Designing A Voice Enabled Smart Mirror for a Personalized Shopping Experience
project overview
IRIS a Voice-Enabled Smart Mirror aims to solve this problem by offering virtual personal styling assistance. The voice bot mainly assists the user with shopping tips, and customized outfit recommendations and offers users a virtual wardrobe to save outfits and clothing items.
​
Team: Anamika Menon, Vanshika Shah
my role
User Research, Product Research,
Bot Personality Design, Scripting Conversation,
Designing GUI Elements,
Prototyping & Testing
tools
Lucidchart,Voiceflow, Photoshop,
Illustrator, After Effects
THE PROBLEM
Over 70% of individuals struggle with the overwhelming task of daily outfit styling
resorting to seeking recommendations from others or browsing the internet for inspiration. The constant evolution of fashion trends further complicates this challenge, calling for an AI bot that can intelligently incorporate ever-changing trends into personalized styling suggestions
THE DESIGN PROCESS
3 weeks
5 weeks
DISCOVER
DEFINE
DESIGN
PROTOTYPING
-
User Research
-
Level of Personification
-
Interaction Goals
-
Level of Personification
-
Key Features
-
Tone of Voice
-
Sample Scripting
-
Intents and Prompts
-
User Flows
-
UI Design
-
Bot Prototyping
-
User Testing
THE SOLUTION
Design a Voice Enabled Smart Mirror aims to solve this problem by offering virtual personal
styling assistance . The voicebot mainly assists the user with shopping tips, customised
outfit recommendations offers users a a virtual wardrobe to save outfits and clothing items
for future reference. The smart mirror would offer features such as:
SHOP
Assist users purchase apparel on numerous clothing platforms for a specific event or by budget.
OUTFIT RECS
Assist the user by recommending styling options on selected apparel or accessory.
VIRTUAL WARDROBE
Recommend outfits from existing wardrobe and online clothing items for various events of day.
A demo of our user interacting with IRIS!
USER RESEARCH
How do we empathize with the wants and needs of our users?
A user survey was conducted to better understand how people perceive voice enabled smart mirrors and get insight into the potential concerns and hesitations that might limit the interaction.
24
who were amongst a mix of fashion students, influencers, regular shoppers.
Participants
09
revolved around the users shopping
behaviors and preferences, their approach towards shopping assistance, their existing knowledge of smart mirrors and their expectations from such a technology.
Questions
WHAT DID WE LEARN ABOUT OUR USERS?
How do our potential users prefer to shop?
25%
Users seek outfit styling/recommendation
on a daily basis
​
42%
Engage in online shopping one/two times a month.
Why would a voice enabled smart mirror be a helpful tool?
65%
Participants said they found it moderately difficult to put together outfits on a daily basis.
Participants said they found it moderately difficult to put together outfits on a daily basis.
80%
Participants concluded that they were fully or moderately interested in using a product like IRIS
According to the users, what are the key features for a voice enabled smart mirror?
"Suggestions from stores with prices"
​
"Recommendations for colors that work well together"
​
"Suggestion with fits that look good on my body type"
​
"Converse back and forth with me similar to an Alexa"
​
COMPETITIVE RESEARCH
What brands offer styling/ shopping assistance? What is their mode of interaction and how do they work?
We conducted a competitive analysis of three brands that offered similar products and services - Alexa, iMirror and ShopExp that aim to improve user's shopping experience. Our goal for this analysis was to understand their success and failure stories while we designed our bot. It helped us identity must-have features to make our user's experience seamless and intuitive.
GIVING OUR BOT A PERSONALITY!
A high level of personification. What does that mean?
​
Our users need to be able to visualize the bot as a personal stylist and rely on it for their day to day fashion recommendations. The bot should be able to keep track of user purchases and form an online directory to extract information from for the daily recommendations. The bot's interaction goals needed to blend well personality traits that were Vivacious, Resourceful, Tasteful, Adaptable, Organized and Mindful for it be approachable reliable and friendly to its users.
