August 2, 2008

Selling With Personalization: Why Amazon.com Succeeds and You Can Too

One of the most powerful features of Amazon.com's service is the recommendations it gives to customers who are looking for books and other products they purchase. Having all sorts of options such as being able to see great photos of products and knowing that shipping is free is nice, but being able to know what other purchased can do the most in helping you decide what product is best for you.

Birds of a feather do indeed flock together.

Although Amazon.com and other retailers have the money to fine tune their personalization recommendations systems these types of services are not out of range. Companies such as Rich Relevance. How does it work? According to Rich Relevance:

• First, they build relationship maps of product inventory and attributes
• Then, they capture consumer intent and combine it with the “wisdom of the crowds.
• On page load, they run multiple recommendation types in real time to recommend the perfect products.
• Finally, they measure the performance of the recommendations, creating a real time-feedback loop that constantly optimizes them on a per placement and per page basis.

I was curious how this system works and Rich Relevance answered some more in-depth questions:

How does one implement a system like this - in summary. If I have a Yahoo Store? eBay Store? Does it matter the shopping cart or eCommerce system?

richrelevance's recommendations engine works in any ecommerce system with which the storefront manager can place a script tag in the code and has access to setting simple variables in javascript like price, product name, category id, and session id.

How long does it take to implement

richrelevance's standard implementation approach--Javascript snippets on each page--is a very straightforward and quick process. Depending on the size of the retailer, integration can take anywhere from one day (for a smaller vendor) to two months for a top 10 eCommerce player (with thousands of SKU's).

What are the costs

We have a performance-based pricing model based on the sales lift we drive for our customers. We have established a no-risk situation for our partners where there is no cost or disruption to their current online activity during integration or execution in regards to richrelevance software.

Our standard rack-rate (MSRP) is 10% of the sales lift we drive. Pricing discounts may vary significantly based on several factors, but primarily depend on a merchant's annual online sales.

What would be the difference in what Amazon.com has and this service

Unlike first-generation personalization technology such as "collaborative filtering," richrelevance's Ensemble Learning brings an entirely new approach to market—a technology that adapts quickly to each different type of shopping behavior. Ensemble Learning determines the recommendation type (choosing between 15+ unique types) and which products to display by running thousands of multi-variate experiments, incorporating the feedback from each customer to determine what's working and what's not. This feedback loop allows the system to self optimize in real time, thus keeping pace with trends unique to the site, distinct page types, and each individual customer. Richrelevance's Ensemble Learning ensures that the recommendations ultimately chosen by the system are the product or service at the right time in the consumer's buying experience for the right outcome of conversion and lift.

Are there any studies which show the benefits of offering product recommendations vs not offering it

Forrester Research recommends e-retailers "add personalization to the top of the priority list" to help up-sell and cross-sell more products. Although "personalization" broadly defines a variety of eTail tools, it is most commonly associated with the use of recommendations: "People who purchased this item also purchased…" as seen on Amazon—the leading master of this strategy.

How important is personalization to boosting sales, compared with showing a logo for a secure site (like ScanAlert), or free shipping, or live chat for customer service

Just ask our clients! richrelevance recommendations have a powerful impact on incremental sales--showing double-digit sales growth.

"richrelevance proved to us, in a split test, that their methodology for creating personalized recommendations adds sales dollars. When they moved from "exploration" mode, in which they test different strategies on different pages, to "optimization" mode, it was like turning on a light switch. Sales per visitor jumped up by 10 - 15% on their side of the test…I've seen many sophisticated-sounding, slick-looking e-commerce products fail to produce results in the field. Rich Relevance stands out because they not only use advanced marketing and modeling techniques, but they produce real world results." - Larry Kavanagh, CEO, DMinSite (referring to richrelevance performance on DMinSite's retail client sites)

tags: crm , ecommerce , personalization , sales | 0 comments