In an interview in the Guardian, I noted that Amazon seemed to be varying prices frequently during the day, so that a shopper that checked back would see different prices. Managing director Brian McBride of, the British version of the site, denied that Amazon did that. There is some confusion; my assertion concerned the US site and his assertion was ambiguous about whether he meant the UK site or both sites.

The UK site is very upfront about its marketing partners, clearly identifying when the prices are from the Amazon itself or another company, and listing several buying options. The US site is not. (The UK site is to be commended for its honest and transparent site design.)

Consider this screenshot (warning: pdf) of the Amazon site. This is the typical result of a search for the Sony NP-BG1 in Cameras and Photos. The Sony NP-BG1 is a camera battery. The result of this search is a single battery priced at $41.95. Underneath is fine print that says "20 Used and new from $31.75," but does a used camera battery sound like a bargain at $31.75? Not to many customers, since the typical reason that one buys a replacement battery is that rechargeable batteries wear out.

I tracked the results of this search over the course a week. This picture (warning: pdf) shows the prices graphed over the course of June 19-20, 2006. The price variation is really quite remarkable, exceeding 7%. Generally prices changed every hour, but occasionally there were short-lived price cuts (around 5 minutes), although not during the two days graphed. Prices went even lower than illustrated on this particular graph briefly on a subsequent day. On other days, the prices were the same all day.

What appears to be transpiring is that Amazon features a price of a marketing partner, and is varying across its partners. These features were never the best price offered, but the better prices are hidden in the fine print. The construction of the US site, unlike the Amazon UK site, hides the fact that there are other offers available.

I identified this particular example by searching 17 products picked from the second page of the list of most popular items in various categories; within a single day three of the 17 items had some variation. This one was the most extreme illustration.

I was traveling at the time the Guardian tried to contact me and unfortunately they printed Amazon's denial without getting an answer from me about whether I meant the US or UK site, or even what the evidence is. I think the evidence speaks for itself.

Here is the actual quote in the Guardian:

Question: Is it true that Amazon offers different patrons different prices?

"Yes. If you search on Amazon, you find they're doing something rather subtle. They don't seem to be targeting prices based on your identity. What they're doing is that with some items they're offering prices that fluctuate - but you don't get the fluctuating prices if you don't clear your cookies. I'm not exactly sure what they're doing, but I have been able to replicate getting different prices from Amazon on different machines within seconds of each other. In general, it seems that people who check prices frequently get better prices than people who don't."

I stand by my statement. The price variation has *nothing* to do with propagation of changes across the internet.