emotions in motion

Human Centered design and engineering M.S. at the university of washington | Adidas

Emotions in motion is a tool for adidas shoe designers that enables them to see the relationships between material properties and human emotional responses.

Problem

Adidas designers and engineers rely on intuition to select materials for new shoe designs. To help increase customer satisfaction, shoe designers and material engineers want to use an evidence-based framework for material selection.

Role

I served as the research lead on a team of four. While I spearheaded the research portions of this project, I was also able to be heavily involved in the design and engineering portions. 

Part 1: research

background research, research study & analysis

The first half of our project was focused on research. In order to build a tool for Adidas designers, we had to develop a recommendation algorithm based on how consumer emotionally respond to various materials, so we conducted a material research study to gather the necessary data to serve as the basis for these recommendations. Prior to this study, we conducted background research to inform the study design.

background research

A literature review was completed to better understand the way emotive frameworks are currently being used across all domains—not limited to just the shoe and apparel industry. The literature review served as a deep dive into the existing knowledge of emotive responses to materials and their properties. We looked at the ways material properties and form factors have historically influenced shoe design as well as the current process in hopes to reveal any challenges with the system today. Specifically, a study conducted by MIT researchers explained the need for a deeper understanding of the emotional response to materials. We also found resources through a toolkit created by the Materials Experience Lab that provided guidance for our study design. This toolkit was an excellent starting point for our research.

A comparative analysis is an opportunity to learn if and how other companies in industry as well as across other domains are using emotive frameworks to help guide design decisions. This analysis looked at the differences in footwear material between Adidas and competitors, specifically in the athletic apparel industry. The broader goal of the comparative analysis was to better understand the industry and to learn how Adidas designs fit into the larger landscape. From the comparative analysis we learned that while many other companies, such as Nike, have developed robust material classification systems, their systems do not take into account emotional responses. This was further motivation for our project, as it allows Adidas to get ahead of the competition. 

Trend Hunter is a leading research and data firm that tracks consumer trends across different industries. This analysis summarized a snapshot which highlights individual shoes that were trending in the market at the time of our study. While the individual shoes were not interesting in themselves, trending materials and build characteristics along with the associated feelings and emotions they invoked were noteworthy.

Physical Material Study

The study consisted of both an online survey and an in-person two-part physical material study. As part of the study, we used a set of materials supplied by Adidas used by their designers and represented a variety of physical properties. We also generated a list of emotional responses to serve as a word bank compiled from both the MaE24 toolkit word bank and a list of emotional responses generated by Adidas. 

Based on discussions with our sponsors at Adidas, we decided to design the online survey and material study with the typical consumer purchasing flow in mind. That is, a customer might view a shoe online first, then go to the store to touch it and see it in person, and then either purchase or try it on. For our physical material study design, we decided to emulate this by gathering emotive perceptions from our participants via an online survey first, and then having them complete the in-person material study.

Mechanical Turk Survey

We also conducted a mass online survey leveraging the Amazon Mechanical Turk platform. The purpose of this survey was to understand how consumers respond to visually examining different materials. We leveraged the taxonomy we derived from the in-person material study, but the goal here was to better understand how the visual interpretation of materials alone impacted emotional responses. With online shopping, consumers’ initial impressions increasingly rely on how they first respond to seeing the product, so understanding similarities and differences in how consumers might visually perceive a product before trying it on is important to consider. The insights from this additional analysis were included in the algorithm that resulted from this research. 

Participant Online survey

Prior to the in-person study, our participants each completed an online survey. In the survey, participants were shown an image of each material and were asked to imagine they were purchasing activewear online and came across an item that contained the material. 

They were then asked two questions about each material: 

1. In one or two words, how does this material make you feel? 

2. Please select up to three words from this list that you most associate with this material.

in-person material study

Participants who completed the survey were scheduled for a two hour in-person session where they completed a two part study. 

Part One 

In part one, participants were shown one material at a time and given time to play around with each. For each material, when the participant was ready, they were asked the following two questions one at a time: 

1. Please describe how this material physically feels. 

2. Please describe how this material makes you feel. 

Researchers allowed the participant to spend as much time and voice as much feedback about a material as they wanted to. Next, the participants were shown a word bank, and were asked to select up to three words that they most associated with the material. This process was repeated for each of the 39 materials. 

Part Two 

In part two of the study, the materials were laid out in a grid in front of the participants. Researchers went through the word bank from part one of the study and asked participants to select the one material they most associated with each word. 

The goal of this exercise was to determine what words may have changed from viewing the material online to feeling it in person.

study findings and outcomes

The results of this research were two-fold. First, there were several interesting findings, some of which influenced the way we built and designed the tool, and some that were passed along to Adidas to consider when integrating this tool into their existing tools and processes. Secondly, the research results were used to create the algorithm that the tool surfaced the results of. 

Certain material attributes had strong correlations to specific emotive responses 
7/7 participants (100%) 

Some correlations between material properties and emotional reactions were stronger than others, with the strongest correlations as: 

Soft-Relaxed (24) 
Stretchy-Curious (23) 
Stretchy-Comfortable (22) 
Stretchy-Energetic (19) 
Squishy-Comfortable (18) 

Emotive responses differed from visual only to tangibly touching materials 
7/7 participants (100%) 

We found that participants oftentimes had different feelings about the materials when physically interacting with them than they did in the initial visual only online study. The most common reason cited for this discrepancy was that the picture did not accurately depict the material. 

