work research play about resume
Image placeholder
overview challenge solution approach metrics reflection links


A conversational user interface that improves the in-store experience for both shoppers and floor associates.

Conversation Designer, Researcher

3 Weeks, Fall 2018

iMovie, Adobe After Effects, Twine, Other CUI Modeling Software


project ask

Solve a problem in the retail space using a conversational agent.

project overview

The retail domain spans significantly varied levels of customer service, from high end retailers to fast-fashions stores.. Our goal as a team was to better understand how we could bring customers back into stores and make the shopping experience more enjoyable!

my role

I worked as a conversational designer, creating flows for how the logic of the CUI would work. I also did extensive market research on which retail environment would be best for a CUI.


This product was not shipped or market tested, but it made me very interested in Conversational Design. I'm currently learning new technologies like Voiceflow and Dialogflow for my Master's Capstone project!


Encourage in-store shopping.

The convenience of online shopping has stagnated the number of in-store shoppers. Ecommerce is growing 3x faster than any other retail sector and 52% of consumers prefer online shopping. We wanted to better understand how we can encourage shoppers to visit stores and challenge our assumptions that in-store shopping is dead.


concept video

Miranda is a conversational user interface that reinvigorates the joy of in-store shopping by building an experience that elevates department stores.

01 personalizing the shopping experience

02 making inventory management + customer service easier

model visuals

The CUI serves as the "front stage" while the stylist manages the "back stage".

Image placeholder

Image placeholder

Image placeholder


ethnographic research

Our team split up and spoke with customers at malls and customers in the process of making online shopping decisions to understand where in-store shopping is lacking.

Image placeholder

current retail cui(s)

Many retailers have already begun to employ CUIs to solve the types of problems exposed in our ethnographic research. We took a look at competitors as well as the overall size of the retail CUI market to consider the potential success of our product.

Image placeholder

Alleviating shopping pain points through our CUI.

Image placeholder

initial design

feature mapping
We mapped pain points to potential features our CUI could have, categorizing them into the four main functions

Image placeholder

scenario flow

There are three main domains in retail shopping: fast fashion stores (H & M), department stores (Macy's), and luxury retailers (Versace). We created a scenario flow and sample scenarios for each domain. See the process book for written scenarios!

Image placeholder


We used the storyboards to user test our concept and decide which domain to hone in on.

Image placeholder

Initial Sketches

Image placeholder

Final Storyboards

experience prototyping

We used experience prototyping to identify potential gaps in our initial brainstorm of features and to think about the technical implementation of our CUI - such as modeling feedforward, pre-attentive prompts, utterances, responses, error recovery and session termination.

Image placeholder

initial models

Our first model was created using Twine.

Image placeholder

Image placeholder

iteration + evaluation

My team was inspired by The Trunk Club - a personalized mid- to high-end clothing service that pioneered the at-home try-on model within the men's clothing industry. We were posed with an interesting question during critiques -

Could we create an in-store experience that mimics the personalized recommendations aspect of the Trunk Club?

We renamed "sales associates" to "stylists" and revamped their roles: stylists collect the items that users tell their CUI they would like to try one, and add items to their fitting room that complement the users' style, based on selected items and purchase history. In this way "we flipped the store", keeping stylists behind the scenes and having the CUI be the major player on the front stage.

Image placeholder

pivoting from fast fashion to department stores

Our initial hypothesis was that fast fashion stores would benefit the most from a conversational user interface since they tend to have the most difficult customer experience due to excessive inventory and lack of organization. Department stores have a high touch service model, and a need for innovation, so we decided to switch our focus from fast fashion to department stores. Luxury stores were quickly eliminated due to their human interaction, high-touch model.

final solution

Service Blueprint

Image placeholder

Modeling Sessions & Dialogues

Image placeholder

Video Scripting + Filming

Image placeholder


This product is not going to be implemented. However, I looked at the success of similar conversational user interfaces to determine important metrics for a retail CUI:

01 Increasing the number of stores that integrate the CUI
02 Increasing the number of referrals/customer downloads
03 Conversion of online shoppers to in-person shoppers
04 Customer satisfcation surveys/increased engagement in stores - with floor associates


01 Building Dynamic Paths

In the context of retail, there are a lot of different questions that can be asked: from getting your size in a pair of pants, to getting helping finding a particular section in a store, to converting your order to an online one. Due to this, it was important to consider all the different turns one could take.

02 Connecting to External Sources

There is a lot of data storage that occurs in this CUI, since it has to remember details about particular stores and particular customers. Additionally, connecting to an online store and creating synchrony between the digital and physical was something I had a lot of difficulty with.

03 Growing an Increasing Interest in Conversational Design

The biggest takeaway from this project for me was becoming interested in non-traditional interaction design. I really liked thinking about how to design for something that may not necessarily have a screen. I also enjoyed all the logic associations in conversation design - it allowed me to really express my technical side.