Designing and Testing Conversational Interfaces

Designing and Testing Conversational Interfaces

Conversational UI - the future of interface



- Nidhi Rai, UXaudit.io
8th May 2018

We're in a new age of computing, where advancements in machine learning and artificial intelligence are creating an increase of interest in conversational interfaces(CI) and natural language processing(NLP). CI have begun to make a vital impact in the business/brand communication aspect. Though many businesses are realizing that CI is probably going to cause a major transformation in brand communication there are many who are doubtful about CIs. CIs have constraints, but they are here to stay and they are getting more advanced day by day. The major problem of identifying spoken input has been solved, but the challenge lies in building a user experience that's fashioned after the natural human conversation.
future of conversational interfaces

Conversational Interface Basics


CIs are basically of two types: Chatbots and Virtual Assistants (Siri, Amazon Echo etc.). The market available virtual assistants are mostly built by tech titans like Google, Apple, Microsoft, Samsung etc. Chatbots and conversational websites are the CIs that are usually used by small-scale businesses. In this article we will discuss the CI designs with respect to chatbots and conversational websites/apps. Certain design guidelines/principles needs to be followed while building a Chatbot in order to ensure friction-less communication.

Designing a CI

  1. Give your Bot a personality

    The aspect that separates a memorable bot from an inefficient one is personality. Give your bot a personality. Figure out is your bot friendly? Is it straightforward? Does it use emoticons? Is it funny and crack jokes? Is there a core value that your company follows in general and that could inspire the personality of the bot? Is there a personality or a fictional character that your bot could follow?

  2. Design a conversation

    Spend some time reflecting on how people move through a conversation. If you can get the conversation right, you can design experiences that are natural to use. Also, the first impression of your bot is crucial. You even need to devise the exit experience. Maybe you can provide the user with contact information or hand off to a person, either way, this needs to be planned.

  3. Plan for a fallback option

    Consider this conversation shown in the image. The user is stuck in a nested conversation with this bot, plan for such scenarios by having the chat transferred to a human to avoid frustration among your users. Questions that be handled by the bot are answered then and there but if the bot is unsure the chat should get transferred to a human operator.

    fallback option for bot

    Source

  4. Authentication

    A San Diego TV reporter on-air talked about a girl who ordered dollhouse via her parents' Amazon Echo and this caused the Amazon Echoes in viewers' homes watching that channel to also attempt to order dollhouses. The reporter said:

    "I love the little girl saying 'Alexa ordered me a dollhouse"

    Alexa in the viewers house could not understand the tense and the context of the conversation and attempted to order doll houses for them too. A minimum authentication should be done by your bot before taking any orders and that could be done by storing information about the users voice or understanding the context the of the conversation.

    alexa as a conversational interface

    Source

Example of conversational interface


Magic is an SMS shopping assistance, the startup guarantees to deliver any item by texting the Magic number (83489), no app download needed! In its first three days after launch, Magic acquired over 1000 customers and delivered around 1700+ requests, varying from ordering take-out Indian food to booking trips.
example of conversational interface

Testing a Conversational Interface (CI)


We are in the age of chatbots and the value of ensuring the quality of your chatbot is very important. It is just like testing any other software but a chatbot needs a specific procedure and attention.
Here are few tips on how you can ensure your chatbot's quality.
  1. Testing the Conversational Flow

    It is most important to test the conversational flow since chatbots are based on them. The most critical factor, the personality of the chatbot is tested over this. Let's take a real-time scenario. You ask your chatbot to schedule a doctors appointment and you are halfway in the process and you need to change the timings of the appointment, in this case, the chatbot should be able to go back and change this detail. If it can't do it and what good is it. Also, your bot should talk in short meaningful sentences, no one likes to read lengthy and bulky lines.

  2. Testing the bot's understanding ability

    A chatbot should be capable of understanding any type of user input like small talk, curse words etc. Even though these conversations are unrelated to the bot's functionality, it should be tested for its understanding ability. Talk about irrelevant topics, and see if the bot is able to respond in that context.

  3. Error Handling

    This is an area which requires a lot of attention, programmers should handle exceptions in a way that users understand the error messages thrown by the bot. The reply should make sense with respect to the situation.

  4. Test Bot Speed

    Bot should be able to respond to user questions instantly, if it takes days to get back to the user then there is no point of having a bot. A Bot should have instant replies ready with utmost accuracy.

As designers discover and learn more - and customers use more - conversational interfaces will become advanced and enhanced. If we go by the trends of the technology industry conversational interface is the future.
With this I hope you would have got an overview of usability testing and all the peculiarities involved. I hope you enjoyed reading this article. Thanks for your time and feel free to add your comments below and let me know your thoughts about this article.




Nidhi Rai, UXaudit.io


I am a QA by profession with 5 years of experience on various automation technologies. I am highly passionate about experience design as an emerging field and an enthusiast to spread the importance of usability and usability testing. When I am not working I am mostly reading, writing, or traveling. I am from India but currently staying in the United States.


Follow me on twitter

FACEBOOK COMMENTS WILL BE SHOWN ONLY WHEN YOUR SITE IS ONLINE

ux audit facebook comments

Get latest articles in your inbox

Subscribe to our newsletter & we promise we won't spam