From Awkward Silences to Magical Moments

Designing meaningful Conversational Interfaces through Relevance, Trust Building & Grace

We’ve grown accustomed to having digital assistants around us. Whether we’re home in the kitchen and ask Alexa to order Lasagna ingredients, query Google Assistant on our smartphone for the best route to the gym or interrogate Siri on our Apple Watch about Stock quotes: Our devices speak to us, and we speak to them. They haven’t quite become the omniscient butlers we’ve been expecting from Science Fiction… yet. But the applications of Conversational User Interfaces (CUIs) are manifold and some of their functions have begun to make a difference in our everyday lives.

“They haven’t quite become the omniscient butlers we’ve been expecting from Science Fiction… yet.”

It is no surprise then, that branded applications for conversational interfaces have developed into a hot venue for companies to present their products and services. Not only is it a good idea to join the ride early as not to miss out, but it’s also a smart way for companies to find entirely new customer bases or broaden current ones while Tech giants churn out smaller, smarter Voice-enabled devices in increasing frequency. Yet, not all conversational interfaces are created equal, and with the diversity in hard- and software involved, it is wise to weigh the pros and cons of each platform. Whether you are considering building a Skill for Amazon Alexa, want to make use of Google’s Dialogflow, aim for Apple’s HomePod or are working on another platform — There are a number of concepts that transcend the device-specific requirements and may be relevant to all Conversational Interfaces.

USEEDS° is a Berlin-based User Experience Design Consultancy with over a decade of expertise in the Banking, Home Automation, and Insurance industries, among others. Along with our background in Customer Experience Strategy, we have been developing voice- and chat-based interfaces in collaboration with our customers and the community in the past years. Our experience helped us understand the many opportunities that Conversational Interfaces entail, but also enlightened us about its many challenges, both technical and conceptual. As a result, we have identified 3 Tenets to creating meaningful experiences in Conversational Interfaces:

  • Maintain Relevance
    Users may have fun playing around with the more whimsical features of a Conversational Interface, but they will come back when it touches on their real-life needs.
  • Build Trust
    Many CUI applications employ user data that is highly sensitive in order to personalize the experience. Establishing trust in your Skill or Action can have a spillover effect on the overall experience with the target interface.
  • React with Grace
    Errors happen and connections get dropped — but reacting gracefully and honestly can make the difference between frustration and ease-of-use
Relevance: What are the Users needs? How can a Conversational Interface further their goals?

Maintain Relevance

As an organization steeped in the idea of a user-centered approach, we were wary to simply adopt a new technology because it was forced upon the market with the enormous buying power of tech and retail giants. However, we understood this technology to be a brilliant chance to re-engage the customer, bringing them back into the loop, and tightening the gap between product and user. Conversational Interfaces have the great advantage of engaging the customer in their private environment, being context-aware, and reactive to their needs. There is a chance of reaching the customer in the moment they seek an interaction with your company, sometimes creating a nearly magical moment of Unanticipated Service Bliss. But for that to occur, it is imperative that the users needs be respected.

There is a chance of reaching the customer in the moment they seek an interaction with your company, sometimes creating a nearly magical moment of Unanticipated Service Bliss.

Interactions become magical when real-world use cases meet context-relevant service that dovetails with how the User wants to be addressed. For this to occur, many hours of research need to take place, and an honest soul-search to pinpoint real user needs.

Just a couple of months back, we worked with one of our closest clients, a leading German bank, to develop a prototypical Alexa Skill for hobby traders. As with our use case, banking and trading customers may not feel comfortable with the thought of having their account balances blurted out, but may enjoy being given the context of their financial well-being when thinking about investing into a given stock.

There are tricky questions that need answering before you should go forward and implement a Conversational Interface right away. For example:

  • Personas: Is the selected Conversational Interface congruous with your target persona or character?
  • User Expectations: What will the User reasonably expect the interaction to be like?
  • User Needs: Can your service help the User fulfill at least one of their needs?

Although facing a cacophony of user voices and the divergence of interests within your target group, it is important to discover a single viable Conversational Interface without overestimating the abilities of the target device. The limitations of what can be understood and processed at any given time need to be leveraged against user expectations. When both align — great! — but if not, maybe decide in favor of a different device or channel. If all else fails, you can always revisit your hours of research into real-world use cases: Maybe the user doesn’t want to wire money to an external account via Google Assistant, but would be happy to transfer a small sum between savings accounts. In that, it maintains its relevance.

In any event, Conversational Interfaces may not have replaced the clerk you brought your piggy bank to when you grew up. But they represent a huge leap towards a more personal, more tailored experience as compared to the still mostly impersonal experience of online banking and trading.

Build Trust: How can business ambitions be consolidated with safeguarding user data?

