November 2017
Text by Nithya Krishnan

Image: © wutwhanfoto/

Nithya Krishnan is a user assistance developer and coordinator. In over ten years in the field of technical communication, she has authored end-user documentation across domains such as healthcare, mobility, database modeling and management, as well as enterprise solutions. With an academic background in information technology, her interests lie in creating a cohesive learning environment for all roles involved in the software development process.




Why click when you can chat? – Transitioning to conversational UIs

Conversational user interfaces have made their way into our everyday lives and they are here to stay. But what does this mean for technical communicators? How can we adjust our writing style away from the matter-of-fact tenor that has been instilled in us and towards a relaxed, conversational tone?

Not all that long ago, conversations between man and machine were a thing of science fiction movies or at least reserved for an unforeseeable, distant future. Today, the technology is here, and conversation is set to become the de facto standard in engaging with user interfaces. Whether servicing your car, checking your diary for today’s appointments or running your daily errands, it is all done through natural language and simple conversation. Say goodbye to conventional clicks and welcome the age of conversational interfaces.

New technology is forever changing the way we interact with applications. Since the advent of Graphical User Interfaces (GUIs), we have become well accustomed to visual commands such as clicking, dragging and dropping, selecting a button or selecting an item from a menu to accomplish a certain task. Conversational User Interfaces (CUIs), on the other hand, dispense with all these actions. You can now interact with your applications with the simple use of text and/or voice (which is later converted into text). Such conversational interfaces make the experience more natural for humans, and can even transcend the need for learning to use an application. Everything works seamlessly with natural language.

Many applications, one mouthpiece

In today’s digital world, we use a multitude of devices and applications and we use them everywhere: at work, at the store, in the gym, in the cab going through the city, and of course at home. GUIs have reached a threshold where our overall digital ecosystem is bursting with apps and interfaces. But most of these do not assimilate information. The interfaces we work with are not harmonized across all devices.  

CUIs provide a common channel through which we can interact with all applications on all devices. In fact, we don’t even need to remember which device or application we have stored a certain piece of information on. Using CUIs, users can concentrate on the task at hand without having to think about what app to use or which UI works best. Digital assistants such as Google Assistant, Apple Siri, or Amazon Echo help us to achieve this. Enabling these assistants with conversational UI capabilities completely transforms the way we interact with systems. And how? By using simple natural language.

Chatbots and digital assistants

CUIs in the current digital setup are conversations between a human and a computer. This is enabled by what we commonly call chatbots or digital assistants.

In the context of an enterprise or a business, a CUI must have the ability to react to questions initiated both by the user and by itself. It must be able to fully understand the business data at hand. In this way, a digital assistant not only provides the data for the request initiated but also provides information for a suitable course of action. Another important point is that a digital assistant remembers past conversations, so that it can pick up where it left off. This includes past interactions as well as preferences. Getting a service performed through natural conversation simplifies processes, enhances the experience and actually involves less training on the user’s part. This holds true even for inexperienced or occasional users.

Writing for CUIs

When we look at CUIs from a development aspect, it seems obvious that the bulk of the work lies in design and programming tasks. However, what we tend to overlook is the crux of the CUI. The conversations! And this is where we, as professional technical communicators, come in.

Looking back, the challenges that we now face creating conversations for CUIs are unprecedented. When GUIs came into the world of applications, we had to adjust to a new space-constricted, keyword-focused style of writing. However, as technical writers, we already knew a thing or two about precise texts, labels, messages, titles, procedures, navigations and so on. But so long our focus has always been on explaining the why’s and how’s.

Most of the time, writers across companies follow certain templates and guidelines to document product-related content. These guidelines have led us to develop an academic and formal writing style that is not suitable for conversational UIs. So, how can we change our writing style from an academic or technical one to a more conversational tone of voice?

Perhaps you are thinking, “Why would that be difficult?” or “Finally, I can really get creative.” Unfortunately, both assumptions are incorrect. The challenges of writing natural conversations should not be underestimated. Language is the major medium framing a conversation. How to keep a user engaged through the course of a conversation, in which sequence the steps should appear, how to respond to user reactions, are all essential questions in the context of CUIs. In simple words, the crucial part of creating conversational experiences is the conversation itself.

