Why I don’t use Card Sorting
Card sorting is a simple technique in User Experience Design where a group of users are guided to arrange subject-headings under pre-determined categories or into groups which make sense to them.
For example, a card labelled “apples” might logically sit under a category labelled “fruit”.
It can be a useful approach for designing menu structures or website navigation paths.
But I haven’t used card-sorting since 1999.
Why? Because since the advent of the Content Management System (CMS) I haven’t needed to. The CMS changed the way I approached information architecture forever. I’ve never again needed to resort to card sorting in order to locate content in a place which is intuitive for ‘most users’.
In my view card sorting is a bit of a hack to resolve the issue of static websites where the content is forced to live in one place.
It’s like going to the supermarket and looking for custard powder. To me it makes sense to look for it within the cooking section next to sugar, baking powder, cornflour etc. But good luck finding it there because you’ll usually find it in the desserts section next to jelly and ice-cream cones.
Am I stupid? No, I just think different (and I know I’m not alone).
But why can’t the custard powder be in both places? And why can’t it also be next to the pre-made custard in the chiller? That would make sense too right?
Custard goes next to custard?
It’s because of physical constraints – supermarkets are dealing with physical products and usually can’t afford the space to locate food in more than one isle.
But, websites are different. We’re dealing with data, which is cheap, flexible and can transform itself into many formats and locations. This means, wait for it… the content can live in more than one place at once!
Why restrict it to live in a place the majority wants? What about the other 49%?
Call me a liberal hippy, I don’t care. I’m all for proportional representation.
PUT THE CONTENT IN THE PLACE IT MAKES SENSE TO [YOU] & [HER] & [HIM]
Doing this with a half-decent CMS is easy. It allows you to create multiple pathways to content by attaching categories/metadata/tags. The trick is to then leverage these tags and make them part of the navigation.
So all of a sudden a whole new world opens up… all roads lead to the content, in a way which makes sense to pretty much anyone.
An example I use when designing intranets is the ‘Leave Form’. Everyone uses it right?
But how might different people think where to find it? Where would they start?
- The Human Resources Manager might be thinking: “HR Form 305” or “Holidays Act”
- The Team Manager might be thinking: “Leave Application” or “Leave Request”
- The Staff Member might be thinking: “I want to go on leave” or “Leave form” or “Holiday”
If the content lived in one place under one label then chances are that some people would spend precious time fluffing around trying to the find the elusive Leave Form.
A few might not find it and email the busy HR Manager. Well, I say, make it available in ALL these places. Why not?
Hereb are a few examples of sites with multiple pathways to content…
Ministry of Transport
Want to find statistics on road crashes? You can get to the right report via a number of pathways…
- By audience type (e.g. “motorists”)
- By transport mode (“land”)
- By topic (e.g. “crashes”)
- By content type (“research”)
- By searching (e.g. “statistics”)
Epicurious
Need to cook vegan chocolate cake? You can get to the right recipe via a number of pathways…
- By ingredient (e.g. “chocolate”)
- By dietary consideration (“vegan”)
- By category (e.g. “desserts”)
- By searching (e.g. “vegan cakes”)
Sustainability
Want to save power? You can get to the right advice via a number of pathways…
- By goal (e.g. “I want to save money on my power bill”)
- By topic (“energy”)
- By searching (e.g. “save power”)
I’m not saying you should throw card-sorting out the window. It might prove useful in some situations.
But it’s time we all starting thinking of content, not as a static thing which needs to be locked into a certain box in some dark corner of your website, but as dynamic.
Set it free! Let it bubble-up to the right place at the right time for your users.
7 responses to “Card Sorting Doesn’t Cut the Custard”
I agree, card sorting can inform, but never enforce the one true way.
I’ve also noticed that this increased adoption of aggregation and multiple surfacing of content objects within the IA has had an impact on how some CMSs structure content. Gone are the days of all CMSs having site trees and pages – ie Expression Engine deals in objects which are surfaced within any number of templates in any number of ways. It doesn’t have an implicit or explicit tree structure – and it feels right.
Zef – I hear you, I really do. BUT – I love card sorting. I still believe you can get all this information (and more) through card sorting. How you run a card sort is the real trick though I think. Why not allow more than one card for the same item into your card sort? I like to leave it up to the users to guide me first as to whether tag based or folder driven can best meet their needs.
I use card sorting in a broader way than I used to. I give the user the right to create more cards for the same thing and stick them wherever they like (I get them to do it on coloured cards so I can pick up the duplicates later). Once I see how many duplicates there are on the table I can tell if we’re going to need to go tag based, folder driven or – my favourite – a mix of the two. Obviously there is a bit of psychology involved in this process to identify and deal with people who are out-of-control card makers (there’s often one), but overall it works well.
