Hybrid Card Sorting for Information Architecture

 

Step 1: Methodology

I considered one of the following methods to categorize the items:

Method 1: Open and Close Card Sorting	Method: Use Open or Close Card Sorting only
The combination of both method could help to understand the mental model of users.	In case using this either Open or close card sorting, we would also combine with the Tree Tests to test if the structure is good and ready for users.

 

Step 2: Methodology choice : Hybrid Close Card Sorting

                                                  Participants: 20 users

 

Explanation of methodology choice:

In the beginning, I intended to use method 1, however, I faced the following challenges:

• I might have to double the time and participants, while we would had time constraints and financial budget difficulty
• I created the open card sorting but the process is too slow, the result is low (screenshot above: 42% users completed the test), despite we facilitate users by adding photo to each item. Users might be tired and frustrated to see ALL 100 items then have to categorize them.
• 
• 

This lead me to the solution of recommending to users the possible categories and combined close card sorting with open card sorting. This is Hybrid card sorting.

Hybrid Close Card Sorting: I can suggest categories for users to put the items in. Users still have the option to create and name a new group if they feel it is necessary.

Step 3: Process of our Card Sorting:

I provided users a link to do the card sorting. I collected data to categorize items according to the users’ choice. The images were attached to items to facilitate users’ recognition and selection.

Brainstorming: The 100 items were imagined to be organized in an online store website. I based on our shopping experience to put items in relevant categories. The following category are introduced to users in closing card sorting:

Heath & Beauty, Kids & Toy, Music Instrument, Grocery, Electronics, Fashion, Bank, Office Supplies, Home & Garden, Accessories, Pets, Autopart and Accessories, Jewelry, Kitchen, Other Category.

Step 4: Collect, Analyze and Categorize the 100 items

41. Collecting and analyzing:

I collected the following data:

• Raw number
• Categories : Percentage of users choose a specific item for a category

 This percentage will be counted, for example, according to the number of users choosing 1 item for 1 category on the total of 20 participants.

Categories chosen by users for the item (By percentage of category chosen for a specific item)	Interpretation of the percentage	Action
Categories chosen by users (>40%) :	Almost clients would be able to find this item in this category	Items will be kept in the category chosen.
Categories chosen by users (30-35%)	One third of our users can find this item in this category
Categories chosen by users (20-25%)	Some users can find this item in this category
Categories chosen by users (<20%)	Very few users can find this item in this category.	Items will not appear in the category if the percentage is below 20%. We may create new categories because for users, the proposed categories are not appropriate/ relevant.
Items which are repeatedly chosen for different categories will be bordered in red.	Users are confused and this items can be either in both categories (for category chosen by users : category percentage >30%)	We may create new categories or consider to keep items in 1 relevant category only.
Items categorized as others or uncategorized	Users could not find any relevant category for the item.	We may create new categories.

I listed the items by categories (percentage) and differentiate them by colors. Afterwards, we make decision to keep or put the item in another category.

 

The items which need new categories (while the proposed categories are irrelevant for users) are :

Flag                                                                Newspaper                            Sidewalk

Streetlights                                                     Greeting cards                      Needle

Thread                                                            Truck

Items are considered to remove from the irrelevant category (by users) :

Packing peanuts ( was put by users in Grocery).

Items repeated at the same percentage of 25%-35% in 2 different categories:

Soda can ( was put in Kitchen) and in Grocery.

4.2 Solution

In order to make sure that users will be able to find the items in its category in a tree (menu) structure, I made the following correction to the categories:

• Remove items chosen but irrelevant to category (i.e: packing peanuts in Grocery)
• Choose more relevant category if items are repeated in 2 different category (For example, Soda can was put in Kitchen and Grocery, we choose Grocery). In this case, users thought that kitchen is the place he/ she can find this item
• Accept to repeat the same items in 2 categories if the categories are both relevant and have same category percentage
• Create new categories for items with category percentage under 20% or in Other category
• Group the categories with items for shopping to under a parent category MALL/ STORE/ SUPER MARKET

I presented a new information structure.

The structure may need a tree test to ensure that users can use them to navigate and find exactly the item in the relevant category.

5. Conclusion:

By applying the hybrid closing card sorting, I understand better the mental model of users.

Here are some notes I took during the whole process:

• Some items are repeated in 2 different categories. The reasons can be:

 

• I confused users if users do not have a context to consider where to put item in.

I could do the task better if we have had a specific context for users for better understanding of the categories.

For example:

Imagine that you go for shopping, where would you find the items according to the following categories?

NOTE

Some items can appear at the same time in 2 different categories

( For example, magnet can appear in both Toys & Kids, and Auto parts & Accessories with almost same category percentage. Tissue box can appear in both Health & Beauty and Home & Garden with almost same category percentage. These categories are both relevant!)

Sometimes the more categories are proposed, the more users get confused. For example, if we do not propose Toy & Kids, may be Auto parts & Accessories will be the only category to be chosen for magnet.

After card sorting, if time and funds are available, tree test can help to test if the structure really works for users.