Card Sorting - a design and analysis technique

I do not know whether you know it, but ...
This is the first part of a series about design techniques and analysis techniques.

The following article belongs to an upcoming series – A series on developing approaches to problem solutions with systematic use of creative techniques. I want to comply with this article and series to wishes and requests of friends and colleagues.

: : : : : : : : : : : : : : : : : : : : : : : : : : : : : :

Card sorting is a categorization technique where users sort cards describing and giving their picture, their understanding and their mental picture of concepts, workflows and information and knowledge.



There are a number of benefits using card sorting - Card sorting is …
  • Quick and easy – It allows a lot of users to be involved
  • Simple – It’s a familiar and common behavior
  • Inexpensive – If we are honest and make a good job in general, the only added cost are the preparation efforts
  • Dependable method
  • Variable technique – to get to know the users

Whether you are designing a complex application, product or software it’s typical to be confronted by a very long list of potential content, functions and contextual topics to include and to arrange. The challenge is to structure these elements in a way that the product and system is useful, meaningful and easy to understand and to use.
Even complex applications can be easy to use … when the application fits the users’ mental picture / mental models. A good tool to see and to understand this “picture“ is card sorting.

It should be in mind, that a card sorting session doesn’t deliver a finished design, taxonomy or structure.

While this instrument might not supply you with a finished design or structure, it can help you solve many questions you will need to deal with throughout the whole project, particular during the first stages of the project. Card sorting creates an overall “idea” to structure your information, as well as an “idea” for navigation, content-appearance and possible taxonomy.

You can identify …
  • Labels
  • Cluster
  • Constellation

… and also their …
  • Relationships
  • Dependences

… in addition you can also record …
  • Significance
  • Value
  • Trends
  • Conflicts

What do the participants like to see?
  • Does the user want to see the content arranged by subject, tasks, process, and business? …
  • What is the individual picture and understanding?
  • Is it always and through out the whole application the same type of structure and arrangement?
  • Are the needs within the groups the same and corresponding? Does the participants discussed long and hard?
  • Are the needs of the different user groups comparable?
  • What is similar and what is different? … and how and why is it as it is.

METHODS _____ : : : : : : : : : : : : : : : : : : : : : : : :

There are three principal methods …
  1. Defined Categories – Users are requested to sort cards into defined Categories.
  2. Dartboard– The dartboard is organized in three, or if it’s needed in more, circles. The users are requested to arrange cards into the circles. The inner circle represents the most critical, important or needed aspects and elements. The second circle represents the medium important content and functions. And the third circle, the outer one, should contain elements that are only used less often, by a few people or in a few circumstances, but they aren’t as important as those in the middle circle.
  3. Open Table – The users are requested to arrange the cards into cluster that they consider are suitable and then they are asked to give the clusters descriptions or labelings*.

The second method can be enhanced by segments, like we all know by the dartboards. These segments are either an initial set of primary groups or just “open segments” without any labelling, in this case the participants are asked to give the segments a description/labelling. * By the way a description is often better than just a word, a defined labelling.




How to do it … _____ : : : : : : : : : : : : : : : : : : : : : :

FIRST _____ I I I I I I I I I I I I
Selecting participants
The primary step in any card sorting study is to collect a list of topic (content, functions and contextual topics) that will becomes your cards. From my point of experience, I’ve found that it’s difficult to do card sorting with more than one hundred subjects. This list should be drawn from a clear focused segment / task.
  • Group of responsibility
  • Business processes
  • Topic area
  • Existing topic and matter
  • Potential content
  • Planned applications
  • Considered functions
Print each topic (content, functions and contextual topics) on a single white card. By the way according to my experience, it’s difficult to do card-sorting studies with more than about 100 topics, but it’s possible, more than 200 is often absolute maximum, but even this depends on the over all requirements and business. About one hour, or maximum two hours, is the most you can usually expect someone or group to spend. Please note – always better one more rest than one less! Although. All this depends on the discrepancies of business, topics and complexity and last but not least according to the user and your participants.


SECOND _____ II II II II II II II II II II II II
Select your participants.
They need to be representative of the target users depending on your application as well as on task and aims.
Card sorting may be performed individually or in groups. I suggest you to do it with groups. A benefit of group sorts is that they typically provide richer data than individual sorts – If you intent to work in groups the participants have to be homogeny.

A card sorting session should have at last four participants. If you have more than eight the group will becomes more difficult to manage. Ideally, all the participants can reach the cards on the table. Usually one or two persons should take charge of placing and moving the cards.

It’s not necessary to tell our participants they’ll be doing a card sorting. As an alternative, let them know they don’t need to prepare ahead of time; they should simply come as they are. Sometimes I went to the users and sometimes I asked them to come to us.

