“Can I order that at a bar?”

The Approach

To convince major grocery chains and e-commerce stores to change the information architecture of their websites, the client needed a strong quantitative justification supporting any recommendations. To achieve this, we conducted a mixed-methods study that leveraged qualitative methods to uncover consumer mental models and quantitative methods to test and validate categorization hypotheses.

Over the course of 6.5 weeks, we conducted interviews and moderated card sorting sessions with online shoppers to understand their thoughts, feelings, and decision-making processes around purchasing RTD beverages. Simultaneously, we ran a series of unmoderated card sorting studies to uncover quantitative patterns around consumer categorization and evaluated our findings using a large-scale tree study. We utilized an iterative approach to our research which allowed us to probe into findings from each workstream and test new hypotheses as they emerged.

Qualitative Research

To better understand the market for RTD beverages and explore consumer behavior within this new beverage category, my colleague and I conducted user interviews with 10 consumers who had purchased alcohol online within the last year and who had, or were interested in, consuming RTD beverages. In addition, we conducted an external audit of eCommerce sites to understand existing alcohol taxonomies. My contribution to this work included:

• Literature Review: I led a review of internal documentation to understand the client's existing consumer knowledge and a literature review of alcohol industry consumer trends to understand the market growth and demand for this emerging category.

• External Audit: I led an information architecture audit of e-commerce alcohol taxonomies across 10 major grocery store websites and food delivery services to identify how the client's retail partners currently categorized RTD beverages.

• User Interviews & Moderated Card Sorting: In collaboration with my colleague, we planned and conducted user interviews and moderated card sorting sessions with online alcohol purchasers to understand their mental models around purchasing alcoholic beverages and to identify categorization patterns.

Quantitative Research

Any recommendations our team made needed a strong quantitative justification to convince the company’s retail partners to make changes to their websites. To accomplish this, we conducted a series of 8 unmoderated card sort studies to identify categorization patterns and test hypotheses. These were followed by a large-scale tree study to evaluate our taxonomy recommendations. As the lead researcher for the project's quantitative work stream, my contribution involved:

• Card Sorting: I identified research questions and key hypotheses to validate during card sort studies, developed the research plan and study design for each card sort, and analyzed data to identify emergent patterns.

• Surveys: I designed surveys to gather quantitative data and high-level qualitative data around consumer purchasing and consumption behavior that could be further explored during our in-depth user interviews.

• Tree Study: Once we had created a final information taxonomy based on our initial research findings, I designed and launched a tree study with over 500 participants to evaluate the effectiveness of our recommendations and test any outstanding hypotheses that weren't addressed during user interviews and unmoderated card sorting.

Synthesis & Insights

By combining data from our qualitative and quantitative research, we were able to uncover:

• Which types of RTD beverages users viewed as being related and the degree to which they were similarly grouped.

• A qualitative understanding of the specific characteristics that caused users to view beverages as being related.

• A deeper understanding of how real-world mental models around alcohol purchasing and consumption inform shopping behavior online.

Leveraging these insights, we developed a new e-commerce taxonomy structure that better reflected consumer mental models of RTD beverages. We made iterations to our final taxonomy recommendations based on the results of our evaluative tree study.