The popularity of artificial intelligence technologies in ecommerce is not surprising: the number of buyers is growing, their demands and demands are complicated, and the level of satisfaction is rising very slowly. Modernization and automation of most ecommerce processes became a necessity.
During 2017, Ocado, the world’s largest online supermarket, has planned a $ 232 million investment in artificial intelligence and automation. Amazon intends to invest $ 5 billion in technical development in India. Alibaba Corporation plans to spend $ 15 billion in research and development of artificial intelligence within the next three years. Forrester Research estimates that investments in artificial intelligence will grow by 300% in 2017, and companies associated with its development will earn $ 1.2 trillion a year by 2020.
In this article, we will consider which AI solutions are used by leading ecommerce companies and how they affect the workflow.
Back in 2012, the retailer bought the company Kiva Systems, which immediately turned into Amazon Robotics. The subsidiary company develops robots, working on machine learning, object recognition and computer vision. Amazon applies the received solutions in the work of its warehouses.
Recommendations of the goods
Amazon users are offered blocks with recommended goods on all pages of the site right up to the complete purchase. Despite the fact that the system of recommendations for Amazon products is imperfect (according to the University of Toronto, the accuracy of artificial intelligence assumptions is only 5%), in the third quarter of 2017, Amazon sales grew by 34% to $ 43.7 billion
As the company receives more and more information about its customers, the accuracy of the forecasts can lead to Amazon delivering goods to people before they order them.
Fighting fake reviews and reviews
In 2015, the management of Amazon sued the companies that paid for placing positive reviews on the site of the retailer. To combat the spread of fake reviews, Amazon has launched a special algorithm for machine learning.
Personal style recommendations
In April 2017, Amazon announced the project Echo Look, a personal assistant based on machine learning. Users can ask Alexa, Amazon’s voice assistant, to take a picture or shoot a video of their outfit and post it to social networks, or use Style Check data to compare their outfit with others.
The Wall Street Journal conducted a small experiment, inviting stylists and comparing their opinion and Alexa. To my surprise, it practically coincided.
Amazon intends to continue investing in artificial intelligence.
Machine learning allows our algorithms to anticipate user requests, improve product search, refine personal recommendations, make sales points more convenient, identify scammers, make correct translations and much more – Jeff Bezos, CEO of the company.
We use artificial intelligence to solve a number of problems: speech recognition, understanding of natural language, answers to questions, constructing dialogues, recommendations of products, – Rajiv Rastogi, head of AI-direction.
Amazon Web Services, the company’s cloud platform, implemented this in the form of the following projects:
- Amazon Lex is a chat-bot, which allows creating interactive interfaces,
- Amazon Polly is a cloud-based service for converting text to speech,
- Amazon Recognition – image analysis technology.
Also together with Microsoft, the company is working on the Gluon project, an interface for developers working on learning algorithms and models of neural networks.
China’s leading retailer invests in AI technology to improve its own service.
Recommendations of the goods
The company has its own technology of recommending goods – E-commerce Brain. Predictive models are based on data received in real time (user-viewed content, purchases made), as well as data obtained from the entire Alibaba ecosystem, for example, AliPay, AutoNavi, Youku and UCWeb.
Artificial intelligence also helps sellers located on the Alibaba platform to create virtual storefronts with goods that will be of interest to the buyer. Technology, among other things, takes into account the history of the user’s purchases, the source data about him and his location. According to Alibaba, thanks to this technology, in 2016, at an annual sale in November, the conversion increased by 20%.
Another AI-development company – Dian Xiaomi, text chat-bot, which helps sellers create and manage virtual shopfronts, as well as respond to users’ requests and solve their problems.
Smart Supply Chain
The Ali Smart Supply Chain platform uses artificial intelligence to help online and offline sellers anticipate changing customer demands, choose the right pricing policy, offer users the right products and ways to deliver them.
70% of deliveries in China (42 million parcels daily) are provided by Cainiao technology. She selects the fastest delivery routes, which reduces transport usage by 10%, and the delivery route by 30%.
Artificial intelligence helps a company process user data, identifying key problems and improving channels for managing client requests. Now AI processes 100% of user requests regarding the status of orders and successfully solves more than half of them.
In October 2017, the company’s technical director said that Alibaba in the next three years will invest $ 15 billion in research and development of artificial intelligence. The company will build seven research laboratories in which they will work on AI, machine learning, and reproduction of the natural language.
The company invests in research on predictive models in order to reduce the time spent by users in finding suitable products. To continue to compete with Amazon, in 2016 the company acquired start-ups SalesPredict and Expertmaker, engaged in machine training.
The possibilities of predictive analysis and machine learning will help our customers better understand the difference between different goods, and sellers will be able to determine the probability of selling any goods at a certain price, – Amit Menipaz, vice president of eBay.
eBay also invests in machine learning and big data within the eBay Research project and sponsors research and conferences related to them. In 2017, the company introduced the Image Search and Find It On eBay applications, which from a user’s photo select products sold on the site. To develop applications, deep learning and computer vision technologies were used.
Another development of the company – eBay ShopBot works in Facebook Messanger and serves as a personal assistant, to which the user can send a search request in the form of text, voice message or picture.
The largest ecommerce site in Japan uses artificial intelligence to better anticipate user requests. According to the company, data on 200 million goods allow you to accurately predict the sale of certain things, and real-time information is used to segment customers.
One of the company’s AI developments is the Rakuten Fits Me fashion application, which uses image recognition technology.
Of the 70 services that we have, approximately 30 will be equipped with AI-chat bots by the end of 2017. Our goals are to significantly improve user satisfaction, which will immediately affect sales, – Sunil Gopinat, CEO of Rakuten India.
The leadership of the largest e-commerce site in India is sure: artificial intelligence will help to more accurately respond to customers’ requests.
A shopping application, which is called Project Mira within the company, is used to process 50% of users’ searches, and it is in it that users add 12% more additional items to the shopping cart.
Work on artificial intelligence, according to the company, significantly improves customer service. Over the past year, the number of users ready to recommend the company has grown by 14%, and the number of dissatisfied customers has decreased by 25%. This result, according to the leadership of Flipkart, was due to chatbots.
One of the UK’s largest online retailers is developing speech recognition systems to influence the behavior of customers. The company also develops image recognition technologies, offering users a visual search that allows an application to match an image uploaded by a user to products on the site.
The company also develops a mobile virtual assistant called AVA, which helps users to search for the right products using customer preferences data.
Artificial intelligence for retail is no longer just a tribute to fashion, but an effective means of communicating with customers, improving interaction with them, a way to solve problems of product selection and delivery.
Despite the fact that only the largest e-commerce projects can allow the development of such solutions, in the future, there will certainly be effective solutions that will be available to small online shops.