Users are asked to sort the cards into logical groupings using no more than a two-level hierarchy.
Workflow for the card sorting session:
· Sort the related cards into groups. It’s often helpful to note the reason and relationship (why, how, …) why this card belongs to the group on the card. If there is one card with no relationship to other cards this card doesn’t have to be grouped. If a card belongs in more than one group, the participants should use a blank card in a different color** to duplicate the card. If there is any new content, the participants should take a cards in a different color** to add the new content that the participants would like to see.
· Once the participants are finished, we should accompany them through a particular task. This helps verify and confirm the results.
· Performing the card sorting session by method one (Defined Categories) or two (open table) - the participants should place each “group” of cards into a small envelope. Label each small envelope with a name that describes the information it contains. Performing the card sorting session by method two (Dartboard) – the participants should save and summaries your results on a flip chart or take a picture with a camera.
· All related groups of small envelopes and any related individual cards should be placed into larger envelopes to create “categories.” Label each large envelope with a name that describes the information it contains.
· Place everything into a folder labeled with the topic of the card sorting session. This folder can contain individual cards, small envelopes, and large envelopes. Please be aware that we used a two-level sort for this particular session. With fewer cards (e.g., less than 40 or so), it may only be necessary to use a one-level sort.


THIRD _____ III III III III III III III III III III III III
A small survey at the end of the card sorting session provides the opportunity to obtain further information about the participants, users and their needs. While it’s important to keep this section brief, it’s nonetheless possible to ask questions like:
· What is Job role
· What are your responsibilities
· Time of experience in this job role
· How do you find information / workflow and methods
· What do you need most – and what least – content? Function?
· What other information or function do you like to have.
· Do you have any other feedback or suggestions?


FOURTH _____ IV IV IV IV IV IV IV IV IV IV IV IV
The fourth step is to explore and study the results of your card sorting sessions and to develop a summary.
It’s often very enlightening to evaluate the results of card sorting sessions of diverse groups of participants. Especial if you discover conflicts like similarities in opposition to dissimilarities. These kinds of conflicts are very typical and it’s more likely to have these than to have not.
Again my analyzing steps depend on the amount of cards and complexity of the project. And well – I wouldn’t keep as a secret – it depends on time and budget of the project. Anyway the last step is always the same – I gather all the data on a spreadsheet. As a first step performing analysis I like to pin or tape the cards and their groups on a board or wall. Or you may be able to see patterns by simply laying them out on a table. But I suggest you to arrange them at a place where you can let them for hours, days or in best case for weeks. That’s the reason why I prefer the whiteboard or wall. To mark and or to arrange patterns and relationships you can use pencils, markers or also threads and tapes. Don’t forget to save your work frequently and regularly (Some time ago I had to face a blank wall and piled cards on my desk at a Monday morning in my office. And the reason was an over motivated cleaner).


EQUIPMENT _____ : : : : : : : : : : : : : : : : : : : : : :

** Colors for the cards – I prefer to use …
  • White … for the original card sets
  • Grey … for duplications of a original content
  • Yellow … for new content
Further more you need …
  • Small envelopes
  • Large envelopes
  • Folder
  • Pencils and Markers

SUMMARY _____ : : : : : : : : : : : : : : : : : : : : : : : :
Remember that the summary often doesn’t result directly into a final taxonomy and it’s also not necessary to jump straight to taxonomy at this point. Your card sort results can be supplemented with additional user research and task analysis.

Regardless of the method you use, card sorting is an easy, inexpensive and variable technique for learning how your potential users view the relations and interaction among the different subjects. The resulting design should be one that allows your users to find the information they’re looking for quickly and easily – but that should be a topic for a upcoming article.

Comments

  1. Hi Holger,

    thanks for this excellent summary of card sorting techniques! I thought I'd add a few thoughts on what happens after the card sort, which you will probably cover in an upcoming article.

    I think most information architects would be familiar in taking the card sorting results, analysing them in a spreadsheet like Microsoft Excel and then perhaps designing the new IA using a diagramming tool like Microsoft Visio. Once you are sufficiently confident in the new structure you might test it with real users on paper using a card-based classification evaluation technique or even a tool like Optimal Workshop's Treejack.

    I've done much of my IA work using the above tools but I felt that there's something missing between the research (e.g. card sorting) and validation (e.g. classification testing) phases. I don't think Visio and Excel really help you design and "get a feel" for the new structure.

    Which brings me to my shameless plug: We have launched a new tool called Naview to help information architects design and visualise a new navigational structure. It's focusing specifically on the design phase and as such it should sit neatly between the research and validation tasks. I'd be very interested in hearing your feedback on the tool: http://www.volkside.com/tools/naview/

    Thanks, Jussi


    References:
    http://www.boxesandarrows.com/view/card_based_classification_evaluation
    http://www.optimalworkshop.com/treejack.htm
    http://www.volkside.com/2009/04/using-treejack-for-ia-testing/
    http://www.volkside.com/2009/07/naview-rapid-ia-prototyping/

    ReplyDelete
  2. Hi Jussi,
    first thx for your comment. I'll check Naview as soon as possible.
    However, I can tell you already now that I'm agree with you that EXCEL and VISIO aren't a good tool to analyze anything - they are mostly just good to show results. That's why I prefer sketching, arranging, setting and building up mental image. pictures and the overall understanding of content, context and their relationships.

    Best regards,
    Holger

    ReplyDelete

Post a Comment