This article was originally published by Jean-Denis Garo Head of Marketing Golem.ai
Conversation has become the new dogma for marketers and salespeople. Many companies have deployed a gimmicky chatbot, without measuring the image impact of a tool that does not meet users' expectations and generates dissatisfaction.
Customer relationship deserves better than a simple chatbot. Artificial intelligence offers, today, the possibility to automate the categorization of messages received on different channels (emails, sms, social networks, chatbots...). It' s time to use AI to provide actionable business insights.
The chatbot disappointments
The first steps of an automated conversation attempt date back to the 1960s, with Joseph Weizenbaum's "Eliza" project; a project then based on the use of the echolalia technique (tendency to systematically repeat all or part of the interlocutor's sentences as a response). The famous MIT computer scientist is also at the origin of the "Eliza effect", which designates the tendency to unconsciously assimilate the behavior of a computer to that of a human being. Yet, despite the progress made since then, chatbots are not as successful as expected and are sometimes even counterproductive, as demonstrated by the suspension of the Twitter account of the conversational agent Tay on March 23, 2016, after just one day of operation by Microsoft. In fact, the technology that supports chatbots does not deserve the bad reputation it has been carrying for the past few years; it is suffering from the fantasy of a conversational tool that would have answers to everything. However, due to a lack of time and financial means, chatbot applications often have a limited spectrum. They often lack depth, because the decision trees are not developed enough. Sometimes they are not even integrated with other customer relationship tools such as the contact center. Nothing is worse for the customer experience than wasting time in a conversation with a chatbot without getting an answer, only to come back last in the contact center queue. The chatbot must support and be integrated into the omnichannel strategy to serve a better user experience.
The challenge of customer knowledge and personalization
The multiplication of communication media today offers consumers an unprecedented choice in the media they will use for their interactions with the brand: from social media, email, sms, instant messaging, to the telephone, etc. The challenge, which is not new either, is to integrate and analyze all the data communicated through the different media used, whether structured or not*. This is the price of the customer experience, and each user must be offered the most appropriate way to communicate with the brand. This linguistic analysis of message content (often asynchronous or fast asynchronous) is a challenge for contact centers. A challenge that symbolic artificial intelligence** will meet. The idea is to enable multi-categorization operations: in short, AI can recognize several intentions in an email (or an instant message, a pdf, etc.) and classify them in the appropriate categories. This recognition of customer intentions is automated and compared with other data in the brand's possession (an order, a complaint, a consumer opinion, etc.). It will allow an enriched contextualization and propose an automation or a semi-automation of the answers of the agent towards his customer. The customer experience is thus enriched and personalized, the agent augmented, thanks to AI.
According to the Axys Consultants barometer, 83% of companies believe that the pandemic has reinforced the importance of customer service and 56% that it has accelerated the deployment of AI. The automation of data processing and analysis favors self-care approaches (offering customers the possibility to solve their questions by themselves), and above all improves customer knowledge and experience. This is the consecration of an actionable AI at the service of customer relations.
Comments