How to design a chatbot: creating a conversational interface

Image for post
Image for post

A conversational interface enables users to interact with a machine through a combination of a natural language with buttons, menus, images, and other graphical UI elements. Currently, two types of conversational UI prevail: through speech and through the exchange of messages with a human or a chatbot. The latter is thus a narrower class of conversational interfaces — a computer program that communicates with the user through a chat interface.

This relatively new form of artificial intelligence is taking the world by storm, but the world doesn’t seem to mind. From now on, it’s not us humans having to learn how to deal with computers, but they will be learning how to interact with us! The benefits that chatbots offer both to business and to consumers are too numerous, especially as the conversation view of the most popular messaging platforms confidently becomes the majority’s home screen.

Image for post
Image for post

If you are going to create a chatbot for your company, here are five essential chatbot design rules:

1. Approach chatbot design with the end-user in mind

Chatbots always should be created to assist users. Service automation may be one of the goals, but chatbots are best to use when:

  • human service agents can’t process requests quickly enough to satisfy the end-user;
  • the user has to supply financial, medical, or other sensitive information which might be judged by humans — but not by a computer program;
  • it is critical to engage the user, personalize each interaction, and qualify the lead in real time, such as in the case of landing pages.

Conversational UI is all about the user experience and satisfaction. Think how to design a chatbot that will be faster and better at providing value to your target audience.

Learn the target audience’s needs and communication methods and offer a UX design which they’ll enjoy without leaving their habitat. For example, messenger chatbots already enable brands to build interaction with those customers who wouldn’t download their apps. If you are trying to reach teenagers and young adults, it’s best to launch your bot into popular messaging apps.

Then, you need to determine what your chatbot interface should and shouldn’t do for the audience. Should it only respond to general support queries? Can it handle the whole onboarding process for new customers? Think of it as a user story and try to keep it simple.

Do some research by testing the bots’ flows, i.e. paths the user can go down to achieve the desired result, across Facebook Messenger, Slack, and other platforms. See how the dialogue, onboarding, suggestions, dead ends and natural language understanding happens. Remember that a useful bot must provide access to the required service or information in as few steps as possible.

Now it’s time to design bot conversations for specific situations. Address the future visitors’ primary issues in the first screen of your interface and guess the subsequent steps from there. One to three main flows are going to make the bulk of your script, but the UX designer should figure out all possible paths the users could take. Instead of focusing only on ‘sunny day’ scenarios, try to predict all cases, including the handling of actual and seeming errors. The more user stories you have and the more flexibility you allow for your bot, the more likely people will use it again.

Prototype your conversation flows and test multiple responses to see which ones work best. Every error can increase abandonment, but can also teach you what users want to do with the interface, what its limitations are, and what opportunities might add value. If you’re paying attention, the bot’s every interaction with a customer is an opportunity to steadily improve your chatbot design.

Сonversational interfaces can and should be adapted to the needs and preferences of different users. Groups of users should be equally able to use one without the need to learn anything. Offer interactive buttons, message menus, cards, carousels, images or audio to make your conversational UI more comprehensible and better present the necessary information.

You can’t customize the visual aesthetics and texts on each page of an app to provide maximal value to each type of users. However, the ability to ‘remember’ a user’s requests, preferences, defaults, and topics discussed can help conversational interfaces deliver varying content in varying ways. A user’s history of orders, previous support issues, communication with different departments, etc., should be shared and integrated into future interactions.

Your conversational UI shouldn’t force users down narrow paths or teach them what to answer (even if you would like them to). But you can and should provide direct option prompts like ‘Yes,’ ‘No’ or ‘Read this story’ to help the dialogue progress. A good conversational experience also implies the validation of user input and managing expectations. Include words like ‘Got it,’ ‘Sure,’ and ‘Thank you’ to tell the user the chatbot is ‘listening’ to them. Make sure the conversation is customer-initiated and can be blocked easily at any time. Create a convenient exit experience, which may include the provision of contact information or hand-off to a human.

And the last, but not the least, piece of advice in this chapter is this: accept the fact that some users wish to speak only to a human. Let the bot automate as much as possible, but leave it to professionals to deal with more complex requests from a user. Provide some fallback and ensure you have a human service agent available if the conversation has been brought to the point where their help is required.

2. Build a simple conversation for a chatbot

Conversational interface design primarily deals with a conversation. Good user experience requires simplicity, intuitive interfaces, and maximal similarity to a natural human conversation. Prior to writing a line of code, think how you expect the dialogues to flow and solve the users’ problem(s). The chatbot’s conversation topics should be kept simple and strictly close to the subjects it was created to resolve.

Image for post
Image for post

Try to narrow the interface’s scope. The less the bot does, the less integration work you have to do and the lower the odds of the conversation going off the rails. Design conversation flows with as few interaction points as possible to give users the shortest path to the desired result.

Selectable button options, prompts, and other graphical UI elements will support the dialogue flow, but the language you use is your primary instrument for delivering a superb user experience. Asking an open question, give an example and show what formats of the required data are valid. If the bot fails to understand a question, it should be able to ask clarifying questions, rephrase queries, and provide prompts for solving errors. Provide one clear call to action for each interaction.

Remember that you are designing primarily for mobile and need to retain the users’ very limited attention span. Be concise and break up long utterances to make them easier to skim. Reduce scrolling by spreading information horizontally, in carousels. Provide smartly designed buttons so that the user can take action without having to go through large chunks of text.

3. Focus on creating a bot personality

A chatbot’s style of message generation and assistance to users differs from what a human can do. Its communicative activities can’t fully resemble those of a human being — but should they? Creating a personality for your bot is a way for the customer to enjoy the service, but also understand it’s not a human they are speaking with. Who then?

