WHY WHAT WORKS FOR AMAZON, WON'T WORK FOR YOU

How and why a small retailer has the advantage in effective cross-sell recommendations.

Recently a client approached me, asking if we could implement the same product recommendations strategy that Amazon.com uses, for his retail site.

If you've shopped Amazon.com lately, then you've probably noticed their product recommendations. The typical "you may also like these items..." sort of thing that always makes me wonder what genius in the merchandising department decided that a toaster and a flashlight are ideal catalog cohorts.

Of course Amazon's recommendations have improved dramatically over the last few years - as they've exploded to dominate the online retail world. But don't assume that just because they're big and have plenty of money to spend, that they have the best of all worlds.

How Amazon's recommendations work is pretty simple. When you select an item to view (a "product of interest") their site simply queries what other shoppers in the past which viewed that same product also viewed. This is certainly an over simplification of process, but describes the core of it.

Why it works for Amazon

Amazon is able utilize this method for making product suggestions because of the incredible volume of both products and shoppers they experience. The massive amount of data collected allows for measurable variance in similar product relevance scores, filtering reasonable options to the top.

Why it won't work for you

Without a substantial volume of products and visitors (probably several thousand products, and several thousand shoppers daily), the likelihood of making logical recommendations is very slim.

Why you shouldn't care

What you may not realize is that this methodology is ultimately not incredibly effective. It's quite likely that Amazon uses this method simply because with the volume of products in their catalog, it's not financially feasible for them to use a less automated method.

The best thing you can do is understand how the current shopper defines their need, then recommend products based on that - not on what past shoppers liked. Your current shopper doesn't care about your past shoppers.

Your product recommendations can actually be far more effective than Amazon's, if you formulate a strong merchandising strategy and implement your cross-sell meticulously.

Stuck in the middle with you.

If you have thousands of products in your catalog, it's probably not feasible for you to manually set all of your cross-sell references. And if you don't have Amazon's volume of traffic, using a similar automated strategy is unlikely to be effective for you either. There is a middle ground. Suite Spot Commerce's Morphandize technology can make automated cross-sell recommendations that are the most effective in the industry!