As AI continues to light up the news, one thing is becoming abundantly clear—it has the power to transform whatever it touches. And more companies than ever are reaching for conversational AI as they seek ways to serve customers better.
In Uniphore’s latest podcast, “Knowing How to Test Your Product,” retail product designer Kayla Orozco talks about what people get wrong about AI and how to build conversational AI products that matter. Here’s what she had to say.
Debunking a conversational AI myth
People would rather speak to a human than a bot.
Despite what some headlines might say, Orozco believes it’s actually bad chatbots that people hate, not chatbots in general. “When I speak to most people,” she explains, “or just through testing, I see that if it’s a good experience, it doesn’t matter that it’s a bot.”
Chatbots can provide a smoother experience for customers when well-designed. Conversational AI meets customers where they are, including in their preferred language and time of day. Chatbots offer a quick and efficient way for customers to get a resolution. In scenarios where the information is straightforward—such as asking about return policies, asking common troubleshooting questions or finding out business specifics like open hours—conversational AI is more than capable of managing the task in a human-like way.
And because there’s no one-size-fits-all approach to human versus chatbot interactions, companies must take the time to understand customer preferences. These preferences will depend on customers’ specific needs and expectations, something that reveals itself in customer data. Conversational AI technologies are constantly evolving, so monitoring customer preferences is key.
Tips for Creating a Powerful Conversational AI Experience
While there may be some general trends in user preferences, it’s essential to consider the specific needs and expectations of users when deciding whether to use a human or chatbot interaction. Here are Orozco’s top tips for getting conversational AI right.
Nail down your bot persona
Creating an engaging and effective bot persona can increase user engagement and encourage users to continue the conversation. When users feel like they are interacting with a bot that understands their needs and speaks to them in a relatable and conversational way, they are more likely to be satisfied with the overall experience.
Key elements of the bot persona include:
Tone of voice:
The tone of voice should align with the brand's personality and values while remaining conversational, friendly and easy to understand.
Language:
Factors such as formality, jargon and the use of emojis or other symbols can encourage users to complete an interaction.
Personality:
The bot persona should have a distinct personality that aligns with the brand image and values—think humor, empathy and helpfulness.
Appearance:
If the bot persona requires a visual representation, it needs to align with the bot's personality and the brand's visual identity.
Interactions:
If the bot persona requires a visual representation, it needs to align with the bot's personality and the brand's visual identity.
Context:
The bot persona should be designed to fit the specific context of the interaction. This includes understanding the user's needs and goals, as well as the platform or medium where the exchange is taking place.
Perform Testing with Real People
Once companies have built their bot, the next step is to get it into real human hands. Testing is critical in pre-production to ensure that interactions with customers are smooth and satisfy customer needs. Even if standard user testing isn’t available, company team members can still help.
Getting real people to interact with the bot has several benefits. First, it can help to ensure the accuracy of the bot’s responses. Developers can identify common questions or inaccurate responses and make adjustments to improve the accuracy of the bot’s responses. It can also help refine the bot persona by allowing developers to make strategic adjustments to the bot’s tone of voice, language and interactions to better align with user preferences and expectations.
Want to Learn More About How to Design Chatbots that People Love?
Check out our discussion with Kayla Orozco on Conversations That Matter.
Iterate Based on Those Design and Testing Results
Iterations help improve accuracy and ensure only relevant information is provided to customers. In preproduction testing and user feedback, companies can refine the bot’s persona and design elements. Iterations also allow businesses and organizations to keep up with technological advances and changes in the chatbot landscape.
Orozco also believes that iterations help overcome one of the most complex challenges of building a valuable AI product—language. Because language is complex and can carry multiple meanings, iterations help designers build these nuances into the product through real user feedback and understanding. This methodology leads to better results despite the often-perplexing nature of spoken language.
Explore Multimodal Options
Multimodal options offer users a variety of ways to interact with the bot, which can improve the user experience in several ways. Here are just a few:
- Accessibility: Companies must improve accessibility for users with disabilities or those who prefer different modes of communication. For example, visually impaired users may choose to interact with a bot through voice commands.
- Flexibility: By offering multiple modes of interaction, companies empower users to choose the one that works best for them in a given situation, such as when they are in a noisy environment and can't use voice commands—the opposite scenario from the previous example.
- Improved user experience: Companies can improve the user experience by meeting customers where they are, providing more natural and intuitive ways to interact with the bot. Again, some may find voice commands for certain tasks a more natural fit, while for others, text input is the most comfortable choice.
- Enhanced functionality: Multimodal options enhance the bot's functionality by enabling it to perform more complex tasks. Giving a bot visual feedback capabilities, such as images or videos, can greatly enhance its responses and provide another layer of value for customers interacting with it.
Building a Conversational AI Tool requires Collaboration and Persistence
Orozco emphasizes the need for company-wide collaboration to create something that resonates with customers. As conversational AI evolves, it offers companies a valuable way to interact with their customers, build trust and loyalty, and meet customer needs again and again.
Listen to the podcast to hear more about how Orozco plans, tests and deploys conversational AI and to understand her journey to building valuable, customer-centric products.