In the not too distant past, each “conversion” depended very much on how the conversation with the salesperson went… What did the salesperson know about the product? Did the salesperson help the client find what he or she was looking for? Did he or she build trust? Did he or she listen to the client? Did he or she know the client’s name? Did he or she remember the client’s preferences? The Conversation Economy is emerging at a rapid pace, giving way to a more personal and personalized shopping experience where clients are treated as human beings.
However, this will only be possible if we are able to design conversional interfaces correctly. This means analyzing users’ needs, their expectations and the context where this voice interaction takes place. We are facing a transformation as important as the introduction of the graphical user interface in the 90s and as UX designers we have to tackle a series of challenges when working in conversational projects:
How can you design a more natural conversation?
One of the greatest challenges is to design a more natural and human conversation. We know that people interact with conversational interfaces in the same way they do with people, despite knowing they are not human. This is why it is especially important to achieve a conversational interface that is as natural as possible, be it an interface based on text (chatbots) or based on voice (voice assistants, cognitive call centers and conversational robots).
There are 4 main aspects which can help humanize and improve a conversation:
- Cooperation. This aspect is based on Paul Grice’s Cooperative Principle which states that when we communicate we assume that those we are talking with will cooperate in order to lead the conversation where we want it to go. This cooperation is based on conversational maxims which depend on the amount of information you provide (don’t communicate too much or too little), its quality (truthfulness), the way in which we express ourselves (clear, concise and structured) and the content (to be relevant).
- Conversation. A conversation is much more than just a question and an answer. The next big step we need to take and work on as designers is “creating a dialogue”. Ideally, we should consider use cases that are more complex and design conversations in a way that the user can ask another question after the conversational interface has given an answer. For example, I asked Siri this morning: “What do I have on the agenda today?” and it replied perfectly that I had an appointment with my doctor at 5pm. Then I asked Siri: “How do I get there?” (the address was in the appointment information), and it replied: “Where do you want to go?”. Siri understood the question independently from the previous conversation flow. Therefore we were unable to “have a conversation”.
- Personalization. Know the user you are talking to, greet him or her with their name and offer relevant information will allow you to obtain an optimized user experience with your conversational interface. In any case, identifying predictive use cases can help improve the experience significantly.
- Disambiguation. In some cases, bots don’t understand everything the user says which is why they can’t always give an accurate answer. In these cases, it’s important to clarify the user’s intention before continuing. Questions such a “Did you mean…?” can help. Furthermore, it’s important to take into account the context given that there are situations where clarifying the user’s intention could cause an undesired effect. For example, if I live in Cartagena (Spain) and ask the bot “What’s the weather in Cartagena today”?, it should offer me information about the city I live in instead of asking me which Cartagena I am referring to out of the 30 that exist in the world.
How can you design the personality and voice of a bot?
Different studies demonstrate that people associate human traits such as gender, personality and emotions to all types of systems: cars, television and computational systems. Keeping this in mind, it’s important to not leave the bot’s personality to chance because it could communicate values we don’t want it to. This is why at the start of any conversational project it is essential to have defined the “persona” or profile of the bot where different aspects will have been decided on: feminine/masculine, age, motivations, frustrations, what values we want it to transmit, vocabulary and tone of voice. We also need to have had defined other key characteristics about its voice such as maturity, intensity and speed. In this respect, it is interesting to carry out a “Voice test” with users before choosing the most suitable voice for your conversational system.
How, why and in what context do people interact with a bot?
It is necessary to understand human behavior and human reactions in order to design better conversational interfaces and better products in general.
It is important to remember user-centered designs when designing conversational interfaces hence the need to understand user needs and context of use before working on a conversational project.
This will help us identify use cases which we want to work on (not all of them will do!) and prioritize those that provide most value to the user. Moreover, as in the design of any product, the key is to test and iterate until we are happy with the final solution.
How do you create a bot that is more credible and likeable?
Interpersonal relationships have been objects of interest since the 1930’s, however, new questions have arisen with the emergence of conversational interfaces.
Studies prove that people can determine whether other people are similar to us and in what terms (gender, race, social class, personality and interests). Also, we have a special preference for those who are similar to us. The same can be applied to conversational interfaces, even though the person knows that they are talking to a “machine”. As a result, conversational interfaces that resemble the speaker are more credible, likeable and inspire more trust.
This fact introduces some concerns when it comes to designing conversational interfaces:
- How can we design a conversational system that is used by several people?
- How can we make the system recognize users’ main characteristics and emotions?
- And finally… how can we choose the most suitable “persona” to interact with a specific user?
What does the future hold?
According to Forbes, voice-fist devices will be the next great step in conversational technology. These devices use voice as their main interaction, but voice isn’t their only interface. One example is Echo Show by Amazon which has a screen that offers a specific type of information to the user.
This type of device offers more sophisticated experiences as well as improvements concerning the cognitive background that the user needs to interact. One of the most clear examples is that the user doesn’t need to remember the different options in order to make a choice because he or she can simply see them on the complementary screen.
IBM spoke with 30 scientists and leaders about the future of conversational technology and most of them agreed that advances made in human-machine interaction depended greatly on Artificial Intelligence. Thus, in the next 3 to 5 years progress made in artificial intelligence will improve conversational capabilities to achieve conversations that are more effective and sophisticated. From there on, a world of possibilities to integrate artificial intelligence opens up. A world beyond what we have already experimented with artificial intelligence until now.
There is most definitely a fascinating revolution underway, it has a voice and it’s here to stay.