Conversational Interfaces - chatbots interaction
New opportunities and challenges for both designers and developers
There are several articles that talk about expectations for 2017 and try to guess which are the trends reserved for the new year and which are the challenges reserved for UI/UX.
It is no longer surprising that, contrary to what happened in the past, it is the companies that now adapt to users according to their needs and habits. These needs are now more dynamic and habits change almost overnight.
Recent studies point to new user behaviors. People no longer install 3 to 4 applications per day. There is no longer that anxiety to know the news that comes out every week.
According to Adobe’s annual report, based on an analysis of 290+ billion visits from 16,000+ mobile sites, as well as 85+ billion app launches (survey with +1000 users in the US), consumers use applications that they already know and that guarantee them the service they need.
On average, the report reveals a 24% increase in existing app launches, as opposed to a 6% increase in app installs. In the best apps (TOP 20), the difference is more significant, the growth of launches is 62% versus 41% of installations.
Users now expect new ready-to-use services that are smarter and capable of performing complex actions
Due to greater competitiveness only the best applications, those that have real value for the user, survive.
On average, 5 out of 10 apps are used less than 10 times and 2 out of 10 are used once. There are several arguments for an app to be deleted but the most popular was “not useful”, which reinforces the initial idea.
Users want less quantity and more quality. They prefer to have only a small number of apps or services that allow them to access a greater number of interests.
While the need to consume digital content continues to increase, users now expect new ready-to-use services that are smarter and capable of performing complex actions.
This expectation influences how services and applications interact with users, thereby changing the way they communicate the intended message. The technological future passes through the ease of use and “human” interactivity and in this past year, the interaction between humans and machines has been the topic of many conversations and debates.
I still remember using my cousin’s computer with MS-DOS to play “Arcade Volleyball” for several hours. The interface was text-based and the interaction with the machine was a bit strict, with several lines of commands. In the 80’s a new type of interface was developed based on interaction through graphics, GUI - Graphical User Interface.
GUIs have transformed the way we interact with devices and made the messaging process easier and more accessible.
The adoption of smartphones, tablets, (…) and apps has grown at an impressive speed and this technological revolution has extended to the development and evolution of the interfaces, making them more intuitive, user-friendly and standard across devices.
However, as a result of user’s own expectations, several companies are interested in new alternatives for interaction and promotion of their services and products, and because of the emerging potential, CUI - Conversational User Interface comes as a new solution.
The aim of Conversational User Interface is the same as any interface - to communicate. GUIs use the traditional format we all know, using buttons, menus, icons, and patterns, while CUIs provide an interactive experience where the user initiates actions and performs tasks telling the system what it wants, replicating what It would be a conversation.
In order to enable interaction between computers and humans, we rely on graphical interfaces, but GUIs can act as natural barriers once the user has to learn how to use this type of system, for example that a button triggers an action or when doing swipe it passes to a next image, etc…
However, through conversation interfaces and their future improvements in language recognition and processing, it is possible for the system to understand everything that is requested and respond dynamically, allowing a more accessible and efficient user experience for certain services.
In fact, bots or virtual assistants are not entirely new. Before they became the topic of the day, some companies already used these systems as a marketing tool to promote products and to help with customer support services or Web searches (e.g.: SmarterChild).
The concept though already implemented, has matured, evolved and has become more intelligent and natural giving companies a new engaging way of communicating.
But then what is a bot or chatbot?
It is an interactive service that is based on a set of rules and in some cases in artificial intelligence, allowing the automatic interaction through chat.
A rule-based chatbot responds to specific commands but in case it works with artificial intelligence, it understands the meaning of the word and learns continuously through each conversation.
Bots can have their own applications, can be run through other apps or through physical voice devices.
Smart assistants are everywhere and have already been introduced by industry giants. Google Assistance (next generation of Google Now) can be found in the Google Allo smart messaging app and has also been integrated into Google Home and Google’s Pixel smartphones. Alexa controls and gives voice to Amazon Echo, a smart speaker developed by Amazon. Apple owns Siri, Microsoft Cortana and Facebook for their chat application developed the M, not to mention examples such as WeChat and Slack that allow the integration of several bots.
Imagine using a single site or base app to perform multiple actions?! Or running a requested task on your computer through talk rather than the traditional tap?!
With virtual assistants you can schedule and manage tasks, configure alarms, control multiple devices, get information, play music, make calls, send messages, shop online, etc…
Conversational interfaces will create new opportunities and challenges for both designers and developers. New opportunities, because we are witnessing the exponential growth of a different approach and a new way of communicating, which will allow us to develop new experiences. Challenges, because new forms of interaction with existing services represent new issues that need to be addressed.
The basic requirements and interaction flows with a bot are quite different when compared to the traditional model. The fluidity of information on both sides is presented through small responses at a time, and its “action-reaction” model for collecting and filtering data is mostly based on the chat system and not through visual controllers.
Reading several articles on this subject, I was able to group some design principles and good practices for the implementation of these interfaces:
Keep the conversation brief and simple as possible, avoiding confusion.
