How Amazon Uses Big Data To Improve Its Online Shopping Experience
Amazon uses big data in much the same way that Google does to stay ahead of competitors. Google, and not coincidentally, also happens to be one of the largest providers of online retailing. At a time when e-commerce is growing at a reported 6% annually, Google’s visibility and influence in the industry make it the obvious choice for companies and brands looking to market and sell. Yet, is Google’s strategy simply too good to be true? If you think so, you are not alone.
Recently, we had an interesting chat with Michael Cieply, VP of Research at Sperry Rand, about how Amazon uses Big Data to drive its business forward. We were curious to learn how exactly Amazon decides what products to sell and at what price, when to charge for those products, and why it takes so long to deliver all of its packages. Mr. Cieply had some fascinating insights into how Amazon uses real time data analytics to respond to the changing consumer preferences, desires and whims. We talked about the recent purchase of Kabbage and its potential to revolutionize the shipping industry, Amazon’s relentless pursuit of lower prices, why it’s expensive to deliver products to consumers in major urban areas, and why it’s so hard to track exactly how many packages a customer has ordered from Amazon in the past.
We also discussed the recent acquisition of Kabbage byocus, a provider of event-triggered e-mail software. Mr. Cieply told us that Amazon is using Kabbage’s technology to help it analyze customer behaviour, including the “peak” moments of a customer’s buying cycle. Amazon uses Kabbage’s technology in other ways as well; it uses it internally for its “customer support” programs, its subscription services and the Kindle, and it also partners with agencies and research organisations to collect and use behavioural data. Mr. Cieply added that Amazon also uses data analytics systems from outside vendors, such as IBM and Deltek, in order to assist in the development of its products and services. Amazon’s use of external expertise carries with it certain advantages that we discussed in the report.
As we’ve previously noted, one of the ways in which Amazon applies the “Big Data” approach is to use it to pre-empt future browsing patterns and to better understand customer behaviour. Mr. Cieply explained that Amazon’s browsing history approach begins by collecting information on a user’s most popular and most visited web pages. From this information, Amazon’s custom-made spiders will build a “web site geolocation” database, which they use to pre-empt users who may want to avoid particular web sites by changing their browser settings to prevent visiting them. Amazon claims that this technique helps its customers “avoid web sites that may have negative content”. According to Mr. Cieply, this geolocation technique is one of the factors that has made Amazon the most successful online in recent times.
Amazon also collects information on users who perform a particular action. For example, if someone were to perform a large number of clicks over a certain period of time on a particular merchant’s website, Amazon would record the IP address of each of these visitors and their location. The resulting geographical information helps Amazon determine the most likely location in which the visitor originated, which in turn helps them make a more informed recommendation to a potential customer. By collecting IP addresses, Amazon is able to provide precise geocoding, which allows them to narrow down the location where a user might be browsing. While other SEO methods may focus on identifying geographical domains and building links between them, Amazon’s focus is on identifying IP addresses so that they can focus on the traffic that should be its focus and which is most likely to convert into customers or sales.
Recommendation systems have long been used in many forms on the Internet and by merchants themselves. Amazon has now extended this functionality to a much broader audience. Starting with its Kindle digital reading device and now including its web browser and Kindle store, Amazon has developed a way for its users to recommend content to other Kindle owners. To do this it requires data on the behavior of one-click shopping users – those who order through the Kindle store or through Amazon’s web browser but then click through to a third party site that offers the item the shopper is interested in without actually clicking through to the site itself. Amazon calls this ‘E-tailment’ and calls it one of the many ways it makes its Kindle more effective.
E-tailment isn’t the only application of Amazon’s new Big Data approach though. In addition to recommending what kind of content to present to customers, Amazon also makes use of its massive customer behaviour database to help identify what types of recommendations would be most effective. This is known as customer behaviour analytics and it covers a wide range of factors related to customer loyalty, online behaviour and product buying decisions. While Amazon doesn’t publish the full list of factors that it uses in its recommendation programme, it says that up to 70% of its suggestions come from its extensive customer behaviour database.
The biggest challenge for e-commerce retailers however is how best to take advantage of this big data. Many small companies simply don’t have the technical skills required to build up and maintain an e-commerce website. There are also concerns about the security of storing such large amounts of data, particularly financial data. However Amazon highlights that even this concern is becoming smaller due to the prevalence of internet security software like encryption and authentication suites that are available for such websites. By taking advantage of all these factors and by keeping the data as secure as possible, Amazon suggests that it is helping to make the online shopping experience safer and more reliable.