Build Trust

Still, none of this is relevant to the customer if they do not feel safe. Conveying a sense of safety is not easy with the limited capabilities of the devices currently on the market. Connecting an Echo or Google Home with your account can only be as secure as the infrastructure of Routers, Switches and Servers between the device and your company. Of course either of the vendors on the market assure their safety, but a huge chunk of responsibility still falls on Third Party developers.

There are many ways of making the most of the built-in security infrastructure of Amazon’s AWS or the Google Cloud, and it doesn’t fall far behind what’s currently possible in online banking (e.g. securing connections with HTTPS, employing Multi-Factor Authentication or using build-in tools to counter and prevent DDoS attacks). But to be frank, the main fault line still mostly lies between the user’s voice and the devices microphone: The intersection between error-prone voice recognition and potentially someone’s real hard-earned money. You should not only guard user data, but also convey a quasi-contractual assurance of its safety within the user interaction.

Speaking about safeguarding data: There are some understandable grievances when it comes to the touchy issue of personal data and how it is being handled by the devices you wish to create for. It is not your customer who needs to fork over data to be one of the hip kids. It is you who has been trusted with one of their most valuable assets.

Yet most likely, your primary concerns lie not with the end user but with your company and the benefits that employing a Conversational Interface may have for you. And that is fine, really. Just make sure that in pursuing your business ambitions, you are not infringing on the trust you are trying to build elsewhere.

Trust is something that, when built up, can be mutually beneficial. Especially in the case of Conversational Interfaces where the engagement may well exponentially increase with the amount of usefulness and trust the user experiences.

React with Grace: How can we create meaningful interactions that handle Errors gracefully?

React with Grace

When a device hiccups — and it will — it is only fair to inform the user. Of course Errors shouldn’t happen too often, but a well-placed reaction (e.g. “I am sorry, but there was an Error in my connection.”) along with an option for next steps (e.g. “Would you like to try again or can I help you with something else?”) goes a long way towards building trust. Explaining what is happening and (whenever possible) why it is happening can be build right into the code of the Skill or Action. Handling errors and giving the User a chance to react on it is at least as important as it is in conventional interfaces.

This is how a downtime or a bout of packet loss can evoke a sense of awareness on the part of the user and build trust. But remember, your Skill or Action is likely not the only relevant one for the user, and they will assess the entirety of the experience with their device. When the Pizza Skill had me order the wrong meal, who is to say that the Trading Skill won’t order a thousand shares in my name and ruin my portfolio?

This is how a downtime or a bout of packet loss can evoke a sense of awareness on the part of the user and build trust

At first, Siris courteous replies to unintelligible ramblings are cute. But when errors or misperceptions are repeated, it loses its charm quickly. After all, even little kids quickly find out that there is no equally sentient being behind Alexa’s soothing voice, but a machine. And that is okay for now, if handled with Grace. It can even play to your advantage. Acting out interactions with your target device is therefore mandatory. This is how you can identify possible errors, hiccups or misunderstandings. Ideally you should have a professional screenwriter or someone with an equal set of expertise on board in your team. Someone who may turn an awkward silence into a funny comeback, a frustrating back-and-forth into a micro-moment that lingers.

Create Magical Experiences

Suffice it to say that the current generation of devices are not yet equipped to cover every possible awkward silence, but they have improved considerably in the little time that they have been on the market, and other devices and channels will follow. We are but at the beginning of an avalanche of digital assistants and it is wise to make them work for you and your company’s best interest. Just keep in mind that the technology in and by itself is not a cure-all. Take a good look at your customers first, gauge their possible real interest in your planned Action or Skill, and tailor an experience to their needs. You should consider your User and build a Conversational Interface that upholds our previously determined Tenets of Maintaining Relevance, Building Trust and Reacting to Errors and hiccups with Grace. It is then that you can transcend the Me-Too’s of the CUI world and create lasting, magical experiences that spill over to other channels as well.

Learn more about our User Centered approach to Conversational Interfaces at USEEDS° and please get back to us with Feedback, leave a comment below and share this story. Thank you :)

Make Sure to also read our other Medium Stories as well as Pascal Becker’s brilliant Intro to Designing for Voice

Further Reading


Action / Skill
Web Applications that interface with the corresponding Voice APIs of Amazon Alexa (“Skill”) or Google Assistant (“Action”)

Conversational User Interface (CUI)
Voice or Chat-based User Interfaces such as Amazon’s Alexa, Google’s Allo, Apple’s Siri. Usually leverages Natural Language Processing to parse the user input and allocate a requested service. Sometimes shortened to Conversational Interfaces or CUI.

Natural Language Processing (NLP)
A way for smart devices or computers to process language input and recognize spoken or written commands. It has existed for over two decades, but has seen a resurgence in combination with Conversational Interfaces.

Application Programming Interface (API)
A set of methods and protocols for different components of a Software or external services to correspond. A Voice API for example may be used to provide Voice Recognition capabilities or NLP to Web Apps to make it easier to develop CUIs