My Amazon Experiment, Part 1

Back in January, I mentioned that I was doing some experiments on Amazon Marketplace to see if it is possible for for individual sellers to influence the market price for an item.  After very limited success, I have concluded it is not possible.

First of all, I'll explain my hypothesis.  For those of you who don't know, Amazon Marketplace is a place for individual sellers, such as I, to sell their products on Amazon.  In return, Amazon gets a cut; I will explain Amazon's fee structure in a future post.  I hypothesized that an individual seller could attempt to raise the Amazon market price for an item by changing the price of his or her item frequently enough.  Obviously, this would only work for items for which there were relatively few sellers.  I experimented with my TI-92 graphing calculator and a copy of Sonic Adventure 2.  Both items had twenty or fewer sellers during my experiment.

When initial viewing the demand for Sonic Adventure 2 on Amazon (I was trying to determine a good initial insertion price), I noticed that prices tend to cluster.  For example, the first seven prices could look like this: $25.00, $25.25, $27.00, $27.00, $27.00, $27.01, $27.99.  The pattern repeats as you go higher in prices; the following four prices could be $28.50, $28.99, $28.99, $29.00.  These clusters of close prices are the floors I mentioned when I first introduced my experiment in January.

I noticed that if sellers noticed a new cluster being created, sellers with prices below the cluster would raise their prices to just below the cluster, meaning that if the prices were $20, $21, $23, and $24 and another seller inserted the item at $23, the price breakdown would frequently shift to $23, $23, $23, $23, $24

This concludes my initial observations.  Next post I will discuss my methods and why, eventually, they fail.