Participants emotions were impacted by their personal histories 
3/7 participants (43%) 

Many responses to materials seemed to be linked to the past experiences that a participant related to a given material. Participants often told stories about a time when they interacted with a material similar to the ones they were being shown.

the recommendation algorithm

The raw data was first sanitized by removing all pieces of feedback that were not strictly descriptor or emotional responses. We also removed all conversational and anecdotal responses from the data. This left us with a list of physical attributes and emotional responses, perceptions, and feelings. 

The second step in the data analysis was to normalize, or code, the data. The goal of this step was to create a single list of physical attributes and emotional responses that was consistent across all participants. The coded data was then analyzed to determine relationships between material attributes and emotional responses. 

In order to compare and contrast different materials, emotions, and physical attributes, we needed to normalize our data. With these normalized scores, we were able to produce visualizations that could be used to better understand the relationships between materials and their physical attributes and emotional responses.

While the visualizations in themselves provide interesting insights into these relationships in isolation, they do not capture the complexities of combining multiple materials together, so we created a basic recommendation algorithm to help with this. We calculated a weighted average score based on the inputs characteristics being designed for, taking into account the primary and secondary attractors and detractors as well as the global attractors and detractors. 

Part 2: Design & Usability study

The second part of this project consisted of designing and testing a tool that would surface the data from the research in part 1. The goal of this tool was to communicate the findings and output from the consumer research in a way that was usable for Adidas designers. To create the tool, we started with the algorithm that generates correlated emotive perceptions or material attributes based on the design inputs. Using this model, we used paper prototyping techniques to generate ideas for the initial wireframe. Then we created a low-fidelity prototype that we tested with Adidas designers and engineers. Through two rounds of usability testing and iteration, we created a final, high fidelity prototype. 

Ideation & Paper Prototyping

Initial Wireframes

After we consolidated ideas and ran through a hero scenario, we found that we had two different versions of the tool that we wanted to test. The paper prototypes were recreated into wireframes in Figma with placeholder images and assets so that we could test them virtually with Adidas designers. 

Usability Testing round 1

For this round of usability testing, we tested Version A and Version B of our initial wireframe prototypes. We ran moderated studies with Adidas designers and engineers remotely over Microsoft Teams. There were six participants for this round. Due to the condensed nature of the capstone project, we decided to approach this initial usability study as a task-based usability study with a user research-oriented interview. The goal of this round of testing was to get initial feedback on the tool concept and the content being delivered to the users. 

Top Findings

• The distinction between emotion and material property differs from consumer to user. 

• Participants wanted the ability to adjust the weights of their inputs. 

• Participants struggled to add supplemental inputs. 

• Participants were unsure how to get started with the tool. 

• The graph is not usable at this fidelity. 

• Not all of the information on the material detail card was useful or scalable. 

  
  The results from this round of testing led us to move forward with one version of the tested prototypes, and resulted in a list of recommendations based off the findings for incorporation into the second iteration of the tool. 

Iteration

After the first round of testing, we discussed the findings and recommendations and determined how the changes would be realized in the second iteration of the prototype. We used sketching methods to play around with different ideas. Once decided on, the features and changes were made in the Figma prototype. For this iteration, the prototype fidelity was increased in order to start gathering feedback on visuals and other UI. The fidelity of the graphs was a specific focus of this iteration, so that we could get more accurate feedback about their usefulness. We also added more color and focused on updating the typography in this iteration. 

Usability Testing round 2

The next round of testing was on a more refined, higher fidelity, prototype. For this round of usability testing, we ran unmoderated studies with Adidas designers and engineers remotely using Validately, an online usability testing tool. Five of the six designers and engineers who participated in the first round of testing participated in this second round. This was a task-based usability study and the goal was to find any remaining usability issues and validate changes made after the first round of testing. The findings from this round of testing led to tweaks and changes in the final prototype.

Top Findings

• Related attributes section of cards was confusing for participants.

Participants want the ability to export or save their queries.

• Participants associated a hierarchy with the left to right column presentation

Final prototype

After the second round of testing, we took the more focused set of findings and recommendations and made some final tweaks to the prototype. Changes included making content on the material and material attribute cards easier to understand, updating the layout of recommendations to reinforce the order of strength of the recommendation, and a number of smaller quality-of-life improvements to address use cases brought up by participants. With the user experience finalized, we turned our focus to visual design explorations.

Visual Design explorations

To determine the visual direction we wanted to take with this prototype, we took some time to explore different visual treatments. We started by creating a mood board containing different interface options that we thought would work well in our tool. After discussing this mood board together, we added some of these visual elements to our prototypes. We ended up with a few different versions and came together to discuss the elements from each we wanted to include in the final prototype. These visual updates were incorporated into the final prototype to create the high fidelity tool we handed off to Adidas.

With the final user experience defined and our desired visual design direction understood and applied to our prototype, we present a select set of screens from our final high-fidelity prototype, EMMA, or the Emotive Modeler Material Attributes tool.