Graph databases, unique in their ability to model connections in data, are finding practical applications in a host of industries – but one of their biggest platforms is in the area of search and product recommendations for online retailers.
Retail giant Amazon first instigated the use of recommendation engines. Today every online store uses some kind of recommendation engine that utilises algorithms and data to search out the relevance of items given a certain context and makes recommendations to consumers. Graph software helps online retailers build a picture of their customers, what they are searching for and what they put in their baskets. If it is simple and convenient, the more shoppers will shop.
Digital technologies have put consumers in the driving seat. They expect a personalised service, tailored to their exacting needs, which includes individual recommendations based on their preferences, personal interests and shopping history. Consumers are more willing to shop in online stores that recognise them as an individual and know their name.
The stakes are rising as personalisation becomes more sophisticated, which makes the concept of a virtual personal shopper key to retaining and growing business against a highly competitive online retail market – where consumers are both fickle and time poor. Virtual personal shoppers are capable of quickly and accurately making recommendations, saving the consumer time working their way through the virtual shelves of an online store.
These intelligent, context sensitive and knowledgeable recommendations, coupled with big data, enables a new breed of customer interaction that enables online retailers and brands to connect with their customers on a more personal level than possible previously. These next-generation recommendations also help retailers create a real sense of need for the products they are offering to potential customers.
This hyper-personalised approach is largely powered by artificial intelligence (AI). This is deployed in conjunction with data-driven smart software, which significantly incorporates real-time capabilities. From the retailer’s point of view, the best part is that often customers won’t know if they are communicating with a human or a bot. The cog that is central to mining these ever-increasing data connections and sources and ensuring it works smoothly is native graph technology.
Talking to your customers
All indicators so far show that ‘conversational commerce’ works and that consumers like it. It saves them time and can seamlessly be integrated into their mobile lives. Take eBay’s AI powered ShopBot , currently only available in the US. This smart personal assistant has been designed to be there whenever the customer needs its services. The app runs on Facebook Messenger and lets shoppers interact with a chatbot using text message, speech or image. The beauty of it is that shoppers do not need to describe the item they are searching for, they can send a picture and the virtual shopping assistant will search out anything similar.
eBay is aware that with an increasing amount of items available for sale on its site, finding the right one can be hugely time consuming for consumers. ShopBot uses AI and recommendation technology to simplify the shopping experience, scanning over one billion plus products. ShopBot is part of eBay’s vision to make shopping on its site as easy as talking to a friend. The more customers use the ShopBot the smarter it will get at its job.
The key to ShopBot’s appeal is the conversation element. The same as for us human beings, speech allows the software to garner more contextual information than a typical search box would be capable of harvesting.
Gauging human intent and delivering highly responsive, accurate help is what makes the next generation of recommendation engines so savvy. Technically, this requires a combination of ML (Machine Learning), accurate predictive analytics, a distributed, real-time storage and processing engine, together with NLP (Natural Language Processing) to work out what a consumer is basically thinking.
Making these real-time virtual assistants work requires the power of a graph database. Graphs help to refine the search against inventory and make connections inside data sources based on shopper intent. This enables the system to build up a picture of the customer in real-time to make accurate recommendations.
The context is then stored, so that the likes of ShopBot, for example, can remember it for future interactions with the customer. It pulls together this information by a rapid traversal of the underlying graph database model. This enables the application to make specific product recommendations swiftly.
By comparison, the traditional relational database way of storing data is under the ‘store and retrieve’ mantra, which can only handle a small number of data links or data sources. In addition, SQL queries are chained by their complexity, which in turn means they are often not able to act in real time. This does not sit well with recommendations, which require real-time contextual information to be accessible quickly. The same also applies to Big Data technologies, which can handle data volume, but which fall down at managing data connections.
Digital customers are increasingly demanding. Online retailers looking to ensure their place in the game really need to start helping their customers by moving to the next generation of automated retail assistance. This means addressing how they can include real-time customer communication in their recommendation strategies through various messenger applications and bots.
Smart companies are finding new ways to interact with their customers that are relevant and personal. These next generation recommendations are a huge opportunity for online retailers and undoubtedly here to stay.
This hyper-personalised approach is set to become a powerful part of the customer journey and native graph software can help make it a reality for online shoppers everywhere.
Emil Eifrem, CEO of Neo4j
Image Credit: Zapp2Photo / Shutterstock