I was out with Josh Grotstein--CEO of Motionbox--last week, and got to talking about how the inefficient market design of our current ad exchanges saps liquidity.
OK, maybe I was talking about it.
Possibly I was ranting*.
Josh wisely changed the subject to an episode of 1987's Max Headroom, featuring the first vision of an ad exchange that we could think of, the steam-punkish Ad Market. It's not entirely clear what the underlying is in this market, since elsewhere in the episode a big advertiser seems to be using Skype to tell a network head to get the ratings up. It could, perhaps, be a pure risk-hedging derivatives market without any delivery of the underlying, which would certainly explain the show's dystopian setting.
Elsewhere in the episode, Amanda Pays makes those 80s style big-shoulder suits look pretty damn good.
--------------
* The thing about a rant is that it can't be kept in. So here it is.
Liquidity in a marketplace is generally determined by depth of limit order books. If the limit order book gets depleted, there is no liquidity. Most research on this subject focuses on the ratio of limit orders (supply liquidity) to market orders (deplete liquidity). This ratio depends on trader patience and market design.
Market design decides time to execution, the key endogenous factor traders take into account in deciding whether to place a market or limit order (mappable, perhaps, to the more classical 'waiting costs.') Variability is a function of spreads, spreads a function of liquidity, and liquidity of time to execution. Highly variable prices drive down the average price level.
In a well-designed market, time to execution can be minimized even with low volumes. Many marketplaces have historically kept TTE low by opening only for a limited period of time every day or every week. By concentrating order flow in a smaller period of time, the chance that each trade would clear quickly was increased.
Our current ad markets seem to be headed in the opposite direction. I consider these markets low-volume. But instead of working to decrease TTE by increasing the concentration of order flow, they seem to be motivated instead to decrease inventory risk by imposing a small TTE, diluting order flow. This depletes liquidity and must, in the aggregate, lower CPMs.
Of course, this is all speculative on my part, since I don't have the data to back it up. Also, this effect is probably a small part of why online CPMs are so low, the relative inefficacy of online ads being the bigger part. But liquidity is certainly more interesting to think about. It also makes me wonder what the heck Hal Varian does at Google.
I'd sync in Brian O'Kelley on this one...
ReplyDeleteThe one thing about ad exchanges is that while they are highly inefficient (will go into some reasons in a minute) they are completely liquid in the sense that there is always a buyer (at least on dCPM) and the inventory perishes so there needs to be an order - which is I think ad exchanges have so much potential, 'instant' sale.
They are no accurate for a bunch of reasons including but not limited to:
1. Lack of quality inventory
2. Terrible publisher driven inventory controls
3. No way to prebuy inventory then flip it later on (where's the futures market?)
4. Too complex order execution systems (where's clean, simple UI and analystics) - back in the 80's Jeff Pulver created a financial software that let people build models in Lotus opposed to using a Bloomberg or something custom. Where is a simple system?
5. No good training (where are the day trading seminars)
6. most volume is driven by arbitrageurs (affiliates) which is a good thing in that it gets activity its a bad thing in that the more middlemen the less transparency (not even talking about efficiency) and the less transparent, the header it becomes to build trust
7. Scummy/scammy ad networks and scummier (but good converting) ads, did someone say Acai Berry? (ps. I am a direct marketer and respect the revenue but seriously, we all know the offers are crap)
8. Inability to negotiate terms adequately
9. The largest ad exchange (right media) isn't a true exchange even, its a hub and spoke model (brilliant btw) like an airline. Inefficiency is built into the system.
10. As a result of all the risk levels , publishers leave off good inventory (back to #1) which results in shitty advertisers running shitty inventory (still making money but leaving off most of the industry.i