Amazon Heuristics

It’s all a matter of mining the shopping data. Amazon knows quite a bit more about me than I thought they did.

From the Wikipedia article on Heuristics:

Heuristic … (from the Greek … for “find” or “discover”) is an adjective for experience-based techniques that help in problem solving, learning and discovery. A heuristic method is particularly used to rapidly come to a solution that is hoped to be close to the best possible answer, or ‘optimal solution’. Heuristics are “rules of thumb”, educated guesses, intuitive judgments or simply common sense. A Heuristic is a general way of solving a problem. Heuristics as a noun is another name for heuristic methods.

From an email today: has new recommendations for you based on items you purchased or told us you own. has new recommendations for you based on items you purchased or told us you own. has new recommendations for you based on items you purchased or told us you own.

If you match this up to my interests as expressed in departments on this website, it would seem Amazon’s batting average is getting more sophisticated in computerized matching¬† to my purchasing habits. If this same email had come from Wal-Mart based on my Amazon purchasing habits, I’d be writing my congressional representatives. As an Amazon customer I’ve seen for years how they match their recommendations to items I’ve ordered, from mystery novels to wristwatches. I have never seen any evidence of Amazon selling my purchasing information.

Let’s see how well Amazon scores against what I know of my interests and purchases.

Here’s my analysis of how Amazon’s recommended items ended up on my email:

  1. The Help. I never heard of this book title or author. Publisher Penguin could be a match on almost anything.¬† But it’s a match: customers who bought this item also bought The Girl with the Dragon Tattoo (Vintage) by Stieg Larsson, which I bought and liked so much I ordered another one in the series. Score: 1
  2. Orion Moon Filter. I have never ordered telescope equipment from Amazon, but I did order several books on Astronomy several years ago. Score: 1
  3. Floodplain: CD by Kronos Quartet, who published a CD of string quartets by Philip Glass which I ordered quite some time ago. Score: 1
  4. Requiem, CD by John Taverner: I have never ordered this composer, but it is still a match to the Philip Glass CD order because several items were recorded by the Royal Liverpool Philharmonic in both cases! Score: 1
  5. The Moment’s Energy: I have never heard of CD artist Evan Parker, and to be honest I can’t see the reason for this match. Score: 0
  6. The Landmark Xenophon’s Hellenika: matches The Landmark Herodotus, ordered a couple of years back.¬† Score: 1
  7. The Ivory Grin: matches Steig Larsson publisher Black Lizard.
  8. Lost On The Way: contemporary Jazz CD, artist Louis Sclavis unknown to me. Unsure what they’re matching on, probably “Jazz”. Kinda pretty tracks, a bit too new age for me. Score 0.5

This gives Amazon a score of 6.5 out of 8 (81%, a solid “B”). But alas, these heuristical guesses rarely result in extra purchase revenues for Amazon. I decide what book or CD I want to order based on what I’m already interested in at the moment, and that’s hard to anticipate in a computer program.

An ad by any other name is still an ad. But it’s fun to look at or listen to the recommendations, and once in a while it actually does drum up new business. That would probably only work when I’ve made a decision to accumulate more of a certain favorite author, recording artist or genre.

1,390 total views, 1 views today