GUEST COMMENT Recommendation engines are powering the next generation of ecommerce

Retailers already know that they need to provide a positive customer experience across both online and offline channels. However, as the online marketplace becomes increasingly saturated, retailers will need to differentiate themselves by implementing more personalised and innovative technology to attract customers and retain their loyalty.

The Infosys Rethinking Retail Survey found that 86% of online shoppers who have experienced personalisation technology have admitted that it has influenced their purchasing decisions. Additionally, 69% of online shoppers said that consistency of customer service across both physical and online stores was of great importance to them. With this in mind, how can retailers make use of valuable customer data and recommendation systems to provide a more seamless experience? And what are the practical steps retailers should take to implement recommendation systems?

At the same time, there may be a consumer backlash brewing. A recent survey by Pew Research shows substantial worries by consumers about the security of their information on-line. Over 90% of adults feel like they have lost control over the way that their personal information is used on-line. This means that while recommendation engines can help shoppers find what they want, care has to be taken in how these systems are implemented.

How a recommendation engine works

Most of us have experienced the power of personalisation technology. For example, you can find a former colleague through LinkedIn’s ‘people you may know’ feature, watch a movie Netflix suggested, or buy an item Amazon.com recommended under the ‘frequently bought together’ section.

In all of the above examples, recommendation engines help provide options that best meet customers’ specific needs. The recommendation systems that these companies have built incorporate algorithms and machine learning that use past data to make precise predictions about user preferences. Most industries, including financial services and media, have a portion of their business driven by a recommendation engine. For the retail industry, however, a recommendation engine is arguably most effective in the context of an online retail or ecommerce operation.

Retail customers benefit from a more tailored and personalised shopping experience, and this positive experience increases the likelihood that they’ll buy more products and stay loyal to the retailer in question. For the merchant, a recommendation engine increases upsell and cross-sell rates, reduces churn, and improves customer loyalty.

Massive amounts of data are collected during online transactions, including clickstream or mobile data, past transactions and behavioural data. By leveraging new big data technologies such as Hadoop, volumes and variety of data can be mined for patterns and outliers to predict what the next optimal product or service should be for that individual customer. Retailers can then offer products based on what other similar customers have bought—providing ‘next best offer’ opportunities.

How retailers can get started

The good news is that there’s a new trend in machine learning and particularly in recommendation engines: very simple approaches are proving to be very effective in real-world settings. Machine learning is moving from the research arena into the pragmatic world of business; thus, the technology is becoming more accessible to retailers and easy to incorporate within their businesses.

To get started with a recommendation engine, retailers should think about the following five steps.

1. Ingesting item meta-data: This helps identify items or products for recommendation. Most retailers already have item meta-data available in a search engine

2. Ingesting log files that contain user history behavior.

3. Analyzing the users’ behavior to create new meta-data that can be ingested back into the search engine to support recommendations.

4. Enabling users to interact with a search index that gets populated with item metadata and user behavior meta-data. This in turn generates new user history that is fed back into the system as part of a closed loop mechanism to help improve the recommendations that are delivered.

5. Improve the system by finding alternative behavioral data that produces better recommendations.

Note that this approach gives the benefits of personalisation and recommendations without exposing any private consumer information.

Next generation ecommerce

This year, online shoppers will spend an estimated £52.25 billion, up from £44.97 billion in 2014. For retailers, this translates into a huge opportunity to stake out a part of the sales growth, and they are relying on sophisticated methods—such as using a recommendation engine—to make use of the benefit from the massive volumes of market and customer preference data.

With a sophisticated recommendation engine in place, the typical customer experience can be highly tailored and optimised. For example, as you order items from Amazon, a section lower on the screen suggests other items that might be of interest, whether it be cookbooks, juggling toys, or collectible ceramics. The items suggested for you are based on items you’ve viewed or purchased previously. Even simple actions like clicking ‘to read more’ help shape your experience. Even Google Maps adjusts what you see depending on what you request; for example, if you search for a tech company in a map of Silicon Valley, you’ll see that company and other tech companies in the area. If you search in that same area for the location of a restaurant, other restaurants are now marked in the area.

We should expect the future of shopping to be a seamless interaction between machine and shopper—with technology guiding customers through their purchasing journey. This next generation of shopping will rely on the volume and quality of data the retailer has, and the infrastructure in place to leverage the data. The resulting personal interaction between retailer and consumer will advise the customer from product discovery to purchase, while building an ongoing relationship based on knowledge between the retailer and the consumer.

Carol McDonald works in education services at MapR Technologies