Carnegie Mellon University_ Masters of Human Computer Interaction
Course: Interaction Design Studio (Ixds)
Team: Yousef Kazerooni_Rob Feeney_Vanessa Kalu
Duration: 4 weeks
Skills: Interviewing_ Literature Review_ Competitive Analysis_ Storyboarding_ Table Reads_ Scenarios_ Personas_ CUI Character Development_ Script development_ Experience Prototyping.
What medium, if not conversation, is more attuned to everyday human experiences?
With the advances in machine learning and voice processing tools, conversational user interfaces (CUIs) more than ever have the potential to facilitate the interactions between users and computers on human terms. During a four-week project, we set out to push the boundaries of conversational interfaces.
I worked in a team of three to explore how CUIs can be used to transform the dressing room experience.
Through research we discovered that a dressing room is as much about the calculations involved in choosing clothes, as it is about emotional, social and trust issues.
By getting rid of the sales floor, and transforming the dressing room into a consultation room with a virtual assistant, we created a personalized experience that is respectful of people’s anxieties and that can engender trust.
I contributed as the project manager. This was a collaborative project, so I additionally created the research strategy and did extensive research. I also facilitated ideation sessions, and contributed to the final design.
CUIs are still in their infancy. This lack of standards means that part of the challenge for us was to create the processes that could best inform our design. Our process consisted of 3 phases:
Research - Phase 1
Our investigation focused on two main areas:
Dressing room Experience.
Strengths and weaknesses of CUIs.
Our literature review was driven by our two focus areas. It was a survey of
1. Findings about the dressing room experience:
Dressing rooms are the most valuable touchpoint in a store. They indicate the customer’s intent to buy. If serviced, customers buy 3x as more as people on the sales floor.
People enter the dressing room with many anxieties, such as whether or not there will be a sale’s associate, or whether a father can take her daughter into a dressing room.
Dressing rooms are a unique awkward social interaction.
2. Findings about conversational user interfaces:
- It is recommended to use mixed media, when considering conversational user interfaces. That allows combining the strengths of people's auditory and visual capacities.
- Voice interfaces can quickly lead to cognitive overload when used to list items, enumerate numbers, or present selection options.
- The strength of voice interfaces is in their ability to mimic a conversation. They can rely on conversational conventions to create a natural dialog. Most importantly, they help create trust, because they are non-judgemental.
With two guys on the team who could barely remember the last time we were out shopping, we wanted to gain a deeper understanding of the shoppers' feelings in a dressing room.We conducted 4 interviews:
2 males, and 2 females.
In their 20s and early 30s.
From three different countries.
Our method of choice was directed storytelling, where interviewers act more as facilitators guiding the story with probes. It allows researchers to get at people’s emotions and intention in their actions.
Social decision making is an important aspect of shopping for clothes.
Dressing rooms can lead to strong emotional experiences. One of the interviewees expressed her feelings as follows:
We focused on:
What interactive technologies are used in the Dressing Room space?
What Market Segments are served by a particular technology?
We looked at 10 different retail stores. we discovered that:
Most stores did not use any interactive technology in the dressing room.
Buttons were used as a low-tech solution in some stores.
Only a few high-end stores had installed high-tech mirrors in their dressing rooms.
Our research found dressing rooms to be the most crucial touch point in a retail store. Today most dressing room technologies explore logistical aspects of improving the experience. People’s unattended pain points encompass emotional, social, and trust issues.
Ideation - Phase 2
Using the insights we gathered from our research, we set out to create an experience that does away with people’s pain points in the dressing room. We went through 3 stages:
Using the insights from our secondary research, we were ready to explore the actual value of design solutions without worrying about the many intricacies of their implementation. Scenarios are stories about 4 sentences long. They are an exploratory method, and helped us generate new ideas that focused on the value added.
Here are 3 of the 16 scenarios we came up with:
Judy is going to Dubai for vacation. She is worried her current clothes would not be appropriate. The CUI is able to tell her about local customs and suggest appropriate clothes. Judy is relieved.
John is a new customer at Top Man. He normally does not shop at stores like this because he has a hard time finding clothes in his size. The CUI helps him quickly narrow down items that he might want that are also in his size. John is happy and plans to shop at Top Man again.
Amanda feels like crap. She has tried 15 different dresses but they do not fit. She is starting to hate the store. CUI detects the frustration in her voice and alerts a sales associate to attend to Amenda. The sales associate walks around the store with Amanda, till she finds something that satisfies her. Amanda is delighted.
