Moe’s Voice Interaction
Team: Kelsie Belan, Myself, Suyash Thakare, Bang Tran
August 2018 – December 2018
UX Researcher: I conducted observations, reviewed background literature, constructed surveys, initiated pivot, wrote script for Burrito Quest, conducted feedback and user testing sessions, analyzed data, presented research results
For this project, Focus Brands tasked us with looking into applications of voice interaction technology for Moe’s restaurants and how it could fit into their brand.
Our research showed that although customers do not feel that voice ordering is easy and efficient enough to be worth using, there are other uses for voice technology that can effectively entertain users while engaging them with the Moe’s brand.
From the beginning, we knew this was going to be an exploratory project. Our sponsors had little idea of how they might want to implement voice interaction or even possible directions to go in. The open-ended format of this project was slightly daunting. How would we tackle such a broad prompt? Where would we even begin?
But it was also exciting. Within this problem space we had the opportunity to really embrace its exploratory nature and examine the aspects of the Moe’s experience, the potential of voice technology, and all the possible ways in which the two could intersect.
“Voice interaction is a novel and increasingly popular technology, and Moe’s Southwest Grill wants to determine whether or not the technology will be a good fit for its brand.”
— Focus Brands Representative
Phase 1: Research Exploring the Moe's experience, voice interaction, and their users
Observations and Desk Research
Why? Given the wide scope of this project, it was essential that we gain a deeper understanding of both Moe's and voice interaction so that we could identify specific aspects to explore in future research.
How? We began by going to a Moe’s restaurant to observe customers and features of the Moe's experience. We then researched the Moe’s brand to learn more about its personality, and also explored its app and website to understand more about the technology it was already using. We also delved into articles on voice interaction technology and its recent developments, functionalities, and user experiences.
Vibe fun, edgy, and laid-back
Features like funky artwork, rock music, and trivia
Pushes boundaries, doesn't take itself too seriously
Fun contests, promotions
Inspired by acronym Musicians, Outlaws, and Entertainers
Used for placing orders, managing rewards
App has more features like augmented reality
Used for simple commands, ordering, entertainment
Users like convenience, but still getting used to concept
More companies using voice to promote their brands
Why? The main objective was to gather data on customers' experiences with Moe's restaurants and voice interaction. Despite the wide scope of the project, we were curious to explore voice ordering since this was the most apparent intersection food and voice.
How? We used electronic surveys to reach a large quantity of participants and gather a variety of data that we could then narrow down into more specific research questions. We reached Moe’s customers through Focus Brands and asked about their Moe's experience. To keep the surveys brief, we asked half the customers about voice interaction and half about digital food ordering.
Digital food ordering
Liked convenience, speed, limited human interaction
Mostly used for pizza, restaurant websites, delivery services
Preferred ordering with screen over voice
Moe's experience preferences
Visibility of ingredients
Efficiency of ordering process
Control/customization of order
Atmosphere of restaurants
Rewards and coupons
Survey Results (n=25)
Used most for phone calls/texting, music, searching
Not comfortable using around strangers in public
None used voice to order food or products
Constructing empathy maps helped us understand the mindset of a Moe's customer and a voice user:
Competitive and Task Analyses
Why? We wanted to better understand the process of voice food ordering, its strengths and weaknesses, and ways we could leverage different features in voice interaction for Moe’s.
How? While our surveys were being sent, we conducted task analyses of voice ordering apps and skills including Domino’s, Wingstop, and Starbucks. Next we tested out some of the most popular Alexa smart speaker skills including The Magic Door, Jeopardy, and Thunderstorm Sounds, which were the top apps outside of smart home integrations.
We tested each program as a group while taking our own notes on steps, features, and any qualities that were helpful or interesting. We then combined our notes and discussed strengths, weaknesses, and ideas. To see an example of our task analysis, click here
Competitive Analyses Takeaways
Voice interaction can be very simple or very complex. Customization must be balanced with simplicity.
The interaction does not have to be disembodied, but can be grounded with visuals.
Chunking information can help users understand, respond, and navigate effectively.
Explaining essential features at start-up can help users be aware of functions and interact appropriately.
Using familiar sounds and phrases can help enhance the experience by anchoring the user in the context.
Voice interaction can be enjoyable, entertaining, and interactive in innovative ways beyond voice assistants.
The Pivot: Iteration of Research Design
We now had plenty of user data to understand what drove customers to Moe's restaurants, website, and app to create an to create an evidence-based design for the optimal voice ordering application for Moe’s. But there was one problem:
Our research showed that people weren’t receptive to voice ordering.
We realized that we had limited our opportunities by trying to fit the Moe’s voice experience into an application that people simply weren’t using. Ultimately people use Moe’s to order and eat food, which narrowed the potential for voice integration. The uses of voice technology are almost limitless, so we needed to learn what people already find useful and enjoyable about voice, and then shape those features to Moe’s for possibilities beyond the expected.
Ordering and eating
Fitting Moe’s into Voice limited the scope of possible applications
Entertainment, lifestyle, humor, productivity...
Fitting Voice to Moe’s yields possibilities for more innovative applications
Why? Following our pivot we decided to take a Jobs To Be Done approach to help us understand what jobs people "hired" voice for and why.
