Sorted

Getting information sorted is as important as sorting spices at home.

How do you sort the spices in your kitchen? You could, for example, sort them by:

• cuisine (eg. Indian, Italian, Thai)

• form (eg. powdered spices, whole spices, paste)

• use (eg. curry spices, baking)

• alphabetically (as the supermarket does, see above).

Each sorting system has pros and cons. For the spice example, some might be used in multiple cuisines and be hard to categorise that way. Powdered spices might be too big a category and need breaking down into subgroups.

Whichever sorting system you choose, your goal for a spice collection is to allow you to find the spice you need quickly without thinking too hard. If your spices are jumbled, or the categories too complex, you may have to take out all the packets to find what you want. Your curry might have burnt by then.

Sorting webpages

Just like spice collections, websites need sorting to make them easy to navigate. Otherwise important content could be lost under misleading section headings or among unrelated information. Users could get annoyed, looking in places and not finding what they want.

How content designers and creators sort information with users’ mental models in view can make a website a lot more useful. Designing how website information is best categorised, an information architecture or IA, can be aided through a treejack test. A treejack test allows you to try out your sorting system on website users.

How treejack tests work

To design a treejack test, you create a set of tasks that ask the user where in the given IA they would expect to find a particular piece of information. You provide an IA for them to navigate and ask them to select the page they'd find it on. For a restaurant website, you might ask "where would you look to see what times the restaurant serves dinner?".

Treejack tests measure:

• How accurately people find what they want - do they get to where the information actually is?

• How directly they get there - do they down wrong pathways and have to backtrack?

• How long it takes to get to their chosen location - do they have to think about it or is it quick?

Recently I got to design and use treejack testing to plan changes to a website. A few things I learnt were:

1. It’s good to keep a treejack test brief. More people will finish the study if it is. I followed the advice of setting no more than 10 tasks. The response rate was high and the test took the average user 9 minutes to do. I've seen tests with 21 tasks. Only the committed will finish these.

2. Don’t lead your users by being careless with wording. Results can be skewed by using page labels in your task descriptions. You need to avoid doing this or it's not a proper test.

3. Multiple rounds of testing give more confidence in the results. They allow you to compare different versions and iterate on your IA. Then you can be more sure you're approaching the right structure for your website.

There are plenty of resources to explore if you are interested in designing and testing IAs. Optimal Workshop's treejack tutorials are a good place to start. .