Next, we developed a persona. There are lots of different kinds of people who use dressing rooms. So, to narrow the scope of the needs and interactions that our CUI would have to respond to, and to give our CUI a human-centered focus we created Roberto.
We crafted Roberto’s character based on the pain points from our research, and on the different needs and behaviors we explored through our scenarios.
Given the time constraint on the project, we decided to limit ourselves to one Persona, and try to create a CUI that is able to meet his needs.
Defining the environment
Next, we settled on a specific environment where the interaction would take place:
- Verbal Outfitters a high-end store with repeat customers. It would be Roberto’s third visit.
In her book, Designing Voice User Interfaces, Cathy Pearl, suggests defining the characteristics of the CUI before creating the actual interactions. We were interested in how our CUI can assist Roberto, given his low self-esteem and social anxiety. So, we designed our CUI to be helpful and polite, but also attentive to Roberto’s demeanor and responses.
As Verbal Outfitters is a store that receives repeat customers, it is important for the CUI to develop rapport with customers. As the relationship develops, there are more opportunities for the CUI to be more sassy.
Prototyping - Phase 3
Our CUI assisted Roberto by recommending items based on his and other customer’s previous purchases. In addition, Roberto had the option of asking for additional sizes and items once in the room.
These features were crucial to attending to the persona of Roberto. Someone who is body-conscious, may not know exactly what to purchase for himself and would not like to interact with a sales associate to request for more sizes or to get an opinion on an outfit.
Bodystorming the Script
The first step in creating the prototype was to develop a script. Pearl suggests keeping the script as conversational as possible, while maintaining the CUI character.
We created a mock-up dressing room, and bodystormed (i.e acting out multiple iterations of the intended interactions) conversations between a customer and the CUI.
This process of coming up with the script had two benefits
Actors can’t anticipate what the other person may say
Conversations are more real and natural sounding
It clarified at which points a visual user interface can enhance the interaction
- There is immediate feedback from the observers.
We then proceeded to create paper mockups of the visual interface. As we found in our research, listing items and providing selections are best done through a visual interface.
In a “Table Read” you read out loud the script to somebody else. It allows you to catch any awkward moments in the CUI and fine tune the tone of voice. For example, we found out that the CUI started every answer with “Ok”, which made it sound repetitive and unnatural.
User Testing and Experience Prototyping
We recreated a dressing room, and had other people interact with it. User feedback revealed that the CUI was very tactical and not really perceived as emotional. The CUI was acting as more of a support for the visual interface, rather than tapping into the potential of conversation as a mode of interaction.
Reframing the problem space
We wanted to elevate the value of the CUI by addressing issues that only conversation could solve.
That led to a light bulb moment:
What if we removed the sales floor from verbal outfitters?
What if the dressing room was the only touch point?
- What if people went into the dressing room for consultation as well as trying on clothes?
Here is the Verbal Outfitters:
Welcome to a store with no sales floor. Choose an empty room and begin your experience.
You’ll start by having a conversation with our CUI, the purpose of which is to interview you, show you things and get a feeling for the type of clothing you like. From there, instead of tediously browsing menu after menu, the CUI will simply make an outfit for you and display it on the screen.
After deciding the things you like and do not like about the outfits, the CUI will create a limited number of additional outfits. Simply click on the clothing items which appear to be worth trying on and send for a ‘runner’ who will bring all of the clothing to your door in a matter of minutes.
The script had to be revised to allow for a more natural flow of conversation, but at the same time making sure that it is a guided conversation, with predictable outcomes that reinforce trust and shopper satisfaction. Here is an excerpt of the final script:
This time instead of paper, we prototyped the visual interaction in form of a slide show. The slide show supported the 4 main visual components of the CUI:
Indicating CUI’s turn in conversation
Outfit creation and item selection
Completing a purchase
- End of conversation
The final presentation was a live demo. We built the dressing room using a projection screen as one wall and two tall wooden boards to enclose it. To avoid mishaps, our team acted out the interaction.
Our final design created an environment that elevated the value of the CUI. Our design created an environment, where Roberto could feel comfortable and not judged; it suggested outfits, not requiring Roberto to rely on his own sense of fashion; and it tried to build rapport.
Finally by eliminating the sales floor, it focused on the most important asset of a retail store, the dressing room, allowing companies to cut costs while increasing sales.