6 steps to redesigning a site’s IA
What is IA?
Organised information environments or IAs (information architectures) are all around you. It’s how books are organised in a library and groceries in a supermarket for example. It’s even how you organise your wardrobe. A good IA organises and labels environments in a way that makes sense to the audience. A great IA helps users find information and complete tasks quicker and easier.
For example, when landing on a clothes website, where do you go to look for jeans? Finding those jeans with ease and speed is the result of a useful and usable IA.
So how do I improve an IA?
Creating or improving an IA starts with empathy for that user in a search. This is then layered with an understanding of how items relate to each other within their environment. Various methods of research can improve an IA, here’s a few I've used:
1. Start with a platform assessment
First assess the platform in question. This could be a website, an app or even a store. Audit and gather the IA tree of that website (the structure of a specific platform in the order of where content sits). I use a simple excel sheet to do this, with each column denoting a category level.
Ask your back-end development team to supply this information, along with the number of items per category. If that’s not possible, try a spider tool (like Screaming Frog).
Looking at the IA without the veneer of the user interface allows you to interrogate its structure. How many categories are there? How narrow or wide is the IA? How shallow or deep is it? Are the number of items in each category the same or vastly different?
2. Benchmark against the market
A bit obvious, but always benchmark your competitors and peers. There are great IAs out there and awful ones too. Don’t assume the bigger the company the better the IA. Test it. For example on a bookstore’s site where would you find a Harry Potter novel? Ask questions along the way: Is it categorised by fiction, fantasy or children? How successfully do you find it? Do you reach dead ends? How long does it take to find?
Don't prohibit yourself to the digital realm either. In researching the IA of a second hand book distributor we benchmarked both online as well as bricks-and-mortar bookstores. Again always asking questions. How do they organise the floor plan, the sections on that floor, the bookshelves, the shelves etc.?
3. Conduct stakeholder interviews
Are there any constraints to implementing a new IA? Speak with stakeholders to find out.
Creating the most seamless information structure is useless if it can't be executed. Meet with various facets of the business to understand the structural capabilities of what can and can’t be done. Serious IA changes could have knock-on effect on operational aspects. Talking with stakeholders on a recent project helped us realised we needed to consider how items were organised in their warehouse too. As a result we supplied an IA that kept infrastructure changes on the physical 'back end' to a minimum.
4. Conduct ‘card sorting’ with users
Card sorting is a cheap and effective method of gathering user perceptions on an IA, and probably the most essential activity when restructuring an IA. In a card sorting exercise, participants organise things or topics into categories in a way that makes sense to them. Give them tasks that are representative of the kind of things your end-users will do.
You can supply cards with categories already written on them (“closed card sorting”) or encourage participants to write their own category labels on these cards (“open card sorting”). For more on card sorting check out the usability book of knowledge.
5. Build and rebuild your IA
Most likely, you would have had ideas on how to structure or restructure certain items while you were conducting your research. Now it’s the time to try them out. Take a bunch of post-its and experiment through rapid trial and error. Probe what you’re looking at. Is it in agreement with your user research and not in violation of business constraints?
Have you considered all the main use cases and any important edge cases? Even if you think you have, you still won’t get it quite right. That’s where validation testing comes in.
6. Validate your new IA
For me, this is one of the most valuable research steps. It involves testing the new IA against your current IA. It’s important to take a step back and figure out key performance indicators (KPIs) for validation. What needs improving when users are trying to navigate your site? This will depend on the context of the task:
- Is speed important?
- Accuracy?
- Number of clicks?
- When users are trying to get to the right newspaper article?
Creating KPIs establishes what a good IA means to both the specific business and its users.
The IA validation software we use is Treejack by Optimal Workshop, where we create a series of questions asking users to find something within an IA"
Usually 10 tasks is enough per user otherwise they lose interest and therefore accuracy. Considering a sample size will depend on how much confidence you need in the test. How certain do you need to be, how broken is the site and how much revenue is riding on it? Other factors include how many versions need to be tested and how complex each of those versions are. For the particular test above we used a sample size of 45 participants.
The validation test doesn’t need to be the complete IA. Test the two or three most business critical areas or categories. As always, iterate as much as it makes sense (or your budget allows):
- create,
- test,
- amend, then
- test again.
Repeat until the IA has ticked off those KPIs established before the test. These lessons learned will inform the labelling across other sections of the platform. Future proofing A well architected site means content is connected in meaningful ways. With screens being increasingly less present in new technologies, there’s never been a more important time for companies to invest in their IA.
Take the soon to be released Google Home or the Amazon Echo, both smart speakers with voice activation systems. Information and the structure of that information is what allows these products to be usable. Paying attention to the IA is future proofing your site for however users access it on the ever emerging stream of devices.
You might also like our article on experience maps – they are an essential tool to help understand the experience a customer has with a product or service.