Conversational AI in retail: robotization, personalization and voice search

McKinsey analysts found that 75% of consumers who changed their behavior in the last year will not return to pre-crisis foundations after the restrictions are lifted. This means that e-commerce will continue to grow strongly: according to Accenture, it was 27% in four months of 2020 against 16% for the whole of 2019.

Additionally, digitalization of retail is stimulated by two factors. First, the general requirements for customer service have noticeably increased, everyone sees good examples of large Internet companies. Second, there is a natural change of generations - purchases are made mainly by millennials and Gen Z who prefer to interact through digital channels. Retailers and experts from Just AI, a developer of conversational artificial intelligence technologies are thinking what is the future of retail.

Integration of offline and online

The pandemic has erased the clear line between online and offline. During the lockdown, many consumers tried the buy-online-get-in-store model, and half of McKinsey respondents liked it.

“Traditional retail is more of a story about a customer's offline experience, which often remains impersonal and does not in any way affect future interactions. And artificial intelligence must constantly learn from the analysis of the received data. We are solving this contradiction for ourselves by developing the OneRetail concept: we create a single digital space and enable customers to seamlessly move from online to physical stores and back and enjoy the benefits of each channel, and companies - to digitize face-to-face interaction with customers through mobile technologies in retail ”, - says Konstantin Rudenok, development manager for the technological direction of M.Video.


Retail automation will continue and will affect the most expensive service sectors. On the incoming line, text and telephone robots will process typical customer requests: delivery, return, order changes, bonuses. In outgoing communications, telephone bots will conduct NPS polls and inform about promotions.

So, the chatbot of Tikkurila the manufacturer of paints and varnishes works on the website, on Facebook, VKontakte and provides the company with automation of answers to 80% of consumer questions about paints and their use. These are 25 thousand completely successful dialogues, which he closes from beginning to end without involving an operator. The bot reacts to text and buttons and helps you choose a product, shade, tells you where to buy a particular product, helps to leave a review and download certificates. In other cases, the bot classifies the question and transfers it to the operator, that is, it participates in the first line of support.

In May 2021, Tikkurila launched another assistant for hotline consultants in Russia - a voice bot. Over the past year, the company noted a sharp increase in the number of calls to the hotline, and one of the most popular questions was “Where to buy paint?”. Hence the idea to add an electronic assistant to answer this question was born. Now technicians can devote more time to consulting, where their knowledge and expertise in materials technology is more valuable and in demand.

And X5 Retail Group is introducing a robotic telephony platform into the Pyaterochka, Perekrestok and Karusel retail chains. Voice assistant processes incoming calls to call centers and answers customer questions. The bot helps to register loyalty cards, block them and transfer points between them, check the balance. On the outgoing line, the bot is involved in NPS polls. The company emphasizes that speech AI technologies develop the functionality of existing call centers. So, every day at Pyaterochka, up to 50% of all requests on loyalty issues are solved by a robot, and the conversion to target action is more than 60%. At the same time, the costs for a bot are 5-7 times less than for operators.


Chatbots perform the duties of online sales assistants. They can work on the website, in a mobile application, messengers and social networks. Omnichannel unites all sites into one system and makes communication between the buyer and the brand seamless. The consumer can start placing an order on the website, and continue on the VKontakte social network exactly where he left off.


According to Microsoft, 63% of online shoppers are more open to the exchange of personal information if, as a result, brands and retailers can better anticipate their needs. AI and machine learning allow us to study customer preferences and make personalized offers to them, including using robotic telephony and chat bots.

“Personalization cases mean a lot in retail. They provide a wow effect that retains customers, ” - comments Svetlana Zakharova, Head of Customer and Partner Relations at Just AI.

Integration of chat bots with messengers and social networks

The popularity of bots is growing when they are integrated with instant messengers - a familiar environment for customers. Thus, the monthly audience of active WhatsApp users is 2 billion users. On the other hand, under the influence of the lockdown, customers began to spend more time on social networks, which was confirmed by more than 40% of Statista surveyed. The data points to an increase in the volume of purchases through the social media channel in the future. According to ResearchAndMarkets, the market will add more than 30% per year and will grow to $ 604.5 billion by 2027.

Voice search

Voice search brings consumer communication with retailers closer to natural form. For example, in the United States, more than 45% of users would like to be able to interact with mobile applications by voice. 35% of US SMEs reported in a survey last year that they had increased their investment in voice marketing. Among the companies developing voice services, 29% offer the opportunity to make purchases by voice, 34% - get general information about products, 28% - check the status of rewards, 32% provide a search function. Voice search will be used in mobile applications, as well as on smart speakers, screens, TV-boxes.

“At Tikkurila, we believe that voice interfaces will eventually become the standard for retail service, even in traditional markets such as paint and varnish. The question is that the costs of implementing and developing voice services are commensurate with the practical value for our customers,” says Egor Pajusov, Transformation Project Manager, Vostok, Tikkurila.

Voice shopping, video shopping

The habit of voice control has been largely shaped by smart speakers and screens. At the end of last year, there were 400 million such devices in the world, in Russia - more than 1 million. 62% of owners of smart speakers in the United States have ever made purchases using their voice. Voice commerce sales expected to reach $ 40 billion by 2022.

The positive aspect of voice shopping is that it leaves their hands free and allows the shopper to do other things in parallel. Among the product categories in voice shopping, everyday household items are in the lead - 25%, entertainment - 21%, and clothing - 21%.

If you need to look at the item before buying, it is much more convenient to use smart displays, which give more expressiveness and video shopping function. Thus, Ozon and Lamoda have integrated their products into the service available on the SberBox smart TV box. While watching a movie, the user can pause and see the product card and its main characteristics. The QR code will redirect it to the site directory for further purchase. At the stage of launching the service, more than 1 million products from the Lamoda catalog were technically available, mainly clothing. Ozon has integrated into the service items from such categories as toys, books, clothing, home and garden, tourism, fishing and hunting, electronics, etc.

The complete catalog of the Samokat online retailer is available in Sber's smart devices - SberPortal and SberBox. As Arkady Chugunov said at Digital Brand Day 2021, Samokat Product Lead, on average, video shopping users increase by 30% per week, and the number of orders is growing by 15%. Conversion to purchase per week is 18% for SberPortal and 15% for SberBox. At the same time, more than 50% of buyers through Sber's smart devices have never used Samokat delivery before, and the average bill, the number of positions and top categories roughly coincide with other channels.

Three Tips for Implementing Conversational AI in Retail

1) Check the performance of CRM and other internal systems At the start.

“We were faced with the need to go to 2-5 systems to get data and make changes to the order. A retailer must be ready to have all the necessary systems, they keep fresh information, have up-to-date access and that the systems respond to external challenges as quickly as possible,” comments Anna Zubkova, CEO of Digital Intelligence.

2) Select a team that will develop the bot.

The integration of bots entails constant work on adding seasonal assortment items to the catalog, finalizing the script and promoting special offers. “The Tikkurila hotline team regularly analyzes the history of dialogues and trains the bot to answer new questions, expands its competencies, adds knowledge on new products,” comments Egor Pajusov, Transformation Project Manager, Vostok, Tikkurila.

3) Test on a small case how the technology works.

“Don't think that only corporations with a bunch of analysts and developers can afford chatbots. I would advise you to start with something small, for example, to disassemble a separate scenario in the work of a call center and try to automate it,” comments Alexandra Kotkova, customer service manager at Philip Morris International.

Link to source