Using Web Analytics to Cross-Sell

Amazon.com delivers an amazing level of personalization based on past customer purchases and behaviors. In my case, it even “remembers” all those one-off purchases from years ago and keeps showing me books on topics such as brewing beer. It’s found a great deal of benefit and success in recommending books, DVDs, CDs, and many other products based on what it knows about each customer and about people with similar interests. Its delivery of recommendations and content is finely tuned to each customer.

The other day I was looking to buy a few gifts and found myself looking at CDs on Amazon. I’d been talking to a friend who wanted the new Coldplay CD “X&Y,” so I pulled it up on Amazon and was preparing to buy it.

I started to look at other offers on the page. Amazon was willing to give me an additional $2 off if I bundled the Coldplay CD with another CD. The retailer isn’t alone with that type of offer; a number of other sites that use this approach. When I scrolled down the page, I found the “Customers who bought this title also bought” link. Again, another common method to cross-sell products and drive up average order size.

Below that was a new item I hadn’t noticed before on Amazon or any other site: “What similar items do customers ultimately buy after viewing this item?” The following four bullets appeared:

  • 42% buy this item (X&Y ˜ Coldplay)
  • 11% buy In Your Honor ˜ Foo Fighters
  • 8% buy Out of Exile ˜ Audioslave
  • 2% buy Strange and Beautiful ˜ Aqualung

    Wait a minute! Amazon just told me its conversion rate for this product page with this product! It convert 42 percent of all visits to the product page for the new Coldplay album. I found it interesting the product it offered $2 off on wasn’t on the list.

    Amazon has always watched, learned, and tuned recommendations based on performance, but this is the first time I’ve noticed it exposing the analytics data on actual buys. I’m interested to see if this has helped the retailer increase cross-sells or average order size. Unfortunately, email to Amazon went unanswered, but knowing Amazon it’s tested this and found for its audience, it does have a positive effect on orders.

    Interestingly, the same information isn’t in the books section and within the electronics section, it appears for some products but not others. I started to question how it might help customers. Does showing the view and purchase percentages help customers make a decision?

    I ventured into the Electronics section of the site, pulled up a product, and scrolled down to see how many people bought it. For this particular product, only 3 percent of the people who viewed it bought it. Additionally, people bought some more expensive products after looking at this specific product. The numbers were staggering:

    • 54% buy Syntax Olevia LT37HVE 37″ Widescreen HD-Ready LCD TV by Syntax Groups Corporation $1,499.79

    • 32% buy Samsung LN-R238W 23″ Widescreen HDTV-Ready Flat-Panel LCD TV by Samsung $849.94
    • 4% buy Syntax Olevia LT32HV 32″ Widescreen HD-Ready LCD TV by Syntax Groups Corporation $1,099.99
    • 3% buy this item (Syntax Olevia LT26 HVE 26″ HD-Ready Flat Panel LCD TV by Syntax Corporation) $724.99
    • 2% buy Sharp LC-26GA5U 26″ AQUOS Widescreen Flat-Panel LCD TV by Sharp

    I started to see how this could really be helpful to shoppers. First I though, “Wow, what’s the matter with the product I pulled up? Why are so few people buying this one after looking at it?” Then I thought, “What’s so great about the other two products that 86 percent of the people who view this particular product page go on to buy one of two other flat panel TVs?”

    It’s an interesting use of analytics. I’m not saying everything Amazon does is perfect, nor what it does on its site with customers (or a customer segment) will work on other sites. But it’s interesting to consider various analytics data uses to improve site performance. Typically, we think of analytics as a way to evaluate performance and identify opportunities. In this case, Amazon uses it in a completely different way.

    If you’ve seen other examples of analytics data in action; data being exposed to site visitors; or data being used in an unusual fashion, drop me a note. Maybe I’ll write a follow-up with a few other examples.

    Related reading

    tencent_emily-ma_featured-image
    kenneth_ning_emarsys_featured-image
    bounce-370x229
    site search hp
  • <