Using the multi-card option can be a great way to indicate to a business that has not previously had any user feedback on their structure, just how bad their basic nav might be. When a customer asks me to do a closed sort on their site as part of including a new section or rejigging one bit of the structure, it can allow me to demonstrate visually how difficult people founnd the first options. I find a photo of the carnage highly compelling to a business owner, it’s the smoking gun on top of the report with percentages and recommendations 🙂
In my opinion there are pros and cons to both tag based structures and more traditional folder based structures.
In tag based structures the context for the content is uncertain – because it depends on how the user got there (using their unique mental map) and this means that the content has to be written to provide it’s own context. Most often content isn’t written this way (although it should be as finding through search has the same impact) so the work in delivering a site this way, unless it’s brand new, can be a much larger job than expected or budgeted for. For example – the custard powder would need to give very clear indications that it was not just an ingredient for generic baking if filed with the cornflour, baking powder, icing sugar etc because it no longer has the context of ‘pudding’. As someone who knows what custard is – YOU would identify it ok – but what about the person who’s getting cake baking stuff for their new flat and has never seen custard before? They would have to pick it up and examine it carefully to decide what it was unless it said ‘pudding’.
In folder driven structures you have plenty of context surrounding your page content but you have the issue of an enforced mental map. However! This can be alleviated through related links and social linking (people who bought this also bought…) In the custard example you may find that in the baking needs section, other people also looked at custard and oxo cubes and soup mix for example. How you decide to derive the information to allow for these other mental maps of the world is up to you but it can be done through… yes! card sorting.
I think the real value in card sorting is in hearing how people think as they work on placing content. I always use group sorting and set it up to encourage a bit of argument (as you know I usually work in intranets so this is a bit easier to arrange). And – it’s cheap and not technology based which encourages more people to feel that it is their opinion we’re interested in and that they’re not being tested.
I’d be keen to hear what other ways you use to get this kind of discussion going about specific content when not card sorting though 🙂 You’re one of the most innovative people I know in this field, and you obviously have winning recipes for getting the good info – maybe you can convert me too?
Just to be clear – I’d never recommend just using card sorting alone to inform a structure.
Good post Zef. And as the author of a book on card sorting I thought I should offer a comment 😉
You are completely right.
We should almost always make information available in more than just one way. Some people think differently to others about information and its relationships, some will expect to find things in different places.
Card sorting can actually help with that – it can help you see how people think about it and what types of things they would clump together. You may find that 90% think of recipes by cuisine and 5% by ingredient and none by category – that would help you make a decision about which facets to offer.
It sounds like you’ve witnessed card sorting used wrongly – as a way to come to the single best hierarchy or as a way to see what cards people would put into pre-determined categories. This is card sorting used as a lazy way of doing IA – and it never cuts the custard.
Thanks for your comments!
There’s a good reason for my unorthodox(?) approach – too often I’ve been called in to design Information Architectures where the bulk of content does not yet exist, is loosely defined or just “plain messy”.
In other cases I’m asked to fix unusable websites or intranets where content has been locked into a hierarchy which was perhaps right at the time, but no longer isn’t.
I think this has forced me to think of ways to create architectures that will work for the people putting in the content (author/editor/manager) as well as end-consumers of the content.
Designing a system where the content author/editor/manager can assigning multiple attributes/facets to content (targeting multiple audiences) means that the navigation structure can easily be adapted during the design/build process and in future as well (without having to re-architect the entire website).
Enjoyed the post and comments thus far.
I manage a wiki-based intranet that doesn’t have common CMS features which would allow pieces of content to automatically surface through multiple navigation paths. Folks find content either 1) through site navigation, 2) through the search engine, or 3) through social discovery (e.g. seeing newly created or edited content in relevant Recent Activity feeds throughout the site).
As such, we are limited to a single navigation scheme, though we can manually insert cross-links and we can ensure findabilty (through the search engine) by making content really good.
Because we don’t use a CMS, card sorting is still a useful tool in our intranet design and redesign processes. I suppose point to draw out is that card sorting should be used in a thoughtful and strategic way, and not as a default IA method…
Given the dicussion above, you may be interested in how to create network visualisations of the results of card sorts, showing: (a) how sorted items are related, (b) how the categories used to sort them are related, (c) how the respondents are related.
Go to: http://mande.co.uk/special-issues/participatory-aggregation-of-qualitative-information-paqi/
[…] While a supermarket is constrained by physical space and storing products in multiple places isn’t feasible this doesn’t apply to content management systems. Once the system is up it doesn’t matter how many pages we show this item on. As a result we can focus on how users look for information and provide multiple classification schemes and put information in places that make sense to different people. Note I’ve built this on a similar concept covered by Zef Fugaz (who was responsible for the User Experience side of the Ministry of Transport Intranet) in his article Card Sorting Doesn’t Cut the Custard. […]