DonnaM » Blog Archive » How I draft an information architecture

How I draft an information architecture

When I teach information architecture, the most common questions aren’t about the principles, but about the process. Just how do you decide on a particular method, how do you choose categories, how do you know what you’ve come up with is right.

As a teacher I’ve spent much time thinking about this, reflecting on my own process and how it actually works for me. And I’ve figured out the answer.

When I teach I tell people this answer. Most are surprised, but there are always a couple of people in the room who nod to themselves and look comforted.

Would you like to know the answer?

Here it is.

Wait for the surprise.

Just make it up.

Yep. As some high profile product says, just do it.

This is surprisingly easy, but there is a dependency. You need information. You need to understand what you are trying to achieve, what users of the service need and know, and you need to know the content well. If you don’t have these things, it will be hard. But if you do have them, pulling them together into a first draft is surprisingly easy.

When you have made something up – and I don’t care whether you do it on a whiteboard, in a spreadsheet or in your head – then start thinking about whether it will work for the users, and whether it will work for the content. Revise and play with your idea until these things start to fall together.

At some point you’ll start to feel good about your made up draft IA. In your head it will look like it will work for the users and the content will fit in. You can see how it will help the organisation achieve its goals. (If you can’t get to this point, it’s likely you are missing some kind of information. You’ll need to do something to fix that, or it will never work).

For me, there’s a funny feeling at this point – it feels simple and clear and makes me wonder why it took me so long to figure out.  That’s when I stop fiddling and start talking to people about it. And I’m yet to be be majorly wrong, so there must be something in it.

Today’s example

So I was doing this today for a client (I won’t say who, not that it’s sensitive – I just haven’t mentioned this to them). I took photos as I played with the new IA & thought you might be interested.

The background is that this is a fairly straightforward redesign of some government content. I know what the team want out of it, have done some basic user research, cleaned up the content and decided what to keep and create. It will be for a simple hierarchical site, so the IA in this case is a set of categories/subcategories to be used in navigation.

I don’t always do it like this, but today I jotted the main content chunks onto sticky notes. Apart from being physical, this helps me get away from the way content is organised now:

Then I jotted things we learned from user research onto a different colour note:

Then I shuffled them around into groups of things that go well together:

I played with these for a little while and moved a couple of things between piles, but it wasn’t hard. It fell together easily. All the user needs are catered and all the content fits in. There are some spots where we need some extra content, and there is no content that someone doesn’t need. I could come up with sensible titles for all the groups.

My followup step from this was to create a quick spreadsheet with the categories, subcategories and main content pages that came out of this shuffle; and also add a couple of special navigation items that will pull together some things across the piles (like forms and publications, which are scattered across the piles according to their topic, but will need separate entry points for some user tasks).

Next steps

Tomorrow I’m going to run through this with the client team. I’m not drawing it up into a sitemap now. I’m going to talk them through it as I draw it on a whiteboard. This lets me describe each section and the rationale behind it, without them being distracted by something I have already written down. I’ve done this before and it is a great trick for communicating the draft IA – it lets me present it as a story, and triggers sensible questions.

Then, of course, there is lots more to do. I’ll be using an existing navigation approach, so don’t need to design navigation. But I need to design all of the index pages – these will introduce each topic and provide deeper links plus cross links.

And we need to revise all the content. As this happens, the IA will change, but if I’ve done my job well it won’t change dramatically.

What, no card sort

OMG, I just wrote a book on card sorting, and didn’t run a card sort. Why not?

Well, I did other forms of user research which gave me a fairly good idea of the main issues and needs for users. Most users only ever need one or two content pieces, so there was little point getting them to do a card sort on things they don’t care about. And I didn’t feel like I had big gaps in my knowledge that would mean a card sort would help. So I didn’t.

We’ll do usability testing on this as well, probably before content rewriting and after.

30 Responses to “How I draft an information architecture”

  1. Krissy Says:

    Very interesting. Always handy to see the other peoples’ processes!

  2. Gary Barber Says:

    Arrrrh but you did card sort … mentally you sorted it and grouped it all, you did the core component, just without user input ;)

  3. Lisa Says:

    It’s always so interesting to see how other people work. IA is not one of my ‘core competencies’ :) so I always worry I’m over complicating the process. Not at all to infer that this is simple, but just that you don’t always have to make it a huge production.