Amazon’s Alexa, Cortana, OK Google, and Siri all speak in a different manner and yet fulfill their purpose. A unique, if not ‘humanized,’ bot personality will not only set you apart from the competitors but also hook the users through empathy.

The first impression of your conversational UI is crucial, especially if you are obtaining users through advertising. Users may imagine the bot’s ‘personality’ or gender even if you hadn’t designed any. To have more control over the bot experience, it’s better to endow it with a personality from the onset. A distinct bot personality also helps reduce the number of topics and the words that will be used in the interaction. Luckily, the text-based format of conversational interfaces provides room for creating any personality you fancy.

Get as much information about your target audience as possible. Take your time to nail both an appealing suitable personality and user’s flow when engaging with the bot.

Facebook Messenger, Slack and other messengers have their own UIs. There is nothing you can do about it but design the interactions and dialogues your bot will have with the visitors. However, if you have an independent app, you can develop a unique visual conversational interface design too. Finally, if you are developing an interface of your own, consider having dynamic avatars that can convey nonverbal information with facial expressions and body language — like we humans do.

Image for post
Image for post

Translate your business’ digital personality into that of your conversational UI. Personality is strongly correlated with the language. What would your brand say if it were a person? Would it be funny or to the point? Would it use emojis? The latter, used as click-through CTA for each story, may be quite engaging. The element of unpredictability in your chatbot’s repertoire will prevent boredom and keep your customers interested.

If possible and appropriate, be creative with the color, typography, imagery, GIFs, and copy. Invest in quality copy to increase user engagement and form a close relationship with your copywriter — every word matters for winning your customers over.

With the recent advancements in artificial intelligence and machine learning, we’re closer to human-like digital assistants than ever before. Emotion analysis enables an intelligent device to understand the user’s feelings behind an utterance and then respond appropriately. Do user research to craft a compatible chatbot counterpart to engage with them. You may even need to go as far as to develop multiple personas for the same chatbot.

4. Simplify input structuring

Interaction with chatbots should always be easy for a user and save them extra typing.

Simplify text input for as many interactions as possible. The chatbot algorithm may offer multiple choice options or confirmation buttons with sensible defaults to let the user only click on a button instead of typing in text. If appropriate, enrich your conversational UI with emojis, animation, or other media, but make sure they are contextually relevant, used sparingly, and align with the brand.

Take advantage of the tools that minimize data entry tasks, such as the need to supply the user’s location or other info. (Such as Facebook Messenger which gives access to the user’s name, so the bot doesn’t have to ask.) Consider allowing the user to send attachments, stickers, videos, and photos to tell the system exactly what they’re looking for.

Design ways to make search easier. For example, a user might need to tap six times: 1) Women’s clothes, 2) dresses, 3) cocktail dresses, 4) black, 5) size 4, and 6) under $200. Not that many, but a user-friendlier bot would better search for a ‘black cocktail dress’ first (guessing the user is looking for women’s clothing) and only ask two intelligent follow-up questions. This approach applies to filtering results too.

Image for post
Image for post

Still, asking chatbots for help through typing and tapping is complicated compared to voice. Amazon Echo, Cortana, and Siri that rely on voice command represent literally conversational interfaces. One of these days, context-aware voice conversations will become the dominant chatbot interface — the required technology is underway.

5. Adjust bot dialogues to messengers

To ensure consistency within a particular messaging platform, chatbots need to be adjusted to the format of the messenger.

Image for post
Image for post

Chatbots must feel native to the messaging platforms. Because users expect your chatbot design to be controlled and adaptable within the guidelines of the particular platform, your conversation flows may look a little different across different messengers.

When you design a bot on Facebook Messenger and other platforms, you will find out that their customization capabilities are limited. Despite having little GUI design control, some messaging apps allow you to add product photos, icons, and menus, edit the color palette, use animation, sound, and video, and insert stickers and conversation landmarks.

Image for post
Image for post

Wrapping up

In 2017, we saw a lot of advances in natural language processing that facilitate the use of chatbots for regular user tasks. Simultaneously, many of the web’s recent success stories primarily leverage an automated conversational UI. As the area heats up, we are going to see a surge of commercial, personal and business messaging platforms, natural language integrations, machine learning, APIs, tools, and UX and UI breakthroughs.

Chatbots are great for onboarding new customers, so more sites, apps and other platforms go conversational. Messengers are the platforms of the future, and users will be increasingly using chatbots to access all sort of services. Take advantage of the extraordinary opportunities that big platforms and conversational interfaces present for your business. With the correct approach to chatbot design, they can become a useful tool for your company to expand your online presence to various messengers.

The main trends in conversational interface design seem to deal with use cases, bot personality, context, and voice-based interactions. Concentrate on creating practical and delightful conversational experiences:

  • Find your chatbot a place within the overall customer experience and narrow down its purpose and objectives;
  • Sketch the dialogue flow of your bot focusing on the key actions it will address for users;
  • Detail all dialogue relationship flows, sentences (verbs), subjects, user tasks, navigation and potential dead points;
  • Design contextual suggestions, options, and prompts for steering the dialogue forward;
  • Simplify data entry whenever possible.

Capturing and precisely visualizing a digital personality further helps design a bot that is useful, engaging, enjoyable, and converting. Would you like to design and build one with Onix?

Image for post
Image for post

FOLLOW US:

Image for post
Image for post

Written by

Onix-Systems provides IT services in website, mobile app and emerging technologies software development. Check our blog -> https://onix-systems.com/blog

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store