was happening on how competing ad networks buy and sell inventory from each other, and the impact on publishers. I figured I'd post it here.
The primary questions were:
1. Why would networks share distribution with direct competitors?
2. What should end publishers do about this?
Ad networks have been trading inventory with each other pretty much since the
beginning. Consider an example. At any given time, Network A may have
more advertiser demand than supply, while Network B may have supply
than demand. Both problems can be solved by Network A buying from
Network B. Network A gets more revenue b/c they deliver on campaigns
they can't deliver on their own inventory. Network B gets more revenue
because they are able to sell space that would have been sold for less,
or gone unsold.
From the network perspective it makes a lot of sense.
It's a little trickier from the Publisher perspective.
Network B will argue that their publishers make more revenue than they
would have if they hadn't sold space to network A. This is true... BUT,
the publisher might argue that they would make more money still if they
had a relationship directly with Network A, and got the buy directly
from the seller. This is also true...and that's the quandary.
Networks generally do their best to create efficient markets, but even
the largest networks do not get on every ad-buy. Sales force,
relationships, market perception all drive buyers to different sellers.
Market theory tells us that the broader the market is, the more efficient
the market will be. Networks share distribution as a means to broaden
the market, and to find the efficiencies in the non-overlapping ad
buys. However, the way this has historically been done is quite
inefficient for the end publisher.
Back to the example.
If I'm a publisher who has a relationship with Network B, I'm likely to
notice ads from Network A serving on my site. My logic would likely be - if Network A has enough demand to be buying from Network B, then I
should consider moving my inventory to Network A. This makes a ton of
sense. Assuming that Network A is taking a 30-40% cut of the ad-buy it
is brokering to Network B, that means diluted earnings to me, the end
publisher. Logically, I'm going to try to move my relationship to
Network A.
But here's the rub. What happens when market dynamics shift and Network
B has more demand than supply and Network A has more supply than
demand. Roles are reversed, and after going through the effort of
shifting my inventory from Network B to Network A, I'm now faced with
shifting it back. It's a lot like trying to momentum trade stocks. You
run the risk of sub-optimizing by missing the supply and demand shifts
that occur often, and very quickly.
It's a tricky problem.
What a publisher really needs is the ability to have a relationship
with both network A and network B (and networks C, D and E as well) and
to have access to their combined advertiser reach. In an ideal market,
inventory will flow seamlessly based on the real-time demand economics
of both partners and the ad-impression will be sold to the highest
bidder on each ad-call.
In order for this to happen, our market is going to have to start
behaving a lot more like the financial markets. What I'm talking about
is single platform efficiency - the ability to see all the bidders for
each impression at the time of the ad-call, and determine for each
unique ad call who should receive the ad. At it's most basic, the concept is to allow the seller to see the broadest possible market which will likely be spread across multiple intermediaries. It's too easy to say that
intermediation is inefficient. In fact, intermediation is necessary in
almost any market (when was the last time you bought a stock directly
from another seller?).
We don't need to eliminate the intermediary to acheive an efficient market - we just have to change how the intermediary operates.