Let’s take a moment and consider one of the more interesting of the pricing models – the auction.  And to see this model at its full potential, we’ll want to analyze the real world example of the company that got auctioning right, and is currently leveraging it to pull in billions of dollars every quarter: Google.

Yes, everyone’s favorite search engine is a giant auctioning machine.  You sign up for Adwords, put in a maximum bid, and BAM!  You’re in the largest and most sophisticated automated auction in the world.  There are many beauties and subtleties of the system.

Subtlety One

One, there is almost infinite segmentation.  If you can search for it, you can put an ad up for it – and with people searching for things like “pugs for sale in east sussex”, “ornithine transcarbamylase deficiency”, “my hamster is afraid of me”, and everything in between, there’s really no limit.  So how do you efficiently price an infinite amount of different ads? The automated auction.

Subtlety Two

A second aspect of Google’s Adwords sales is that the transactions need to be made quickly and effectively, because so many of them need to happen daily (on the order of billions per day, I’d estimate).  You therefore must have the auction automated, or else you could never hope to reach the scale necessary to accomplish the task.  So what else does the automated auction do for Google?  It provides the ability to scale.

Subtlety Three

What about a third aspect?  How about the fact that it’s impossible to know what the actual value is for a given ad space for a set of keywords at any given time?  How do you set a price for something like that, especially when it’s changing so often?  Again, the answer – the automated action.

It really is a genius concept.  But now we must be dutiful innovators and ask ourselves, what else could the automated auction be applied to?  Or, what other markets and product sets have the same kinds of characteristics as the three we just enumerated above?


How about the insurance market?  Do we see infinite segmentation?  I think we do – you have people’s age, socio economic status, yearly income, health levels, gender, risky behaviors, etc. etc. etc., and all the combinations of all of those things – all of which could be used to determine a premium for health insurance, car insurance, home insurance, life insurance, etc. etc. etc.  Do transactions need to be made quickly?  Perhaps not on the daily basis that we see happening with Google, but with the number of people that could use an insurance policy (everyone) and the frequency that some of the above factors could change, it may be useful to have a system that would allow for more transactions to take place without demanding extra time.  And the third – impossibility of knowing the exact value of any given set of insurance policies for a given individual.  That one is in the bag – even with all of the sophisticated data crunching algorithms the industry uses today, we still haven’t quite been able to pin it down.

I wonder how the concept of the automated auction could transform the insurance industry.  (or any other industry, for that matter, that has a semblance of these same three characteristics.)