    And the thing is, that’s where skill is most evident, when you know the best process without thinking about it… ie when you know a card sort would be over kill :)

  4. Donna Spencer Says:

    @Gary. No, I didn’t do a card sort. I know you at least skimmed my book. And I know that I said somewhere in there that I think that card sorting has to involve users. Otherwise it is just shuffling sticky notes around a desk, which is obviously fine, but not card sorting ;)

  5. Jessica Enders Says:

    Thanks for sharing this Donna, very interesting.

    I had a real “you too?!” moment when you said at some point “it feels simple and clear and makes me wonder why it took me so long to figure out”. Exact same thing happens to me with my designs and then I know it’s ready for the next round of input/critique (e.g. from the client). Also vice versa – when I have a niggly feeling that something still isn’t right it usually isn’t so I need to keep plugging away.

    Fascinating, really.

  6. Donna Spencer Says:

    @Jessica – see. you are one of the people who, in my workshops, nod & think ‘OMG, it’s not just me’. Experience lets us internalise and feel whether things work or don’t. It takes a lot of hard thinking to deconstruct all the things that contribute to that feeling ;)

  7. Donna Spencer Says:

    @Lisa – your point about skill is a good one. I know that we don’t need a card sort because I know that I’m comfortable with what I know and know I won’t learn anything hugely useful by doing it.

    One thing I see a lot is that we (i.e. our field) pitch user research to our clients as a thing that we just should do, because it is part of the process, or part of the right way to do things. We’d be far better if we said ‘hey, I need to do user research, card sort or other method, because I don’t know enough to do this confidently’ and not pretend that it is anything but something to help us fill gaps in knowledge as designers.

  8. Patrizia Bordignon Says:

    I love your insights into the way we work Donna. You have verbalised the design thinking process I go through when I am creating an interaction design. Good input is critical, then I work at the design until it is as simple and clean as possible. At some point, I will either think the design feels right or some aspect is not quite resolved. That’s when I get the client and/or the target users to take the design for a spin and see how it works and feels for them.

    Today I ran evaluations with landscape architects using a new online tool that has been designed to make their work much easier and quicker to do. I love this moment in the process. Aha moments! They shine lights on the design with their unique perspective on the world, the stuff that only comes from their world of experiences and the way they need to do things. It is ultimately so satisfying to design a new tool that is going to make someone’s working life much easier, for the users and the client.

  9. Ruth Ellison Says:

    Thanks for sharing this Donna. It’s fascinating seeing the processes that each person takes in the design process.

  10. Silvia Reitsma Says:

    Really interesting. I agree with you that the IA process should be somewhat flexible. I just wonder what are the criteria you use to determine the right approach.

    Sometimes I have the feeling I know what I’m doing, user research supports my ideas (or actually my ideas are supported by the UR), but then the client wants something done in their way. How do you deal with that?

  11. Lisa Rex Says:

    It’s great to read it’s ‘normal’ to make things up as you go along. IA planning is definitely not a one-size-fits-all situation, is it? One of my current projects is a simple portfolio-type site with only a couple content types. I don’t even need to spend time on sticky notes or card sorting (I don’t think?). Drawing everything out on paper will be sufficient before taking it to the site map and wireframes stage.

  12. Donna Spencer Says:

    Thanks for your comments.

    @Silvia – if you and your client aren’t agreeing on the way to approach it you either haven’t communicated well or are missing some information. You may need to spend some extra time finding out why the client wants it done their way – they may know something about the detail of the content, or about how it has failed before. And you may need to work harder at making sure your idea is a good one and communicate that. But ultimately it is the client’s call – it is their business. Yours is to provide your best advice, with a solid rationale, to help them make that decision.

  13. staplegun Says:

    I so agree with “just make it up”, because every situation is different. So in the end this really comes down to the (quality of the) person doing the IA – their ability to think in the abstract and make good connections. You can’t really teach this, though if you don’t have that ability I’m guessing you wouldn’t gravitate to being an IA (same with BAs). You can teach methods to help that person gather the right inputs to their brain, but after that it’s up to the magic between the ears.

    This may sound kinda elitist, but I’ve (only) recently realised I look at the world in a different way to other people. While sometimes this can make things difficult, one way it shines is that it makes that IA kinda thinking a synch for me. I find others just can’t see those kinds of connections I’m making – if I couldn’t do that, I can’t imagine how I’d do IA work.

