Amazon Housekeeper
Role(s): User research, UI design
Design Space
The goal of this project was to understand how consumers shop for household items within the general contexts of physical and digital stores. To gather information, we implemented a survey to answer five main questions:
Who buys household items?
What factors affect their purchasing decisions?
When do they purchase household items?
Why do they use specific channels over others?
How do they purchase household items?
Insights from a survey of 101 participants showed that consumers who buy household supplies digitally do so via Amazon’s desktop website vs. the Amazon app. These findings helped narrow down the first problem statement:
Why are consumers not using the Amazon app as often as they use the website to purchase household supplies?
Interestingly enough, no evidence was found from the open-ended questions in our survey of these consumers considering the environment when purchasing these products, thus leading to the development of our second problem statement:
Are consumers placing any consideration on the environment when purchasing eco-friendly products? Does Amazon provide features that support consumers with this?
User Needs: Investigation Methods and Findings
In order to gather well-rounded feedback to inform our user needs and validate the problem space, 4 different research methods were used: Semi-structured interviews, Focus groups, a Task Analysis, and a Cognitive Walkthrough. These methods answered the following questions:
What tasks do users complete to achieve their intended goals on the app?
How well does the app allow the user to complete those tasks?
What pain points do users have when using Amazon’s app to purchase household products?
What are user’s behaviors while shopping via the app?
Design Requirements
Using the data acquired from performing the four research methods, several design requirements were uncovered. Personas created by analyzing the research data allowed for the evaluation of the design implications through targeted user flows.
Ideation
Using the design criteria and personas, several different concepts were sketched for a total of over 20 designs. The top four concepts were chosen based on creativity and relevance to the design criteria and user needs. Low-fidelity wireframe prototypes were tested with four participants whose criteria fit the intended user audience. For the low-fidelity wireframe feedback sessions, there were four participants from the identified user group who reviewed 3 concepts: Landing page, Product page, and Housekeeper. The Search and Compare wireframes were excluded in the session because according to previous feedback it had no value-add to improve the purchase process.
Prototypes
The prototypes were created in Balsamiq (medium-fidelity) and Adobe XD (high-fidelity) by incorporating feedback from all of our sessions. The Amazon Housekeeper feature would potentially live within the current Amazon mobile application. Using the medium-fidelity prototype, heuristic evaluation sessions were conducted with four usability experts.
Medium-fidelity
High-fidelity
Evaluation Goals
The main goal of heuristic evaluations is to identify any problems associated with the design of user interfaces. These sessions helped identify the main usability issues within the design. The experts examined the design with the awareness that the interactive prototype is not fully-functional and may have certain imperfections that are not directly relevant to the usability. The goal of this part of the usability testing was to establish and validate user performance measures during the purchasing process, and identify potential design concerns to be addressed in order to improve efficiency, productivity, and end-user satisfaction.
The specific objectives were to:
Determine design inconsistencies and usability problem areas within the user interface and content areas.
Establish baseline user performance and user-satisfaction:
The satisfaction of the process of selecting and browsing for household items;
The ease of use and understanding of the housekeeper feature;
Usability Testing
Using the think aloud usability testing method, allowed for the discovery of key pain points and thoughts throughout the entire purchasing process. This gave the ability to listen to the cognitive thought processing of participants and discover the root causes of their issues. One limitation of this method is that when a moderator is prompting or clarifying the tasks, this interference may have an impact on users’ behavior. In such cases, the resulting behavior may not represent the participant’s intended action since the assistance may have biased the user.
The usability test began by giving participants a scenario about purchasing paper towels then introduced the tasks one by one. During each task, notes were taken on points such as emotion, confidence, and actionable design recommendations, as well as behavioral observations. After the completion of tasks, quantitative data was collected from participants to understand the ease of use of our application for each task. Lastly, we reserved a few questions at the end to gather users’ thoughts on future improvement.
Lessons Learned
The biggest takeaway from this project was learning how to filter and prioritize user feedback based on repeating suggestions versus iterating with everything that users say. Also highlighted was the importance of organization. Aggregating/coding detailed notes into an organized spreadsheet also turned out to be extremely efficient.
Thanks to the efficient structure of the heuristic evaluation, most experts were able to complete the evaluation in under 30 minutes. Following this, they mentioned that they were very satisfied with the process and that the tasks were easy to follow.
The usability testing was also very well received and probing further questions led to very specific insights rather than general comments, making sure to give the participant appropriate instructions on the task they needed to finish in order to not bias their actions.
Lastly, as a designer, I realized that the fidelity of prototypes establishes participant’s expectations about the functionalities. It is very hard for them to merely analyze and evaluate the design without these expectations. This was especially important due to the real-world connection to the concept and Amazon’s application.