Chatbots are the new black in UX design. Facebook already has over 11,000 chatbots since the official announcement of chatbot support made earlier this year. The question is: Should you build a chatbot for your business?

Yes, if your goal is to create a more personalized experience with your customers without hiring additional staff. You shouldn’t be perceived just as “faceless” website or a product page. Your bot can appear as a witty and helpful assistant with actual character.

Our app development company decided to hop onto the bandwagon as well and created a simple chatbot to talk with prospects on Facebook. April was born, and here’s the blueprint we followed to create her.

1. Prepare your toolkit

If you need a simple, intuitive, and robust tool to get started with chatbot design, try Chatfuel. It’s free, and it features a drag-and-drop interface, so you don’t really need to write any additional code. That’s the tool we used to build Alty Bot v1.

AI Weekly

The must-read newsletter for AI and Big Data industry written by Khari Johnson, Kyle Wiggers, and Seth Colaner.

Included with VentureBeat Insider and VentureBeat VIP memberships.

To create a more advanced Facebook messenger bot, look into Api.ai (recently acquired by Google), Wit.ai (helps to convert voice commands to texts), and Twine, an open-source tool for testing non-linear conversations.

2. Teach your bot to render different types of questions

Rob Ellis created a great library called qTypes, which has over 40 sub-classifications on how different question should be answered by chatbots. qType indicates the type of reply the users expect, and qSubType indicates what the question format is:

  • CH: Alternative Choice Question. The chatbot is asked to choose between two alternatives: Is this apple red or green?
  • WH: all the questions starting with who, what, when, where, why.
  • YN: Yes/No questions. Example: Do you want to order pizza?
  • TG: Tag questions. Not actual questions, but rather an option to keep the conversation flowing. Example: The weather is great, isn’t it?

3. Offer hints

Clearly outline in the opening lines what’s your chatbot is capable of doing:

  • Buttons and Facebook lists immediately hint at the next step and describe what the chatbot can do for the user.
  • Typing lengthy texts on mobile isn’t that comfortable. Besides, you’ll have to teach the bot to catch and render common typos and abbreviations, such as “Tue” for “Tuesday.”

To deal with the latter, you can use normalizer — it’s an open source library to help you convert British/Canadian spelling to U.S. English, fix 4,000+ misspelled words, and render common abbreviations.

4. Take advantage of Facebook native formats

Facebook lists can be used to display various kind of information based on the user’s question.

5. Always confirm user input

If the user’s answer is valid, confirm it before moving on to the next question to create a more natural conversation flow.

OK, travel apps, then. And what’s your project budget?

6. Always suggest what went wrong

You don’t want your chatbot to sound dumb and annoying, right? When the user typed an odd command, explain to them exactly what went wrong and summarize again what kind of information the bot can provide.

Bonus tip: Think twice whether a certain user input is critical for proceeding at this point. Teach your bot to make educated guesses and come back to the same question later in the conversation.

7. Work on the micro-copy

You might not need to make your chatbot as eloquent as Hemingway, but don’t forget that your chatbot dialogs stand for your product style.

Hence, take off your designer’s hat and wear a copywriter’s instead. When crafting the conversation copy, follow these tips:

  • Use the same user flow as when talking to an actual person.
  • Don’t make your bot sound too clever by using overcomplicated grammar and language structures. Keep it simple and be concise.
  • Don’t use gender-specific pronouns, as you never know who’s on that side of the interface.
  • Add a set of slightly different canned replies to make the conversation more human-like.
  • Add help messages and help intents whenever a user feels lost.
  • Create witty replies for unsupported topics so that the bot doesn’t appear dumb.

8. Sort out the navigation

Unlike standard interfaces, the message flow for chatbots goes more chaotically, as humans tend to have nonlinear discussions. When interacting with a chatbot people don’t just use it, they talk to it.

When crafting the chatbots navigation and user flow, take these points into account:

  • Don’t overweigh your bot with graphics. Keep it to the bare minimum and opt for third party integrations whenever applicable to keep the speed fast.
  • Reduce the required data input to the minimum, as a lot of people will use the bot from mobile.
  • Offer two ways of navigation: menu/buttons and free text input.
  • Always offer a clear path to the next step, using buttons or menus.
  • There’s no need for a back button, though, since users can just scroll the chat history.
  • Offer an easy way to connect to a human assistant when the bot can’t help.

9. Infuse the right type of personality

Last, but not least, make your chatbot seem more human. Obviously, your aim here shouldn’t be to pass the Turing Test (unless you wish to, of course), but to create a pleasant experience for the user.  Here are some tips to make your bot more likable.

Using emoticons and smiles makes your chatbot appear more human-like and appealing. Adding rich media — illustrations, stickers, and pictures — is a good way to make your bot stand out and improve your user experience at the same time.

Create your bot’s backstory. Creating a believable personality is another integral element to creating a bot capable of leading a human-like dialog and attract users. Adding self-awareness, irony, wittiness, and jokes, plus a mix of serious, scientific, and slang speech is another point to consider. “I’m just a bot, I don’t even have arms” always works.

VentureBeat's mission is to be a digital town square for technical decision-makers to gain knowledge about transformative enterprise technology and transact. Learn More