Conversational Internet is digitizing the other half of the world - Learn More
Build Gen-AI Powered Chatbots - Explore

Five tips to create a well-behaved chatbot

On May 23, 2016 | 3 Minutes Read
ChatbotsConversational CommerceConversational MarketingConversational SupportGeneric

“The medium is the message” – Marshall McLuhan

The number of chatbots is on the rise and the behavior of these chatbots will be a defining metric of its usability. A chatbot uses the conversation as its interface. Users do not appreciate magic, and that can be prevented by understanding the expectations of the user.

 

Here are five tips to create a well behaved chatbot.

  • Always introduce yourself – When a user initiates a conversation with a bot for the first time, it’s imperative for the bot to introduce itself. This means saying hello, establishing context to what the bot exactly does and introducing a couple of keywords to the user. Many bots also ask the user for his/her name or for a simple input to establish ease of conversation. Here’s a screenshot from the Techcrunch weather bot on Facebook.

Another good example is the CNN bot that does a good job of setting context and giving few examples to what the user can type in. It also gives the user the option of clicking a button to read ‘Top Stories’ or ‘Stories For You’

  • Never leave a user hangin’ – The cardinal mistake a chat bot can make is for it to not reply to a user when the user has asked something of it. The premise is instant conversation over a chat interface. When a user asks a bot a question, the chat bot has to always respond, even if it doesn’t understand what the question is. Some of the bots demoed at F8 received flak for being unresponsive. Most messenger apps provide a ‘typing’ indicator which is also useful as it reduces the anxiety for a user who is interacting with a chatbot.
  • Account for typos – Most people use mobile phones on the go and it’s easy to make typos. Damn you autocorrect, anyone? To make a chatbot as intuitive as possible, try and account for common typos to keywords that your chatbot responds to. After all you don’t want your user to repeatedly try and type the same thing.
  • Give easy access to ‘help’: An end-user might interact with anywhere from 10 to 100s of chat bots in a day so it is unfair to expect him/her to remember keywords for each bot. So it is imperative to provide easy access to a help menu that assists the user with context. A great example is the Poncho bot that is always on hand to help you interact with it.

  • Offer levels of customization: As a bot-creator you need to remember that each user has different messaging preferences. A die-hard soccer fan will want a play-by-play commentary for a soccer game but a casual fan might just need a score update at the end of a game. A good soccer bot will provide different levels of customization that adapts to each user’s preferences.

Many messaging apps such as Facebook and Slack have approval processes in place that check if your bot behaves as it’s supposed to. Guidelines such as these will ensure that your bot is approved easily.

Now get crackin’ on building some bots!

Team Gupshup
Team Gupshup

Blogs you will want to share. Delivered to your inbox.

Business Email

Recommended Resources

How Social Media Can Boost Customer Service Success?

Boost customer satisfaction with social media customer service. Learn how AI tools like Gupshup help streamline...
Read More >

Exploring the Benefits of Artificial Intelligence in Finance

Discover how AI in finance enhances customer satisfaction, streamlines operations, and drives smarter decision-making for financial...
Read More >

Data for Decision-Making: How Gen AI Chatbots Are Helping Businesses Achieve Better Outcomes

Discover how Gen AI chatbots transform business decision-making by turning conversational data into actionable insights for...
Read More >
×
Read: Why the Future of Bots will be Multi-Platform