From data
To meaning

Or: how to learn to design conversational interfaces and to satisfy users (and customers).

Let’s make a fantasy effort and go back to the 1960s: the only people who are able to interact with a computer are professionals or computer experts. Communicating with a machine means having a set of specific programming skills and knowledge that are additional to the skills possessed by other human beings. It is the human being who must express himself in a way that is understandable to the machine.

In 1975, the advent of the Personal Computer overturns this paradigm. Anyone can now become a potential computer user, regardless of their technical skills. The computer becomes a tool that meets usability and immediacy criteria: that is, everyone must be able to use the machine based only on their innate and cultural intuitions.

A new discipline is born, the Human-Computer Interaction, which aims to model machines starting from the human cognitive model. Gradually,  it’s the machine that expresses itself similarly to how one human being communicates with another.


Voice Design? Voice App Design? Voice User Interface Design? VUI? CUI? Conversational Design? Conversational Interaction Design? Voice UX?

This Human-centered approach to technological design has become the starting point for the creation and development of new interfaces and experiences. Conversation-based user interfaces, in this context, represent the most natural way for humans to communicate, both with other people and with machines. In fact, a human being normally begins to express himself through natural language around the age of 2 … and then never stops!

The recent proliferation of natural language interfaces – whether textual or vocal assistants – demands for new effort in the field of User Experience design. even if only in terms of wording and definition, there is quite a deal of confusion! Luckily, CELI has been investigating requirements, good practices and key concepts for some time now within the discipline generally called Conversational Design.

In Conversational Design, users are the focus: conversational experiences must converge to the mental model that people possess on how a dialogue should unfold. Obviously, communication between humans or communicating with a computer has its differences. Users often have a wealth of prejudices or misconceptions about the behavior of a machine: they can expect too much, or too little, or attribute human prerogatives (such as lying) even to a computer.

A good designer creates a bridge to successfully transition the user from a  human communication to a human-machine dialogue.

Designing the bridge

Like all bridges, our conversational interface will have to join two sides: on one hand, the needs of a supplier – to innovate and give a new voice to its brand – and on the other hand, the needs of an end user: obtain information, actions fulfilment, to be entertained, …

Between the two shores, a sea of opportunities and potential rocks. For this reason, we begin by mapping the path of our bridge, taking care of the User-Interface interaction through a joint definition of the Customer Experience that our product will provide.

The first step is to understand the level of involvement that the user perceives when interacting with our reference brand: involvement can be high or low, according to a series of parameters. High involvement means longer decision times before making a choice (for example, to buy a product or not), high monetary stakes, a long-term interest in the purchase made; the user makes a cognitive effort to process the various steps in a rational way. Low involvement, on the other hand, means a momentary decision, which implies a low monetary cost of the product in question and it is linked to emotional or sentimental reasons; that is, the user relies on affective intelligence to choose which product to buy, and whether to even buy it or not.

The level of involvement affects the type of Customer Journey that the end user makes when interacting with our brand. For example, a product (or experience) with high involvement ecompass  a very strong interaction in the phase before the purchase, and a milder interaction in the post-purchasing phase. For a low-involvement product, there is an interaction even during the purchasing phase itself, but above all the attention is focused in the after-sales moment.

By differentiating the paths  we can quickly identify the various pain points that each interaction will present. These critical points actually constitute our opportunities, where a conversational experience can intervene to improve the Customer Journey.


Building the bridge

From theoretical analyses, we move on to practice. First of all, it is necessary to highlight which thoughts and emotions users feel when they are up to the pain points. And, consequently, how users formulate their thoughts and in what tone they do so: “I changed my mind about my trip, is it possible to ask for a refund?” and “I’m not satisfied with the service at all, I want my money back!” presuppose a similar intention on the part of the user: to obtain a refund. But the words and sentiment expressed in the two sentences are very different, and a good conversational assistant must be able to handle both cases.

Starting from the primary needs of users, the pillars of our bridge build ideal interactions between user and computer that lead to the accomplishment of those needs. Happy paths – satisfying conversations – are the first scripts of the conversational interface under construction. From these scripts, through tools known to the design disciplines, we obtain maps, flows and cards to fully represent the human-machine interaction.

However, not all interactions will be satisfactory in the first attempt: there will be errors, misunderstandings and unexpected events that will have to be managed. Therefore, remaining in our metaphor, we need to strengthen our bridge and equip it with safety barriers so that users do not derail from the conversation, but they may manage to reach their goals regardless of how they linguistically express their demands. We must not forget accessory elements that also serve fundamental purposes: a good conversational assistant knows how to help the user, but it is also savvy on politeness. It knows that when somebody says “Thank you”, the answer is ”You’re welcome!”.

We also believe the testing phase to be of paramount importance. Before going in production, the system is put into the hands of those who do not know its deep functioning logic and better represent the agnostic end user. The variety of natural language ​​is enormous and it is not easy to find all the ways to express the same concept at the first try.

The added value of CELI

Conversational Design is a unique discipline, since it combines paradigms of cognitive science and psychology with elements of linguistics, copywriting and human-machine interaction. However, the design phase is only the first step towards creating a product. CELI linguists and developers have the skills to guide the customer through all phases: from defining the Customer Experience that you want to create for  your users, to choosing the best touchpoints – with special attention to technical constraints – to developing for these touchpoints, as well as the integration of the product into the omnichannel.

From ideas on post-its to the development of complex technologies integrated with written and vocal systems, CELI has been helping companies innovate for over 20 years.

If you are interested in these themes, CELI organizes introductory workshops that allow companies to deepen the theme of Conversational Design and, in general, on the development of vocal interfaces. For more information fill out the form!

(Francesca Alloatti – Computational Linguist)

I have read and I agree with the privacy policy

Tags: , ,