When we compare the design work for websites and applications and that of conversational interfaces, we come across a long list of differences: for example, in the first the user visualizes all the options that the web or app offers through the different menus, the language is different, the experience is practically standard for all users and a long etcetera of factors that change between one field and another and that This article from Amazon counts very well.
However, the main dissimilarity is above the purely visual. The aspect in which the GUI and VUI design are most different is an almost philosophical question, related to which of the two parts holds power during the interaction . And here comes the novelty: in the voice interfaces, it is the user who carries the singing voice and the technology who, for the first time, folds to the needs and the way of interacting typical of humans.
The voice, the main means of human communication is more humanized if possible to better reach people and respond more efficiently to their demands.
To design voice interfaces capable of meeting the conversational expectations of users and offer personalized experiences that take conversational interfaces away from the robotic aura that many people still assign to it, it is very interesting to situational design approach (I recommend this article by Jesús Martín, in which he explains it and proposes to work it through Airtable).
I will try to summarize it in the following:
The situational design states that voice interfaces should not be limited to launching one response or another depending on the decisions that users are making, but must also depend on the context and the situation in which they are.
What does this translate to? For example, at the same start of the conversation: the greeting should not be the same the first time a user enters a skill as when it is a recurring user nor should the responses be the same throughout the different times of the day.
In this article I will try to address some aspects that, in my opinion, should guide the situational design. For this, I will talk about how it should be, taking into account this approach, a totally invented skill that recommends ebooks according to the tastes of users (and that also offers immediate download). Let’s go there.
Variables in the conversational experience
Several factors come into play in a skill that recommends books. Knowing them and adapting our assistant’s responses to them helps to achieve a more personalized and successful experience. Let’s review some of the most important:
The first time someone accesses our skill and without a menu that allows them to orient themselves through the navigation of our interface (as it happens in graphic interfaces) , you need a explanatory welcome message which allows you to know what you can do using our skill.
If, on the contrary, we detect that this is not the first time that person has spoken to our interface and we use the same message every interaction, not only is that we are giving too much information and our user does not need so much degree of explanation, but it can give him a negative feeling, of automation.
Therefore, just like that happens between two people when they are gaining trust, we must adapt the greetings . I propose the following start messages adapted to various situations:
First interaction : Hello Sara, delighted. I’m Miguel and I’m here to recommend your next reading. I can help you choose history, romance, adventure, suspense, travel, or specialty books. What do you prefer?
Second interaction: Glad to talk to you again, Sara. How quickly these two months have passed. I hope you are great. Do you have any kind of book in mind or do you need some guidance?
Third interaction: Mmm, let’s see. The last two times you have chosen detective stories. Do you want to continue along that line or do you want to change?
In the same way, the greeting should be different if the user has just left the skill without completing the process, in which case resume the conversation Where you stayed would be the best experience. Also, if our skill has incorporated new functionalities since the last time you used it, our interlocutor will value this small update and will feel well taken care of by our bot.
The importance of timing
Issues such as time or day of The week in which the interaction occurs also influences the design of a good conversational experience. This does not only translate into greeting the user with a good morning or good night depending on the time, but it goes there.
For example: maybe not It would be a bad idea if there were a change in the tone of the conversation using more energy in the morning and less momentum and loquacity as we get closer to bedtime, at which point our user will probably be tired.
But now let’s apply this approach to our example. This is how our messages could be transformed according to the time variable:
Marching some summer recommendations, perfect for those deserved holidays! Hmm… I think you might like it a lot Terra Alta the latest from Javier Cercas (variable: July month).
And why not use the user’s location to offer a better experience? Let’s see an example:
I just realized that you are in Valladolid, so my recommendation cannot be other than Las ratas by Miguel Delibes. I am sure you will love it! (Variable: location).
Or, for example, get the atmospheric data from that location to offer recommendations that combine well with a sunny or rainy day.
These are just some basic questions that we could take into account in a situational design, which always puts the user’s context at the center of the experience.
In summary, the situational design allows us to adapt the conversational experience to the context in which the user is, leaving him with the feeling that he has maintained a personalized and effective interaction that he will easily want to repeat again.