Amazon’s Emetrics Don’t Suck

AmazonOK, first off this data is at least three years old, but I still think it’s fascinating.

I ran across it at Web Pages That Suck (I looked for a permalink, but couldn’t find one.  Look for "When Simple Design Doesn’t Work").  It’s a PDF of an Amazon.com presentation from 2004 that discusses, among other things, Amazon’s A/B testing.

There’s some great examples starting on page 14 showing differences in their top navigation tabs, a scaled tab that never was, and, the most interesting part for me, a look at simpler pages vs. more crowded pages.

Now, if you believe the current Web 2.0 trends, which I like from a design standpoint, you’d think more "white space" and less crowding would be more attractive and, hence, more profitable.

Well, it turns out that for a retailer like Amazon with lots to browse and recommend, more is better.  Cart adds were down and abandonment was up.

Amazon is well known for leaving nothing to chance, and letting the numbers speak for themselves, so, looking back at my own redesign, I’m really glad I looked closely at how they organized their home page.

I’d love to do some intensive A/B testing when time and budget allows.  Not only would it yield some interesting results, I think it’d just plain be a lot of fun.

Anyway, check it out.  It’s pretty fascinating.

Technorati Tags: , ,

2 thoughts on “Amazon’s Emetrics Don’t Suck”

  1. One of the times when I was young and thought maybe I was in the wrong field I went to work for a variety store as an Asst. Mgr.

    The previous manager had been fired and his replacement hadn't arrived yet. On my own with my advertising training I decided to revamp the displays. When the new manager arrived he informed me that we weren't Macy's. The windows were cluttered back up.

    He looked around to see what wasn't moving heaped a dozen or more items on a table all mixed together, stuck a special sign on it but didn't change any prices. People gathered around that table digging for treasure. Most of those slow items were gone in 24 hours. different strokes for different buyers.

Comments are closed.