Card Sort: Product Taxonomy

Background

Cardinal Health aims to design and build a single unified pharmaceutical ecommerce experience to replace its 8 current storefronts spanning desktop, mobile, and tablet experiences. With the completion of the project, Cardinal will retire the 8 current B2B storefronts.


Problem

In the current state, users rely on search because browsing for products by category in Cardinal platforms is limited. Consequently, it is difficult for users to find products if they don’t know exactly what they are looking for, or how products are listed in the system.

Moreover, a competitor analysis of B2C pharmaceutical sites revealed that the way that over-the-counter products are classified varied considerably.

The research goal of this study was to understand how current customers mentally group over-the-counter and prescription drugs. This data would be used to create robust product categories and subcategories that are intuitive to the user, inform design work, make it easier for customers to find the products that they need, reduce bounce rates, and increase sales.


Research Approach

Over-the-Counter (OTC) Card Sort
Moderated sessions
Group of 25 OTC medications, derived from a list of the 100 most-ordered OTCs
Mixed card sort, wherein participants were asked to group like products into applicable categories, or create their own category
Second round with ability to subcategorize products and create labels
15 participants (joint sessions with RX Card Sort), based on Nielsen Norman’s recommendation to test 15 users to reach a correlation of .90

Prescription (RX) Card Sort
Moderated sessions
Group of 25 RX medications, derived from a list of the 100 most-ordered RXs
Mixed card sort, wherein participants were asked to group like products into applicable categories, or create their own category
Second round with ability to subcategorize products and create labels
15 participants (joint sessions with OTC Card Sort), based on Nielsen Norman’s recommendation to test 15 users to reach a correlation of .90


To analyze the research:
I created rainbow matrices, a way of color-coding results in Excel to visualize patterns in the data. I calculated the percentage of participants that grouped particular products in specific categories. This data was used to determine which categories were intuitive or unintuitive and show how closely customer groupings matched our initial groupings. Moreover, it highlighted areas where users’ mental models did not correspond with our categorization.

I used dendrograms, a tree diagram showing taxonomic relationships, to show which products users mentally group together to make more intuitive taxonomy recommendations.

I coded qualitative data (new user-created labels, subcategory labels, user comments, and answers to probing questions) to better understand the mental models underlying participant classification choices.


Findings and Solutions

Results:
For OTC, there was significant overlap in products assigned to Health & Medicine and Home Health categories.This led to the recommendation of For example, 50% of participants placed Aspercreme under Health & Medicine, but 35% placed under Home Health.
Qualitatively, participants noted being unsure of the differences between these categories

Recommendation: combine Health & Medicine and Home Health categories and create robust subcategories based on product purpose of use

For RX, many participants questioned the need to browse for products. When they order RX products, they know the exact name of the product that they need to order, because it is based on patient prescriptions. It is easier to type the name into search, then to browse for the product. On the other hand, they saw value in browsing for OTC products because they are not always sure how Cardinal has that item listed, leading to guesswork when using search.

Recommendation: Focus resources on creating a robust search experience
Recommendation: Prioritize OTC categories over RX categories

Mental Models:
The test indicated that our customers’ mental model when searching for OTC and RX products does not align with our current website taxonomy. Based on our findings, I made 29 product recommendations that would better align the website’s overall structure and subcategories with customer expectations.

Top recommendation: Focus resources on creating a robust search experience for RX products
Recommendation: Prioritize OTC categories over RX categories

Participants were guided by:
The therapeutic purpose of drugs, e.g. placing products used for diabetes together
The bodily system that drugs treated, e.g. placing products for eye care together
The form of drugs, e.g. subclassifying oral versus topical pain reliever together
Participants expected some products to be in multiple categories, reflecting varied purposes of use. For example, alcohol swabs are needed for testing blood sugar for patients with diabetes, but they are also used to treat minor wounds


Outcomes

Based on my research findings that users first look at the therapeutic purpose-of-use for products, the program adopted the Uniform System of Classification (USC), dividing RX products by therapeutic category.

For OTC products, the program focused on adopting subcategories based on therapeutic use, e.g. Digestive Health and Ear Care.

These product categories were released in December 2024, with the unveiling of the MVP version of Vantus HQ to a limited audience.