How to Strike up A Conversation – 5 Best Practices for Successful Chatbots

5 Best practices for successful chatbots

Have you heard the term “conversational marketing”?

It’s essentially a euphemism for AI-enabled chatbots (bots). In the near term, these bots are expected to revolutionize how companies interact with existing and potential customers. In fact, Gartner predicts virtual agents (bots) will participate in a majority of commercial interactions between people and businesses in less than three years.

The rise of bots doesn’t really come as a surprise to us. We’ve been implementing conversational apps, complete with natural language processing (NLP) into client solutions since 2010. We’ve developed bots across a wide range of industries - from retail to home security to healthcare – and along the way, we’ve developed the following five best practices for successful chatbots:

Understand the End User

1. Understand the End User – All the bells and whistles don’t matter at all if the bot doesn’t serve the needs of the user. So, understanding the reasons a user might initiate a chat, the experiences that led them to take that action and their resulting attitudes are critical to designing a successful bot. In our process, we use design thinking and empathic design to understand where stakeholders are coming from and what their needs are, so we can build chatbot features and functionality to meet them where they are.

Preset scenarios

2. Preset Scenarios – Regardless of whether or not you’re using AI to inform your bot’s capabilities or not, you need to pre-set your bot with a few scenarios that it is bound to encounter. These may be things like: onboarding, missing inputs or vague or irrelevant requests. The more cases your bot is programmed to handle, the better your bot will perform and the more satisfied your users will be.

Give your bot a personality

3. Have a Personality – Just because it’s a bot doesn’t mean it has to sound like one. By applying deep neural networks to NLP, today’s bots can communicate more “humanly”, and in turn, more effectively. With this in mind, give your bot a bit of personality - just make sure it’s one that aligns well with your brand. For example, for brands with a serious, pithy personality, a bot interaction might begin with a “Hello,” while more fun loving, or light brands might initiate a session with a “Hey there!” or even a “Hi!” Regardless of the personality that your bot adopts, make sure that it’s consistent within every interaction.

be bot-obvious

4. Be Bot-Obvious – No bot is going to sound 100% human 100% of the time. Rather than confusing and frustrating users, we suggest being upfront with them. Tell them directly that they’re interacting with a bot. This builds trust with the user and makes them more understanding if things don’t go right. Here are some ways to let users know that they’re interacting with a bot rather than a live representative:

   • Put the word “bot” in your bot’s name – e.g.

   • Initiate the first conversation with an intro from the bot and a brief description of what tasks the bot can perform

Handoff to a human if needed

5. Offer Human Contact – Even the best bot in the world will be insufficient at some point. A user will make a request that the bot is not programmed to handle or will speak a language that the bot isn’t coded to understand, and human interaction will be required. Make sure that your bot offers a clear way for users to contact a human who can help them when/if the bot fails.


These are just a few of the chatbot best practices we’ve gleaned over the years. To see these best practices in action and more information about how DVmobile creates conversational apps, integrates NLP into existing apps or helps companies develop their bot strategy, contact us!


 
blog-danny.png

Danny Sanchez

UI/UX Designer at DVmobile, Men-at-Work fan, and 3D printing guru.

z-blog-separator.png