By Tugdual Grall, Techincal Evagelist, MapR Technologies
The UK’s online retail market has grown drastically in the past year, with Euromonitor valuing its growth at 17 percent in 2014. This has largely been driven by the increase in high street brands harnessing the advantages of combining ‘bricks and clicks’.
But as the online marketplace becomes increasingly saturated, retailers must work to find ways which not only provide a positive customer experience across both online and offline channels, but also differentiate themselves from their competitors.
Innovative personalisation technologies provide the perfect opportunity for brands hoping to attract and retain customers. In fact, over 85 percent of online shoppers who previously experienced personalisation technology admitted it has impacted their purchasing decisions, according to the Infosys Rethinking Retail Survey.
But it’s not just about getting it right on one channel or the other. 69 percent of online shoppers considered a consistent level of customer service across both physical and online stores of great importance. So retailers must look at their customer data and recommendation systems to see how they can provide a more seamless experience to their customers.
The nuts and bolts of a recommendation engine
From LinkedIn’s “People You May Know” feature, Spotify’s ‘”Discover Playlist” to Amazon’s "Frequently Bought Together" section, personalisation technology is everywhere and helps provide options that best meet a customer’s specific needs.
Recommendation engines incorporate algorithms and machine learning that look at a users’ past data to make predictions about their personal preferences. And with the significant information that websites garner, it is undoubtedly ecommerce operations that can most effectively deploy a recommendation engine in the retail industry.
Enormous amounts of data are collected in every online transaction, from past purchases to behavioural information, clickstream to mobile data. Leveraging new big data technologies, such as Apache Hadoop, helps companies drill down into the volumes and variety of data for patterns and outliers to predict what “next best offer” opportunities should be suggested to a certain individual.
Tailored and personalised shopping creates a more hassle-free, positive experience and ultimately increases the likelihood that a customer will buy more products and stay loyal to the retailer. This helps merchant’s increases upsell and cross-sell rates, reduces churn, and improves customer loyalty.
Machine learning – and recommendation engines in particular – have moved from the research lab to being deployed effectively across a range of real-world business settings. As the technology becomes more accessible, retailers are incorporating it into their business development plans to drive sales.
Any retailer considering a recommendation engine, should consider the following five steps:
- Ingesting item meta-data helps to identify further products or services for recommendation - for most retailers, this tends to already be available in search engines.
- Ingesting log files that contain user history behaviour help gain a greater understanding of the individual based on past preferences.
- Analysing the user’s behavior to create new meta-data that can be ingested back into the search engine helps support future recommendations.
- Enabling users to interact with a search index that gets populated with item metadata and user behaviour meta-data helps generate new user history - this is fed back into the system as part of a closed loop mechanism which then improves the recommendations that are delivered.
- Improving the system by finding alternative behavioural data ultimately produces better recommendations.
With online shoppers estimated to spend a massive £52.25 billion in 2015, up from £44.97 billion in 2014, recommendation engines will prove an important tool for businesses who want to increase or retain their slice of the sales growth.
When effectively deployed, a sophisticated recommendation engine can tailor and optimise a typical customer’s online experience to drive sales. For example, when looking at clothing on ASOS, you’ll also be suggested other items “you may like” based on items you’ve viewed or purchased previously, but also items that users "similar to you" have viewed or purchased.
In the future, retail will be a seamless interaction between shopper and machine as the technology guides a customer through the purchasing journey. But the quality of these recommendations will vary significantly between retailers, with accuracy dependent not only on the volume and quality of data the retailer holds, but also the infrastructure it uses to analyse it.
As the interaction between retailer, consumer and machine becomes ever-more important and intertwined, those retailers that invest in the right mechanics and keep fuelling their engines will finish first. Driving purchases through relevant and useful customer recommendations will help not only secure those all-important sales, but will develop a personal relationship between a customer and the brand.