Everyone’s Guide to Designing Great Conversational Interface Experiences
Conversational interfaces (CIs) are rapidly shifting how consumers interact with brands, and the vast range of CI platforms is making poses a lot of challenges for organizations trying to embrace the technology. While there are a range options ranging from facebook messenger chatbots to Amazon Alexa voice user interfaces (VUIs), certain design elements remain consistent across platforms.
1. Why Are You Designing a Conversational Interface?
The first consideration is an honest understanding of what problem you’re trying to solve and who your intended users are. Why would these users interact with a CI instead of more traditional channels (email, call centre support, etc.)?
A true understanding of who your users are their unmet needs will ensure that the CI you design resonates with customers and increases the likelihood of strong user adoption.
The ideal way to gain this understanding is to speak directly with your target users. User interviews, focus groups and surveys all present specific methodologies to help collect qualitative data that should inform the design of your CI.
Quantitative research is equally important and can take the form of web analytics, call centre data and user/customer call histories. Analyzing historic data of how your customers have interacted in the past (across a range of channels) can inform how they would be most likely to communicate with a CI.
A combination of quantitive and qualitative research will help ensure that you’re building a contextually relevant CI. Where possible, primary research (interacting directly with your customers and their data) will result in more meaningful results than secondary research (reviewing research studies or evaluating competitive systems).
2. Consider the Range of CI Platforms
While there are a range of CI platforms, one of the key ways to segment these platforms is between text-based and voice-based systems.
Text-based CIs (known as chatbots) are more evolved, and as such, there are a range of common platforms. While not an exhaustive list, some of the platforms include:
Facebook messenger
Skype
Whatsapp
Slack
When it comes to voice-based systems, there are a few main platforms:
Amazon Alexa
Apple Siri
Google Home
Each of these voice-based systems requires more specific and deeper technical integration. For that reason, some thought should be given to an abstraction platform that allows organizations to create a conversational data layer that defines an interaction architecture that is agnostic to a range of CI platforms. API.ai is commonly used to provide some scale and future-proofing when it comes to designing Cis.
3. Evaluate CI Design Patterns
Conversational interfaces can take a wide range of forms, and depending on the context for which you’re designing, consideration should be made to the most appropriate approach.
Question & answer
In this design pattern, a user is taken through the interaction process in a step-by-step manner. Designing a Q&A CI involves creating a decision tree outlining the different paths that a user can proceed down.
In general, the questions that the CI asks of the user should be close-ended questions, rather than open-ended. Close-ended questions reduce the risk of your user responding in a manner that the CI does not understand.
Menu-based CI
Menu-based interfaces resemble traditional interactive voice response systems (IVRs) that we’ve all experienced while dealing with a central customer support line. Traditionally, these have taken the form of “dial 1 for sales, dial 2 for service…”
When you don’t necessarily know a great deal about why your users are seeking support, or they may not know the full range of options available to them.
Depending on the specific CI platform you’re employing, certain considerations should be made. For example, Amazon Alexa does not permit users to interrupt, so if you’re providing a list of 3 or more options, users may get frustrated waiting through the full list of possibilities.
Intelligent menu CI
Intelligent or secret menu patterns involve asking your user a question, without explicitly providing a list of the potential answers. Apple’s Siri voice-based CI is a common example, given the wide range of actions that Apple’s assistant can provide.
The intelligent menu pattern works well when there are a range of choices, but the options are all contextually related and relatively predictable.
An example where this pattern would work well is contacting your mobile phone carrier. While there are a wide range of options (add-on services, technical support, cancel a subscription, etc.) users generally have an idea of what they are looking to accomplish, and it’s unlikely for any “left-field” requests to come in.
Take the Next Step
Once you’ve completed some initial research and thought through not only how the CI should work, but also what platform(s) will best resonate with your users, it’s time to take the plunge.
Where possible, consider prototyping the solution in advance of doing a full build and release. There are a range of rapid prototyping tools for both chatbot-based and voice-based CIs. The tools below all work without the need for coding or technical development. While the list is increasing almost daily, some of the top runners in this group include:
Finally, keep in mind that launching your CI is just the first step. It will be important to get real-world feedback on how users are embracing the tool and continue to evolve the system based on this feedback.