And if the world of marketing studies had any interest in appropriating these new tools that are chatbots (or French version “conversational agents”) which, in the space of a few years, have taken an increasing place in the devices relational brands – consumers.
This is one of the ideas advocated by the institute Audirep studies , with the bias to experiment these devices for study purposes. Why ? For what types of use? What are the benefits and potential limitations? These are the questions answered by Laurence Hua, director of development of the institute.
Q1. For a few months Audirep has been looking into the possibilities of chatbots for marketing studies. What triggered your interest?
Laurence Hua(Audirep): This is part of our ongoing search for innovation, particularly on our collection tools. The key idea is that the more connected studies are to people’s habits and the tools they use in their daily lives, the better. This is an essential condition for collections of data to be both qualitative and representative of “real life” and “real people”. The use of chatbots being very strongly diffused these last years in the relation marks – consumers, we told ourselves that there was a real interest to see how the studies could appropriate this conversational mode. An interesting figure to note, from the AFRC (French Association of Customer Relationship):
Q 2. I imagine that this interest may also be related to the difficulty of reaching certain targets via more traditional tools like the phone?
In fact, this is a big part of it. It is the sense of history for marketing studies to move towards a form of multimodality with an increasingly wide range of tools for interacting with individuals. The phone for example is still very useful today in the BtoB universe or on some difficult targets.
The challenge is to propose the best collection tool according to the type of data to collect, targets and where they evolve naturally. One of these spaces is that of social networks, the use of which is completely trivialized, which poses to the institutes the question of how to find interesting information for brands. The chatbots offer from this point of view a promising opportunity and will go among other things to look for young and connected targets.
Q 3. In what way are chatbots particularly interesting to “plug” on social networks?
A little over a year ago, the biggest social network – facebook not to name it – opened the use of Messenger to brands, with the possibility for them to integrate chatbots, via an API. This initiative received a very positive echo, the figure that circulates being that more than 10,000 branded chatbots were set up on Messenger in just 4 months. Since 2011, 100,000 chatbots were built solely via the Facebook API according to Maddyness. So we thought it was also an opportunity for the studies.
Q 4. Where are you today in the study of these chatbots?
We are engaged in a logic of experimentation. We have shared the lessons of our first tests in the last Spring of Studies, and are today in discussion to go further with three of our customers. The idea is to see with them how it works, what kind of results can be obtained, with the aim of having first results in early 2018.
Q 5. What are the types of issues for which chatbots can provide interesting answers?
We see today mainly four. First, the opportunity for brands to collect the opinion of users on their Facebook page. Why and what do they react to? What types of information did they come for? What is the attractiveness of posts? … One of the other obvious applications is mobile advertising, again on facebook. Still according to the same principle, you can activate a chatbot as soon as a person has been exposed to a campaign, to measure its impact and the reactions it provokes …
Q 6. In both cases, the use is limited to the world of social networks, and even that of facebook (messenger) ….
This is true, with the advantage, however, to face voluminous individuals absolutely huge, the number of users now approaching 1.3 billion monthly active users. But beyond facebook indeed, it is quite possible to integrate chatbots on almost all websites, with interest probably very particular for e-commerce issues, for example to understand the reasons for abandoning a shopping cart, or for many other reasons. The decisive advantage of the chatbot device is that it allows an immediate interrogation, hot. We find this same advantage for tests related to events (an event organized by a brand or an institutional player, a festival … or just an experience of visiting a sales area). In this case,
But what we have just mentioned are only examples, the field of possibilities being undoubtedly extremely vast.
Q 7. Some advantages seem obvious, in particular the principle of a real-time collection, 24 hours a day and 7 days a week. But the very particular relationship mode that induces the chatbot is so easily compatible with a logic of study ?
It is clear that with the chatbot, we leave the dominant framework of studies and highly standardized questionnaires. But it seems to us to be an interesting line of work to take inspiration from what has changed in recent years in the relationship between brands and consumers, where we are much more conversational and playful. This has an important consequence, it is that we must take the tone of the brand, and therefore work more closely than we usually do with the communication services of our customers. The chatbot must look like the brand, communicate its “values”, for example by playing on humor or rather on its warm side.
Q 8. I imagine that one of the main constraints of these tools is that it imposes particularly short questionnaires … What other limits should be taken into account?
Chatbots lend themselves to short interrogations, the first tests are done on 10 to 20 questions but this threshold can evolve to a few minutes depending on the type of study. We will probably come there one day, but with chatbots, we are not yet fully in the field of artificial intelligence. We are dealing with software agents who can be given a certain personality to match the brand, and who can integrate aspects of scenario, but these remain fairly simple. We are therefore rather in closed questions. But it is nevertheless perfectly conceivable to integrate relevant open questions since they are triggered according to the context, in addition to closed questions on very large sample sizes.
Q 9. Can the developments of artificial intelligence considerably amplify the interest of chatbots within a few years?
Certainly. But we must nevertheless take into account the economic equations of our customers; with current chatbots, we are likely to be on budget levels quite close to those of a cawi, unless we add open questions that necessarily require additional resources. Since we are going to integrate artificial intelligence with more sophisticated semantic interpreters, that may be to increase the budget.
We are only at the beginning of the exploration of this new collection of playful and responsive, it is by experimenting and working with our customers that we will improve both the intelligence of our robots, the proximity to the targets, profitability and efficiency of the tool.