7 Things You Know in UX That Will Help You Design Chatbots

At BEEVA we have been working with chatbots for some time now. In addition to the design framework I previously published, I’m going to share seven things that I brought from my UX design background that helped me approach the chatbots universe.

Visibility of System Status

It’s the first heuristic from Jakob Nielsen and one of the first you will find in every usability checklist. Applied to the chatbot interaction, it’s a must to have a kickass welcome message that sets user expectations and makes it clear which functionalities the chatbot will cover. It may seem obvious that chatbots only cover a small scope, but if you don’t make the scope clear, users will crash with everything your chatbot is not able to meet their expectations.

In addition to the welcome message, we need to remind users where they are and what they are doing. Users that are not paying attention to our chatbot can get lost many times on a single conversation.

Example of welcome message used on Maya a in-house chatbot for BEEVA

Personality

It’s a concept commonly underrated on interfaces, but it’s one of the fundamental basis of chatbots. It was perfectly studied on Designing for emotion, and the best example ever designed is the personality crafted for Mailchimp (here).

Because the interaction is happening on a chat, the user implication level is really high, and many users will receive the conversation as something real. Because of this, it’s extremely important to work on the chatbot voice and tone, the bot’s humor, and even its hair color.

My advice here is to create a user persona profile with your chatbot’s personality to guide all of your messages. If you have no clue about where to start, here is a template you can use.

Personality template

Interaction Elements

Words. That’s it… nope!

There are already some great articles talking about the interaction elements we have nowadays (this one with seals is my favourite one).

We could also make a timeline about the evolution and adoption of the different interaction elements, but the conclusion would be the same. The trend is to avoid any conversational interaction, replacing it with all sort of complex interactions, and transforming the old chat into something like an embedded web page.

Here you have an interesting article explaining how Google paved the way some time ago: How Google’s AI paved the way for the next generation of bots.

We have experienced real conversations bringing headaches and malfunction problems. This can be avoided by offering the users buttons and links to click on. Artificial intelligence and NLP are able to do amazing magic, but they don’t meet user expectations yet. You can read more about it here.

Every platform has different elements, and they change all the time, so you better update your knowledge before starting a project.

Some of the interaction elements on Facebook messenger

Information Architecture

Even in a chat we need to decide which information will be offered first and which information we are hiding. We can understand the chatbot flow like a screen flow with little information in every step, and different interaction elements on each part, where some actionable “buttons” are just words users can type.

Because of that I treat the information architecture as in any other interface by defining which interaction elements I will show on every step.

Navigation Flow

We need to design the navigation map for our chatbot in which we’ll define what will happen in every single step.

We should be thinking about every possible scenario on every step: which interaction element the user will be using, what will happen if they misunderstand any step like “What if they want to go backwards,” “What about skipping steps?” etc.

You can find more about how I do it on my Design framework for chatbots.

Basic navigation flow for Maya, a BEEVA chatbot

Prototyping Tools for Non-Developers

The same way we can find interface design tools, chatbots tools are beginning to spread. I differentiate them in two groups: low fidelity and high fidelity.

In the first group, there are tools that help us explain our chatbots and in a short time, we can get great results and be ready for testing much more quickly. There are a bunch of these, but my chosen ones are:

  • https://botsociety.io: Allows you to export the designed conversation on a video. They are adding more functionalities in order to allow more complex conversations and even different flows on a conversation. They are probably the leaders right now in this area.
  • https://botframe.com: you can generate a quite customizable conversation that you can export to png.

For high fidelity tools, we end up developing a real chatbot without writing a single line of code, but obviously with limited functional complexity. About them I would recommend:

  • https://octaneai.com: the position they are reaching in less than one year is something to be admired. I like their proposition about convos and the poll functionality.
  • https://chatfuel.com: it’s probably the rival to beat, but they haven’t changed very much in the last few months and their competitors have. They offer chatbots for Facebook Messenger and Telegram.
  • https://rundexter.com: It was the first tool I discovered even when they weren’t focused 100% on chatbots, but they remain quite aside, and they seem to be oriented to developers.

Analytics Tools

There are already a few players willing to become the Google Analytics for chatbots. Here are some of them, plus some other tricks that will help you understand what’s going on with your chatbot:

  • http://botanalytics.co: They offer a lot, but the results they give are still quite far from expected.
  • https://www.dashbot.io: The best one we have tested, but I still have some information missing… Besides the figures they offer I’d love to have information related to my chatbot performance and specially about funnels and how users are leaving them.
  • In-app solutions. Some of the chatbot platforms are already offering some data, like chatfuel and octaneai itself.
  • Direct access to Facebook Messenger conversations. Every chatbot on Messenger is associated to a Facebook page which means you can review the automatic conversation at any time, and even take part on it. Basically you can read every single thing a user says to your chatbot. Remember that the next time you say any bad word to an assistant.

Conclusion

Chatbots are one of the buzzwords nowadays, and they may eventually replace most of the interfaces we are used to. It will take time, and maybe the definitive chatbots and virtual assistants won’t look like those we have today. Don’t worry though, as a UX designer you are already a master of most of the things you need to work on its design.


7 Things You Know in UX That Will Help You Design Chatbots was originally published in Chatbots Magazine on Medium, where people are continuing the conversation by highlighting and responding to this story.

Source: <a href="7 Things You Know in UX That Will Help You Design Chatbots« >TheOldReader

Laisser un commentaire

Votre adresse de messagerie ne sera pas publiée. Les champs obligatoires sont indiqués avec *