By initially suggesting the type of response we want, we avoid asking ambiguous questions to the user;
Work on good conversations.
It’s not enough keeping conversations brief or nicely-formatted, they have to make sense. Chat flow is the key element for successful bots. Some companies are even hiring copywriters to help with UX and bot development.
Understand user needs.
Prototypes and experiences with chatbots help to learn and understand the habits and behaviors of users. When we test a platform we get context for its design, many of them completely unexpected.
Break the monotony.
Conversations are naturally dynamic and the use of animations helps to capture the user’s attention by adding rhythm to the interface.
Do not use “human” indicators.
People should know when they are talking to a bot, at least I’d like to know. By using such indicators as “is-typing …” or artificial waiting times we are perpetuating a lie that can end up in frustration.
Give personality to the bot.
Although bots are not human they can and should have personality. To help with brand representation, it is important to structure behaviors. How the thing is going to behave? What will be the tone of the bot? Will it be more serious or more relaxed and friendly, robot type or crazy scientist?! Finally, it may make sense to be the user to choose, who knows ?!
Visually format bot responses.
Although we reduce all interaction to a simpler conversational interface, bot responses do not all need to be presented in plain text. Responses can and should be visually structured and formatted in order to prioritize information and improve the user experience (e.g.: card presentation).
Use graphical inputs.
There are several ways to present the same information. In some cases in response to a question, for instance, displaying “yes” or “no” buttons can save users time instead of forcing them to type in the text field. It may also be useful to use graphics when we need feedback from users, incorporating icons to complement the scoring system (e.g.: stars icons or hearts). Another great advantage, with the implementation of graphic inputs, is to limit the number of options available, controlling thus better conversation.
Reconfirm user options.
One of the biggest challenges with bot interaction is that its flow follows a story that hardly allows you to “go back”. It is important to give users the possibility to confirm an option, either because they are asked to reconfirm or in case there are buttons their choice is only sent after clicking “send”. The presence of buttons can even allow the inclusion of small explanatory captions before the user confirms their option.
Use autocomplete suggestions.
If these systems are based on chat platforms with text input why not take advantage of the autocomplete suggestions?! So when the user starts typing, suggestions for possible options appear.
Here are some examples of bots:
ChatBottle - Repository of bots, organized by category and by platform.
Your old new friend from the folks at uxdesign.ccuxdesign.cc
Adrian Zumbrunnen’s conversational website - Website turned into a conversation interface.
Here is also the article about the whole process:
I made my website conversational. Here is what I learned.uxdesign.cc
Cody chatbot website - It is still under development but I leave here the article about its construction process:
Bots should be used to enhance the user experience.
When developing a solution it is advisable to first understand if a bot is indeed the answer to the problem and if it brings any benefit to the end user.
For this reason, my opinion is not that doomsday is about to come, designers will no longer have control over visual interactions and flows, developers will be limited to pure logic scripts and that all we know and the interactions we experience will all be replaced by bots.
There is always the question that is rarely addressed and which in my opinion is intrinsically linked to all human, which is the visual attraction.
We are naturally attracted visually to everything around us and we make many decisions at the expense of that same attraction.
By saying this I mean that the visual interfaces run the risk of causing user paralysis because the user has to understand how the system works (not so intuitive), but the conversational interfaces can promote the opposite, total disinterest, nothing to analyze or anything that appeals to us and that in the end, we will not have any conversation at all. Bots are not human, they do not have to look like humans and do not offer opinions or suggestions of new ideas with feeling.
By interacting with an app or website, we establish emotional connections through visual integration and animations, enhancing the user experience.
I agree that conversational interfaces will be extremely useful for many applications, even because they perform actions much faster and more efficiently. Probably everything that involves tasks will be replaced by bots but the future of interaction with the multiple devices will be executed in several formats, either by integration of voice, gestures, conversation and visual.
My prediction for 2017 will be the greater adoption of chatbot technology and of CUIs - Conversational User Interfaces in more industries, since there is a great opportunity and interest in creating solutions that contrast with more traditional and static applications and interactions.
What is to come will certainly be promising and anticipating short-term advances and challenges, I would say it all boils down to the development of truly user-friendly UIs, betting on the symbiosis of environments where graphical interfaces support conversation interfaces.
Some services that can be used to build a bot:
IDEO Stanley-Bot - Stanley is a custom SMS messaging platform to prototype early services, products, and content.
Text It - Build multi-platform messaging & voice bots.
Botkit - A toolkit for making bots for Facebook Messenger, Slack, Twilio IP Messaging, and other messaging platforms.
API.ai - A tool to create a conversational Messenger bot.
Wit.ai - Easily create text or voice based bots.
Motion.ai - Create a bot for Facebook, SMS, Web-based, Slack, email and for most of the bot platforms.
Chatfuel - Chatbot creator for Facebook Messenger and Telegram.
Facebook Messenger Platform - Facebook platform that lets you create a bot via their official page.
Beep Boop - Hosting platform to developers who intend to create a chatbot for Facebook Messenger or Slack.
If you prefer to read in Portuguese or for those who want a good challenge in learning another language, this article is also published here :)