Lately I’ve made a good few purchases from Amazon’s ‘Amazon recommends’ section on my account. I just made one more as I type, and I clicked on another one.
It’s starting to get a bit worrying, I’m spend spend spending. But on the plus side I’ve got/getting lots of stuff to keep me occupied!
I guess it started when every item on the list was an Atheism related book and I bought nearly all of them.
How well does Amazon know you? How many purchases have you made from Amazon’s list of recommendations for you? And did those purchases turn out to be money well spent?
(p.s. gotta love Firefox’s ability to spell check in web forms!)
Not too well, as I do a lot of necessary work research for books that have nothing to do with books I am personally interested in buying. I have to fix my recommendations a lot.
Hmm. Amazon recommends that I buy a lot of Bob Dylan, a little Loreena McKennitt, two books of cryptic crosswords, The Faerie Queene, The Unfortunate Traveller and Other Works by Thomas Nashe, and King James I’s Demonology.
I’d say they know me very well indeed, if you forget about the fact that I already own most of these items
When it comes to something I’d like someone to buy for me, they know me better than my friends, for what that’s worth. And that’s not much. If Amazon were to buy me a present right now, it’d be 50/50 based on my desires or the desires of someone else I shopped for. No, I don’t want a cat calendar; that was for my cat-freak friend.
My company does software that does this. Not for Amazon, but for many of their competitors.
The more data we have that correlates purchases of one item to another, the more we can accurately predict what may appeal to you. The more items in your purchase history, the more we can tailor the recommendations to you. There are also item-to-item correlations that aren’t personalized but just recommend titles/items that relate to the target.
It’s all a numbers game, grinding lots of data and coming up with correlations that are borne out by the purchase history.
I take the time / click the box to ask Amazon not to base recommendations on anything I buy for gifts, as well as click “not interested” ass appropriate; consequently, the suggestions are often great. I loves it.
In Amazon’s case, much like Netflix, you can get even better recommendations by rating books you’ve already read.
Amazon knows me very well. Currently there are five different types of 120 film, four different Holga accessories, a Diana+ camera, two different Nikon lenses, a speedlight, a lens cleaning kit, and a book on digital photography. Unfortunately, Amazon doesn’t realize that I already own several of the things on the list. Oh well, more camera equipment never hurts and I’m just going to assume that amazon knows I need two speedlights.
Amazon knows me not at all. I’ve never done business with them, and never will, precisely because they made it clear that they consider their customers’ information to be a commodity.
All of this can be done within a company’s data; there’s no need to sell your data to anyone. Besides, by the time we get it there’s no personally identifiable information. You’re just numbers to us, no one could track it to you.
Yeah this is why I make great use of the “wishlist” function. It lets me bookmark all the crap I want without having to actually buy it. Then xmas and bday time I can look over the whole list to see what I really want the most. I do something similar in real life - when I see something I “must have!” I take a photo of it and that satisfies most of my immediate greed. Works especially well with collectibles - I collect turtles, but I don’t have the room or money to buy every turtle knickknack, poster, and tshirt out there.
My Amazon knows me ok, I guess, but not in an uncanny way like “my tivo thinks I’m gay”. Usually it will recommend something obvious like a book or cd by someone who has authored other things I’ve bought, or a sequel to a movie I bought. I find it most useful in wishlist mode, to add multiple items that are similar in nature. For about two weeks the “items similar to what you just added to your wishlist” page was replaced with “most popular items on Amazon page” and I was very unhappy, but then it got restored and all was well with my universe. I love the user lists too!
One thing Amazon doesn’t know about me, and which doesn’t fit their predictive software, is that I have an interest in many varied, different things. Their software asumes that if I like X, I will like more things like X and not much else.
It’s tough to make recommendations based on multiple items simply because of the sheer number of possible combinations. For a site like Amazon that carries a few million items, they’d have to consider trillions of pairs of items, most of which have never been bought by anybody.
Not too well. I get a lot of recommendations based on gifts I’ve purchased for other people, so it’s always bothering me about Nintendo DS games and other things that I have no interest in. Also I bought a DVD movie or two and now it’s bugging me about new Blu-Ray movies all the time. I wish there was a selective opt-out on stuff like that, it seems to be all or nothing.
As you can see, there’s a lot of complicating factors to recommendations. I’m not sure how the Amazon system is implemented but you may be able to get around the gift problem by rating those items poorly. Most systems that take ratings into account won’t recommend poorly rated items. Many also have a “Don’t Recommend this to me Again” feature for exactly this purpose.
Amazon gives you the option of not using a specific item for recommendations. It’s a nuisance to go through and edit them all, but you’ll get better recommendations if you do.
I just wish that I could tell Amazon not to offer me books by certain authors ever again. No matter how much other readers of books I’ve bought liked them.
It also recommends things that I have, on more than one occasion, checked as “Not interested” or “I own”. How many fucking times do I need to tell it that I don’t want to read any of the Eragon sequels? (I’m still trying to bleach the first one out of my mind.)