TONE OF VOICE
The user's interaction with the bot should be casual, like having a conversation with your stylist where you feel heard. The bot's recommendations should also feel attainable yet solving the needs of the user and therefore expert. Through research,we also wanted the bot to sound excited, approachable and always ready to interact with it's cool and professional demeanor.
​
SAMPLE SCRIPTING THE CONVERSATION
We carefully wrote scripts of all the different ways a user could interact with IRIS based on the bot's interaction goals and personality. The primary task flows we developed were:
​
1. Outfit Recommendation
2. Outfit Recreation and Purchase
3. Inventory Search and Virtual Wardrobe
CONVERSATIONAL UX FEATURES
INTENTS AND SLOTS
​
For our user to have a seamless conversation with IRIS, an extensive set of intents and prompt slots were designed. They were also altered and numerous updates were made based on the feedback we got from the user testing the bot.
​
FLOW DIAGRAM AND PROMPTS
​
A complete flow diagram was designed for each of the three major tasks performed by IRIS was created to rectify limitations in the training data and ensure a seamless interaction for the users.
​
This helped us identify the possible conversational routes and error or dead end scenarios.
​
​
​
DESIGNING THE GUI
IRIS provides users with a visual output to help users engage in a responsive interaction with the mirror. The visuals will also help users remember where they stand in the conversational flow thus making their interaction as seamless as possible. The visual output will act as a reinforcement for the audio output helping users gather the required information easily and reducing the overall cognitive load.
I contributed by designing the following visual UI elements to represent visual output from the mirror.
HEY IRIS!
Beginning of interaction
Initial Prompt Options
Users will be shown three options to start begin with their flow:
​
Outfit Recommendation Option
User will be asked if they require an outfit recommendation using new unowned clothing or with an item they already own in their virtual wardrobe.
Accessing the virtual wardrobe
Users will be shown all the items saved in their virtual wardrobe which they can access by saying the item name.
Styling the item from the virtual wardrobe
Users will be able to select an item with the option of styling, shopping similar items or deleting the item from
the virtual wardrobe.
Curating an outfit based on user style
IRIS will present several outfit options with the selected item for users to pick from using “Option #”as a response.
Securely placing an order
Users will be able to review their order with the product names, brand and prices. Users can respond with a YES
to make the purchase.
THANKS FOR SHOPPING!
End of interaction
PROTOTYPING
A higher fidelity interactive prototype built on Voiceflow was tested with 2 participants tounderstand how the training data influenced the nature of their interaction with the voicebot and to measure the success of the interaction. Tap below to speak with IRIS.
and watch our demo interaction with IRIS on the left. The video showcases our user getting outfit recommendations from items in their virtual wardrobe with the help of IRIS.
TESTING OUR PROTOTYPE
How were our users interacting with IRIS?
​
We conducted a moderated in-person user testing with IRIS, where our participants represented different segments of our target audience. To give our participants the full experience of interacting with a smart mirror, the test was set up using a real mirror with an Alexa device placed behind it that played our VoiceFlow prototype, as a way to make our users perceive as if they were talking with the mirror.
​
The participants were then asked to give us feedback on the interaction which we used to further analyze and alter our task flows and the prototype. View our full testing summary (pg.19-21) report here.
​
FUTURE SCOPE
USE IN RETAIL STORES
In a broader use case, IRIS can function as a virtual shopping and trial assistant in shopping malls and stores. IRIS can be integrated into store mirrors, providing the ability to virtually try on outfits in stock, creating and recreating looks on site, as well as purchasing clothing items from the store to enhance our user’s shopping experience.
ACCESSIBILITY AND INCLUSIVITY
To ensure that this product is accessible to users who might have speaking and hearing limitations. This is product can be expanded to allow for touch and visual input to aid functionality such as the ability to virtually try on outfits in stock, creating and recreating looks on site and online shopping.