Conversational Services

6 steps to successful conversational design

Share this post:

As anyone who has navigated through a complex interactive voice response system knows, it’s time for a different approach to automated customer service. Customers now more than ever demand experiences that demonstrate understanding of their needs and that empower them with the service they need, when and where they need it.

Fortunately, AI assistants can transform previously frustrating customer service experiences into opportunities to foster engagement and loyalty.

Designing Conversations

So, how do you build conversations that provide the best possible customer experience? Let’s dive into the fundamental aspects of conversational design.

Conversational design is the process of planning for the needs of your users, building dialogue for them to respond, crafting the flow of typical conversations, writing responses and creating opportunities for your AI assistant to learn and grow with your organization. The process takes in 6 main steps.

1. Question collection and understanding utterances

In the first crucial step, you’ll need to collect your users’ frequently-asked questions. Resist the urge to anticipate what your users might ask. Instead, train your AI assistant on what your users actually do ask on a regular basis.

From there, plan for how your users will speak. Teach your virtual assistant the shorthand, slang, acronyms, and all the other ways your users already communicate. You don’t have to provide every possible utterance. Just give the system enough examples to capture the typical ways a user expresses key concepts.

2. Ground truth mapping and intent clustering

Ground truth mapping allows you to align action to and provide clarity on like concepts your customers are likely to reference when engaging a conversational AI solution. This entails grouping phrases with the same meaning (intent). A good rule of thumb is to make sure you have at least 10 versions (‘utterances’) of a question for each intent.

For example, a user who says: “I’m frustrated; I haven’t been able to login in to the online billing system,” can be mapped to the intent Password Reset, which could trigger a corresponding action.

3. Designing the dialogue

Designing your AI assistant’s dialogue involves customizing the words, phrases, and Q&A that make up the user experience.

Designing the dialogue can be described in terms of three P’s.

  • Personality: The tone of the AI assistant and how that informs interactions.
  • Positioning: The purpose of the dialogue –its job description. Does it inform or take action for a user?
  • Proactivity: Defines how much your AI assistant leans forward and directs users versus sitting back and letting the user guide the experience.

4. Crafting the conversational flow

What steps do you want your users to take as a result of dialogue? Conversational flow leads them in the right direction.

To begin, think about your outcome first. Where do you want customers to go? What do you want them to achieve? Understanding that will inform the rest of the process.

To start a dialogue, welcome customers and tell them what kinds of questions the AI assistant can answer for them.

Next, guide users throughout the experience. For example, you might offer a number of different actions to choose from through multiple choice questions.

To finish this step, think about how to end the flow. You may want to transfer customers to a human agent, or end the conversation by saying goodbye.

5. Designing responses

You can design responses to make your virtual assistant’s interactions feel natural. Adjusting for tone, restating the intent, and introducing variation in the conversation all can help your users stay engaged.

At the same time, make sure your answers stay concise so that users can get the answers they need quickly. Inject personality into your AI assistant, making it more engaging, intelligent, and even fun, without asking your users to read a novel.

6. Continuous learning

Finally, build in continuous learning processes to ensure that your virtual assistant gains capabilities as your business develops. You will want to keep your solution up-to-date on any business and legal changes that will impact customer service, for example.

You should also monitor customer usage patterns. How your users interact will give you a lot of insight into how to update your AI assistant to make it more user friendly and helpful.

Avoiding Pitfalls

Along with all the benefits of conversational design come some potential pitfalls. Here are some common pitfalls we see that you should avoid:

  • Failing to leverage representative data
  • Trying to solve every problem at once
  • Failing to establish success criteria
  • Underestimating the time needed to train your bot
  • Lacking vision and alignment in your organization

Now that you’re armed with the keys to success for conversational design and know what pitfalls to avoid, you’re ready to create an amazing customer experience to meet your business needs.

Visit the Masterclass landing page and register for Episode 5

Director, Watson Expert & Delivery Services, IBM

More stories
September 21, 2018

Announcing the AI solution with voice for contact centers

IBM Customer Care Voice Agent is easy to set up and connect to existing call center infrastructure or telephone networks. That's part of the reason Forrester named IBM Watson Assistant the leader in their report, “The Forrester New Wave™: Conversational Computing Platforms, Q2 2018.”

Continue reading

September 12, 2018

6 steps to successful conversational design

Conversational design is the process of planning for the needs of your users, building dialogue for them to respond, crafting the flow of typical conversations, writing responses and creating opportunities for your AI assistant to learn and grow with your organization.

Continue reading

September 7, 2018

IBM named a “Leader” in Forrester Predictive Analytics and Machine Learning report

Watson Studio was named a “Leader” in the Forrester Q3 2018 Wave report on Multimodal Predictive Analytics And Machine Learning Solutions. According to the Forrester report, IBM Watson Studio is a “perfectly balanced PAML solution for enterprise data science teams.

Continue reading