How it Started, How it’s Going
Reflections and lessons learned building our first voice app
City Guide — New York City Bars is a voice experience that helps people quickly and easily find suggestions for where to go out for a drink in NYC during the pandemic.
The experience was designed and developed by the remote design team of
, a UX Researcher and Strategist based in New York City, and , a Product Designer based in Toronto, Canada.How it started
In early April 2020, right around the time that our current reality had set in, we met through the Voice Tech Global Slack channel, and over a Zoom coffee chat, connected over our love of travel and interest in crafting digital experiences beyond the screen.
Though pretty new to designing for voice, we were both keen to translate our interest into practical experience. So equipped with a healthy dose of research and design expertise, beginner conversational design knowledge, zero conversational development experience, and an outsized dose of optimism and confidence, we began what would be an 8-month journey to the MVP release of City Guide — New York City Bars.
With the MVP launched, we wanted to take a pause and reflect on our design and development journey, namely, what worked for us, and what didn’t. We hope it will encourage you to take the leap to try it out for yourselves!
What Worked
Testing early and often
This project has been highly data-driven and iterative. Throughout the design process, insights from interviews and concept testing directly informed all design decisions, pivots, and iterations.
Our initial concept set out to combine storytelling elements with practical recommendations aka. “infotainment.” We imagined an app that would offer crowdsourced travel stories, followed by practical recommendations for restaurants, bars, city parks, and cultural activities.
Initial use case: Explore and dream about traveling to another place, experiencing a place through other people’s experiences and get practical recommendations for future travel plans.
Through Wizard of Oz concept testing, we quickly realized that this storytelling experience was not a good fit for VUI. We needed to provide more “info — “ and less “ — tainment.”
Kill your darlings
Even when we were pretty into an idea, phrasing, or feature we kept ourselves honest by making data-driven design decisions.
The user feedback that we collected from our first concept test made us realize that as enchanting of an idea we thought we had, we might just have to kill our darling. Taking user feedback we pivoted toward the design of a practical task: helping people entertain themselves during covid by giving them recommendations for places nearby.
Leaning into process
The conversation design process that we used to develop the app followed the non-linear design thinking model. By leaning into a familiar process we were able to better navigate the uncharted territory of designing for a new interface.
Since we were collaborating remotely from kickoff through to launch, it also proved helpful in keeping us aligned on next steps. By following process and establishing a cadence of weekly meetings we were able to keep up momentum and use our face time to synthesize, define and build. Special thanks also goes out to collaborative tools like Google Workspace (docs, sheets, draw, etc), Miro, and Figma!
What Didn’t Work
Best Practice…IS best.
At the outset of our ideation process, we were pretty stoked by the idea of creating an app that could provide both information and entertainment value. The magical travel stories our user interview participants shared with us swept us in a rather romantic direction…one that our first concept test quickly revealed was not quite right.
As touched on above, the data from our initial concept tests informed a pivot that aligned directly with conversational design best practice of keeping things simple and practical. Once redirected, we set out to create an experience that provides value by helping users accomplish a task easily.
Use the right tools
Though more familiar with the research and design side of things, we were both somehow quite confident in our ability to develop a voice app. 🤷♀
Turns out Google Actions Builder isn’t as user friendly as we expected it to be, and so without advanced knowledge of how to code in the environment, we found ourselves limited to designing around basic functionalities.
For this reason, midway through the project, we considered switching over to Voiceflow — a prototyping platform for conversational experiences. Moving our concept to Voiceflow would have meant pushing back our release date, however. So to stay on track we opted to release an MVP with limited functionality and explore working on a new platform for the app’s first release once we had user feedback.
…and how it’s going
The MVP for City Guide — New York City Bars was approved by Google in early November and is available now through the Google Actions Directory. Check it out here!
We’d love to hear from you
Have any thoughts about this process and the result? Tried your own project and want to share? Comment below! Like the article? Hit the clap button! 👏
Last but not least
Let your (over)confidence be your guide! But really — if you’re interested in Voice Tech and are keen to try it out for yourself — go for it!
Interested in learning more about designing for VUI before taking the leap? Check out “The Ultimate List of Conversation Design Courses” to help get you started!
