The Lies of Efficient Frontier

In June I published an article at The Search Agents’ blog called ‘Dirty Little Secrets of Portfolio Theory‘. The post attracted a few comments from employees of Efficient Frontier, the company whose methodology I examined.

In one of the comments on my post they said, “Efficient Frontier literally calculates and predicts outcomes for every possible combination of bids to identify the optimal yield scenario at any given point in time.” Of course, this is completely false.

For AdWords bids from $0.01 to $100.00, there are 10,000 possible bids. (Actually, Google permits bids to be for a fraction of a cent, too, so there are really more than 10,000 possible bids. But, for the sake of argument, let’s say that there are 10,000.) Now, if an account has 2 keywords in it, then there are 10,000 possible bids for the first ad and 10,000 bids for the second ad.

That means there are 100,000,000 (one hundred million) possible ‘combinations of bids’ for an account with 2 keywords. But most accounts don’t have just 2 keywords, they have thousands. By Efficient Frontier’s own reckoning, there are about 40,000 keywords in the average account they manage. So, there are 10,000 multiplied by itself 40,000 times (or 10,000^40,000) possible combinations of bids that EF needs to examine…per account…each day.

This is simply impossible. For comparison, there are 10^6 dollars in a million (10 multiplied by itself 6 times) and 10^9 in a billion. There are estimated to be about 10^78 protons, neutrons, and electrons in the entire Universe. So, there are about 10^79 particles in 10 Universes. And a googol (after which Google is named) is 10 multiplied by itself 100 times. So, even if every proton, neutron and electron in the Universe was a computer processor, Efficient Frontier couldn’t possibly examine every possible ‘combination of bids’ in less time than the Universe has existed, let alone a single day.

However, based on the claim that they actually can do this, Dr. Siddharth Shah of Efficient Frontier has repeatedly said that “Portfolio theory GUARANTEES THE BEST PERFORMANCE FOR ANY GOAL SUBJECT TO ANY BUSINESS CONSTRAINT” (capital letters in original). Last week, at Search Engine Land he repeated this assertion when he said that EF’s approach is a “a rigorous mathematical method that guarantees the best outcome for any goal.”

When I said in ‘Dirty Little Secrets of Portfolio Theory’ that EF’s approach actually does not by itself guarantee optimal performance, an Efficient Frontier employee wrote that my claim was “a red herring argument. Any smart marketer would make use of the best math and technologies on the market, and utilize manual oversight/strategies to optimize performance.”

Now answer me this: If bidding by Portfolio Theory, as Dr. Shah claims, is guaranteed to produce the optimal bid set, then why in the world would EF’s system need “manual oversight” and manual “strategies to optimize performance”?

It seems to me that Efficient Frontier is simply unable to get their story straight.

~~~~~~~~~~

Note: As I say on my ‘About Bradd’s Blog’ page, “The purpose of this blog is to cover topics that can’t be covered well in my blogging at The Search Agents, Search Engine Land and Search Engine Journal. Nothing I write here should be interpreted to reflect in any way on those other fine sites.

Advertisements

  1. 1 Exploring the world of Google Metrics « excapite

    […] an insight into how we may discover the clues to finding  the right answer in his recent post on The Lies of Efficient Frontier. In this post Bradd provides us with an exploration of the economics of the AdWords Auction process […]




Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com 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 )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s



%d bloggers like this: