X5 Retail Group, a leading Russian food retailer that operates the Pyaterochka, Perekrestok and Karusel retail chains, has begun to implement a robotic telephone platform. The system, which uses speech technology to carry out various business functions, is a proprietary development of the X5 Big Data department. The project also uses Yandex SpeechKit technology, part of the Yandex.Cloud online services platform.
X5 voice assistants are already helping the company’s retail chains to handle incoming calls and make outgoing calls without human intervention. Voice assistants simulate dialogue with customers and help to resolve their issues. The bot has more than 300 different phrases in its arsenal, with new conversation scripts created within a single day. Most callers say they feel no difference between talking to a person and a bot.
When a client contacts the call centre, the interactive voice response system directs the caller towards resolving his or her issue. After the caller formulates a question, the bot uses complex machine learning algorithms and accumulated data to identify the subject of the request. The bot can then maintain a dialogue with the caller and learn his or her name, as well as collect the caller’s date of birth, phone number and customer loyalty card information and enter this information into the system.
The main aim of developing speech technologies is to enhance the functionality of existing call centres. In the Pyaterochka retail chain, robots already resolve up to 50% of all requests on loyalty issues each day, and the conversion to target action is more than 60%. At the same time, the costs to maintain a bot are 5-7 times lower than that of an operator.
The X5 robot is already successfully processing incoming scripts for registering loyalty cards, transferring points between cards, and checking a customer’s card status. A script for checking the card balance and blocking cards has already been implemented. The bot also works with outgoing calls by contacting customers and conducting Net Promoter Score (NPS) polls. Cold-call campaigns are created in the robot’s feedback module and are automatically sent to X5 through a configured integration system. As soon as the system receives a call request, the robot calls the customer, communicates with him or her and asks questions about the quality of products and services in order to get an opinion about an individual store. Depending on the question, the client’s ratings or responses to open-ended questions are then recorded in the system.
Sergey Dobryakov, big data development director at X5, commented: “The main focus of our team right now is to scale the use of voice robots in the Pyaterochka and Perekrestok retail chains and to create a convenient app with a visual interface. We plan to further develop the speech services platform to carry out functions that include order and delivery confirmation, screening interviews when hiring employees, carrying out surveys and research, attracting more customers to our mobile apps and loyalty programmes, and consulting callers on a wider range of issues. The technologies we have developed allow us to make standard calls even more functional and efficient.”
Yandex SpeechKit makes it possible to use Yandex speech technologies, speech recognition (Speech-to-Text) and speech synthesis (Text-to-Speech) for various business solutions. The service synthesizes text in real time, is able to take into account the lexical and stylistic features of oral speech, and supports several languages: Russian, English and Turkish.