This shift from graphical user interfaces to conversational ones necessitates a new skill set; it requires user experience professionals with a deep understanding of narrative and conversational design and writers who can help frame the conversations in a natural way.

The integration of conversational UI capabilities into digital assistants humanizes the way we interact with computers. These digital assistants make conversations more context-based and intent-based by taking advantage of technologies such as Artificial Intelligence (AI) and machine learning, resulting in a more human-like interaction.

Turn-taking, context, and threading are all part of a cooperative or fruitful conversation, an idea popularized by linguistics philosopher Paul Grice. Grice called this the "Cooperative Principle". He also developed Grice’s Maxims to define the essential conversational rules he observed – namely, that people should be as truthful, informative, relevant, and clear as possible when talking with each other.  A conversational UI should follow these inherent rules of cooperation as well.

  • Quality – the speaker conveys only truthful information
  • Quantity – the speaker provides as much information as he can
  • Relation – the speaker provides relevant information that is pertinent to the topic being discussed
  • Manner – the speaker organizes the information and avoids ambiguity and obscurity

By leveraging the conventions of natural conversation, chatbots can be created in such a way that people intuitively know how to use them and feel comfortable with them.

The future of conversational UIs

It seems evident that CUIs are here to stay and will keep moving a notch higher each day on the technology adoption curve. With more AI and deep learning algorithms, CUIs will certainly gain popularity. As we know, emotions are the factors that influence conversation the most. The better CUIs can master human emotions, the more mature they will become in catering to varied user needs. Interpreting emotions through voice or facial expressions, or even understanding physical conditions (for example based on your body temperature), are areas that will enable this advancement.

With the proliferation of devices and their varied interfaces, CUIs will help bridge the gap between humans and technology and enable users to carry out tasks using simple and natural language. In the near future, CUIs and digital assistants (as the extended version of CUIs) will become the preferred UI for digital natives.


Building your chatbot

So you’ve decided to build a chatbot to cater to your users’ needs? Here are some basic steps that you can follow:

  1. Setting goals and defining a persona
    Before constructing your chatbot, understand the service that your CUI needs to offer the user and set the desired goals. We build relationships not only through the information we exchange with one another, but also through the emotions with which we deliver this information. A chatbot must convey a personality of trust and emotion. Therefore, it is crucial to define the persona of your chatbot and instill in it a character that reflects your brand. Instead of delivering information to users in an automated format, the chatbot must aim to be conversational and cater to the service it is created for.

  2. Structure and flow
    Structured conversational flow is at the core of building effective and engaging conversational interfaces. Any breaks in conversation are generally very disturbing and interrupt the flow. Your chatbot must be intelligent enough to bridge these breaks to make the experience more fruitful. It should always drive the conversation forward, for instance by helping users discover additional functionality, or by providing actionable phrases or buttons to redirect them to a place that might solve their issue. Keep conversations natural by giving the user space to speak or write.

  3. Content matters
    Of course, content is the foundation of every chatbot. Determine the various entry points to the CUI, possible moods of the user, more and less likely questions, and so on. The intention behind a chatbot is to save time and to direct users to the right source to get the task done. If your chatbot doesn’t fulfill this task, it’s not much better than any website or app. While designing the content for the chatbots, you have to create boundaries for the conversation by giving your users either buttons to select or actionable commands to use. Don’t leave the conversation open to interpretation. It is also a good practice to repeat the information back to the users, so they feel more comfortable knowing you got it right.
    And remember that you don’t want your users to feel like they’re talking to a machine. Use friendly, inclusive language. Attaching context to the conversation helps. This gives a more personal touch and a whole new dimension to the conversation.

  4. Build your conversational script
    With the flow and content in hand, you can create chat clusters and determine (on paper) what the overall conversation could look like.

  5. Consulting with developers and deciding on a platform
    Understand how you can digitize your overall script with the help of your colleagues from development. Upon completing the coding of your chatbot, you can decide which platform best suits your business needs such as Facebook Messenger, Slack, Telegram, WeChat, and so on. Select the platform based on your target audience and the user experience it offers.




Page 1 from 1
#1 Ambarish Palnitkar wrote at Fri, Nov 24 answer

Nice write-up!