How? We constructed a brief survey asking participants what they used voice for and why, and what they found most useful and enjoyable about their interactions. We sent this survey to our previous Moe's customers who indicated that they used voice, but we also shared it with the general public on social media to capture a larger, more diverse sample of voice technology users.
Survey Results (n=90)
Used for practical assistance when voice was more more convenient than completing a task themselves
Also used voice for entertainment like music, games, and jokes
Used in cars primarily for hands-free functionality like texting or GPS
Wider variety of interactions when used in the home
No one reported using voice for ordering
User Needs and Implications
From our user research, we were able distill the following user needs and design implications:
options and process
Ability to control and tailor experience
Completion of tasks
in a timely and convenient manner
Ability to interact with voice discreetly
Access to opportunities for saving money
Phase 2: Concept Ideation How can voice capability fit into the Moe's brand, and what is most appealing to users?
With our user needs and design implications in mind, we moved forward with the ideation of several concepts before settling on three to sketch and evaluate:
An ordering application that
can be used with both voice
and a touch-screen
A voice-interactive game
with weekly trivia on Musicians, Outlaws, and Entertainers
An interactive story about a
cowboy searching for the legendary “Golden Burrito"
We ran feedback sessions of all three concepts with four participants. We ran the sessions by introducing each concept and then running them through a storyboard (for multi-modal voice ordering) or reading through an interactive script with the participant (for Moe’s Trivia and Burrito Quest).
Following each demonstration, we asked questions about their impressions, likes and dislikes, and recommended changes. At the end of each session, we asked them to rank the three concepts.
Feedback Session Results
Liked simulation and flexibility
of voice or visual
Thought voice complicated ordering process and wasn't necessary
Liked being able to play trivia
and earn rewards
Found ranking system confusing
Had fun engaging with the script
Liked having different paths and mini games
Concept feedback session using interactive script.
Photo used with participant's permission.
Phase 3: Prototyping and Evaluation How do you prototype and test a voice interface?
Although both Moe’s Trivia and Burrito quest were both highly rated, we moved forward with Burrito Quest
because interactive storytelling is a newer application of voice technology with more possibilities than trivia.
We wanted to explore a new territory that would benefit more from our research efforts.
We had to carefully consider how to build a prototype that would allow testers to fully experience interacting with a voice interface and choosing the path of the narrative. Our solution was a script that provided options, but ultimately led to the same outcomes, giving the participant the illusion of choice.
I outlined the structure of our story and then wrote a complete script. Our team members then recorded lines of the script, each of which were placed into slides in a Keynote presentation. The Keynote format enabled us to play the appropriate sound clips that corresponded with the participant’s choices
We conducted two sets of evaluations. The first session provided general feedback on the concept, its features, and its presentation in the prototype. We used the results of the first session to refine our research questions and gather data on specific issues that were raised.
To evaluate our prototype, we used a Wizard of Oz approach that simulated interaction with a smart speaker. In this scenario the participant actually spoke to a small external speaker that connected to a laptop where the Keynote presentation prototype was controlled by a team member acting as the “Wizard.”
After running through the prototype, we asked participants about their general impressions, preferences, understanding, and engagement, as well as their thoughts on some of the features like mini games and reward points.
Evaluation session with Wizard of Oz set-up using external speaker and Keynote prototype.
Photo used with participant's permission.
Sample page from Keynote simulation prototype
Burrito Quest Story Map
Evaluation Session 1 Takeaways (n=4)
Understood flow of story and input required
to proceed at each step
Liked storyline, voice acting, sound effects;
lots of laughter and smiling throughout
Enjoyed playing the shootout mini game
and the variety it added
Overall participants found it to be fun and engaging
Participants confused about how points were earned and what they meant
What is the best way to explain Moe's Rewards and how they were earned?
Some participants thought there was too much narration, others did not mind it.
Is the length of narration a matter of usability or a matter of preference?
Some suggested including a visual component to improve their engagement.
What visual could enhance the experience and would people find it useful?
Handout provided to participants with story map and explanation of reward points system.
Evaluation Session 2
In our second evaluation we aimed to answer questions that came out of our previous session. We asked participants specifically about their preferences for narration versus interaction and also provided a handout with a story map and information about Moe's rewards, to help us evaluate the value of a visual and whether it would help people understand the points system better.
Session 2 Takeaways (n=5)
The length of narration was a matter of preference where some really enjoyed the storytelling more than others
Participants still had difficulty understanding the points system
Gameplay should be more flexible, less linear
People enjoyed the visual component, though some found it more useful than others
This project was a challenge in exploring the possibilities of a novel technology within a novel context. At first it was hard to think outside the most obvious solution, but by listening to what users really liked (and didn't like) we were able to take a more creative approach that was very well received by our industry partners at Focus Brands.
Future iterations should explore how to accommodate users who prefer more interaction over storytelling, without sacrificing elements of the narrative that others really enjoyed. It will also need to test out how to best allocate the task of explaining the points system to users. Perhaps the system could provide a thorough explanation on the first play, with the option of repeating it on subsequent plays or viewing instructions on the Moe's website.