April 21st, 2006
In recent user tests, we’ve noticed something very peculiar. Something I call the Amazon Effect.
As we do in many of our studies, we allow our test participants to create their own tasks. This way, they’re much more likely to be shopping for something they’re truly interested in, as opposed to going through the motions for testing purposes. We also give users real money, which makes the incentive to purchase very strong. If someone is truly motivated to purchase, then any problems we uncover that prevent purchasing directly affect the bottom line.
Despite the freedom we give test participants to create their own tasks, we have to be more strict about which sites they visit. The reason is straight-forward: we’re testing the effectiveness of particular sites for a particular purpose. In order to see what works and what doesn’t, we need to watch several people on the same site (especially the site of our client). Otherwise, we wouldn’t have a large enough population to be sure that we weren’t just seeing outlying issues that only affect one or two people.
Amazon.com, however, can make this difficult. When we test sites that sell similar products as Amazon does, (a list that is always growing, as Amazon seems to sell everything nowadays), we always get testers who want to go to Amazon before they go to the site we ask them to go to. “Can I just check out Amazon first?”, they ask. Because we’ve already included Amazon in the testing, we must remind them that we are also interested in learning about other sites, and that they may or may not be asked to visit Amazon later in the test. “OK”, they say reluctantly.
I call this the Amazon Effect. Amazon is so strongly rooted in people’s minds that it overrides their desire to try out other sites, even if it’s during a test! And it was funny at first, when we would have to say things like “well, we may or may not ask you to test Amazon today, but either way, you can go there on your own after the test”. This always seems to make folks feel better.
But our biggest insight into this behavior came when we started questioning why Amazon was so powerful. Why do people have such an urge to visit Amazon before they go anywhere else? This wasn’t happening with other sites, only Amazon. What makes Amazon so special?
One of our first hunches was that people were so strongly leaning toward Amazon because they already had accounts there. It turns out that this isn’t the reason, though. Even though their credit card information is stored on Amazon’s servers, the few moments they’ll save not having to type it in again isn’t the reason why they want to go to there. It’s not the prices, either. People expect low prices at Amazon, but they don’t seem to think that Amazon has the lowest prices.
In fact, the testers don’t necessarily want to purchase at Amazon. They don’t say “I would rather shop for this on Amazon”. What they say is, “Can I go to Amazon first“. This subtle difference in wording turned out to be the clue to what drives them there.
The real cause of the Amazon Effect is research. Amazon has become the Consumer Reports of the Web. The primary reason people go to Amazon is to do product research. They trust what they find there and it heavily influences their purchasing decision. Some people we’ve tested claim that they go to Amazon and read the reviews there before every purchase they make. They want to be sure that what they’re buying has been positively reviewed.
Another interesting behavior we’ve seen is that some people only read the negative reviews, as if they’re looking for a deal-breaker review that just makes it impossible to purchase. We’ve heard comments like “I know what the positive reviews will say, but I don’t know what the negatives ones will say”.
The Amazon Effect is very strong, and we’re continually faced with it. However, it doesn’t seem to affect the usability of the site, as we find just as many problems on Amazon as on other sites (especially with third party sellers). But even with its own usability problems, Amazon has an effect unlike any other site we’ve tested.Tweet