We marvel every day at the potential that these conversational robots have from a marketing point of view.
What interests us most about chatbots is their ability to produce consumer insights for brands, especially (and logically) among the under 35s. This is due to new consumer behaviors and the very nature of chatbots.
Introduction to chatbots
Natural language processing
This is a field of machine learning that allows the human ability to communicate through natural language to be modelled and reproduced by machines. We are talking here about training models to give close representations (in the form of vectors) to words/sentences/documents with similar meanings and, on the contrary, distant representations to words/sentences/documents with opposite meanings.
A vector representation of almost all words in the English language exists and is available: it is called word2vec. This model for transforming words into vectors has been implemented and trained on billions of words and by looking at the context in which each word occurs – the other words around it. In this representation, we can see, for example, that the path from “man” to “woman” is exactly the same as the path from “king” to “queen”.
In our opinion, we should start by asking ourselves what the different types of chatbots are. Many bots are not based on natural language interpretation. And most of the time, they are not because they don’t need to be. The most common examples are chatbots developed on Messenger for e-commerce groups or for media. These are tree-like paths to guide the caller to the elements that will answer their query: the “bot” asks you a question with several answer buttons and depending on your choice, guides you to another question in the same form. These bots represent a new form of customer journey to meet their needs. On Slack, the bots are command interfaces: they only react to the appearance of keywords or sequences defined by rules. In this POC, the idea was to develop a chatbot with which one can exchange freely, using natural language. The interlocutor is therefore not supposed to know any rules before starting to interact with a bot. His statement alone should allow the bot to “understand” what answer to give or what question to return.
On the consumer side, what’s going on?
72.2% of European people own a smartphone, a rate that rises to 95% among 18–24-year-olds in 2017. In the United States, it reaches 77% of the population. As for the messaging applications to which chatbots are connected, their use is expected to exceed that of social networks in the coming years, with more than 2 billion users worldwide expected in 2018 and a global market penetration rate of 80% among smartphone users in 2019.
Alongside this growth, the emergence of chatbots on these messaging channels has led to some startling findings about user behavior. The WoeBot bot and the academic work that preceded its development demonstrated the tendency of users to reveal their innermost feelings and desires to a bot more easily than in the presence of a human, when they know they will not be subjected to any value judgment.
What about the chatbot?
A chatbot is a computer program that you speak/type to in order to trigger actions. One of the advantages of such a program is that it generates a large volume of data, hence the multiplication of analytics platforms for chatbots such as Botmeter, which we have developed at Botfuel for our bots.
In the context of a conversational commerce chatbot, whose goal is to allow users to configure a service or browse a product catalogue and buy, I like to compare it to a customer journey in a shop. Imagine being able to follow all the actions taken by your customer in the shop: the first product they see, the ones they grab, the ones they take out and put back on the display, their journey between the different shelves, etc. A bit like on an e-commerce site. Yes, but with a significant new dimension.
Instead of impersonal clicks and searches, your customers express themselves in natural language, in writing or by speaking, a fundamentally personal means of communication. This gives marketers unprecedented insight into the relationship between their brand and their customers. It is no longer just the keywords they used in their search engine that you have access to, but potentially all the words (and emojis!) they used to access your products, to qualify the products the chatbot offers them.
Finally, some human in analytics
In the formulation of insights and the understanding of consumer behavior, chatbots bring a new dimension absent from any other web analytics, the human dimension of your consumer, who is then no longer reduced to a series of clicks and conversion rates on a dashboard.
As chatbots gradually become part of consumers’ daily lives, they open up a valuable source of information for marketing teams to exploit.
Would you like to implement chatbots to your business? Contact us today.