  14. Donna Spencer Says:

    @staplegun I totally agree.

  15. Making it up | one man writes Says:

    [...] And I think I’m at that point in my career when experience plays a large part and, despite any processes that are in place, I have to agree with what Donna said. [...]

  16. Max Design - standards based web design, development and training » Some links for light reading (11/8/09) Says:

    [...] How I draft an information architecture [...]

  17. Bruce Bowden Says:

    In the past, where I have done card sorting exercises, I’ve often felt that i hadn’t really covered all the audience groups, I hadn’t really done it ‘properly’. But I had reached the point where I personally had a handle on how the content was going to come together. So, thanks to your article, I now feel retrospectively less guilty. Of course, it helps that the feedback once the site went live was very positive.

    The only problem now is when someone who didn’t have a clue pulls an IA out of thin air – Its very hard to argue that they are wrong just because you have more ‘magic’ than them. Card sorting was such a nice concrete thing to use as a defence.

    In the end that particular problem was solved for me when they offered me lots of money to go away. Now the organisation is trying to work out how to justify a complete (professional) redesign on a web site that’s just been launched.

  18. Donna Spencer Says:

    @Bruce. Thanks for your comment. I’ve found that even inexperienced people can create a good IA out of thin air, and experienced people sometimes create really bad ones (actually, this happens far too often). The test is whether it works well for the content and for the users – and to check those you need some information ;)

  19. links for 2009-08-14 | Ip's. Says:

    [...] DonnaM » Blog Archive » How I draft an information architecture "Just make it up. [...]

  20. Jason Says:

    I attended the LGweb09 conference on 20-21 August 09 and sat in on your presentation. I found you extremely informative and you cleared up much of the confusion which I had regarding IA. When I next approach this once daunting topic, I’ll be more confident when the task is at hand.

    As a side note, you feature in this week’s eGov newsletter at


  21. Donna Spencer Says:

    Thanks Jason ;)

  22. Matt Balara, Freelance User Experience Designer in Sydney Australia : “ Get Your Thoughts Together: How to Organise a Workshop” Sep. 3rd, 2009 Says:

    [...] at having committed to doing a workshop, and inspired by my friend Donna Spencer’s blog post, How I Draft an Information Architecture, I went back to a method I’ve often used in the past to organise my thoughts for a pitch [...]

  23. Nathan W Says:

    Donna – its refreshing to see an expert in their field speak honestly about what they do. Im a big fan of the “make it up” approach.

    I get asked all the time if my brain is wired in some strange way because this all comes “naturally” to me. As you say though, card sorts are only 1 way, there’s usually other research or observations you can use too.

    As for the post-it notes, they also work really well on a glass wall…

  24. Joca on stuff (about internet, product management, agile, swimming and etc.) » Blog Archive » How should we draft an information architecture? Says:

    [...] it’s quite simple, as Donna Spencer points out: Just make it [...]

  25. DoingMedia : One Way to do an Information Architecture Says:

    [...] found this post, How I draft an information architecture from the DonnaM blog through Jared Spool’s UIE Tips email and I thought I’d share [...]

  26. Noj Says:

    Great article Donna.

    One thing that stood out for me has to do with choosing the appropriate method.

    I recently worked on a relatively small site but found myself starting to think about card sorting. I stopped because I realised that I know the content and users quite well. I’d already spent considerable time in discovery with the client. So I was able to draft my spreadsheet and get stuck into it.

    A card sort exercise, in this instance, would be an overkill. I could better spend the time getting the copy written correctly for example.


  27. Donna Spencer Says:

    @noj. Exactly, and that knowledge of what to do when really comes with experience…

  28. Deepesh Joseph Says:

    Nice description of how one visualizes the whole IA process for a website design. This is a good starter for those who venture to do IA stuff for the whole enterprise systems.

  29. Ricardo Says:

    I’m not entirely comfortable with this approach. You can’t possibly “know” you won’t learn anything useful. You’re *assuming* you know enough of the users to put yourself in their place – I thought that was one of the basic mistakes designers make.

    A simple card sorting session with a small group can take less than 20 minutes, I don’t see how that would be overkill. At the very least it will confirm your assumptions. I haven’t read your book yet though, so I don’t know what’s your exact POV here.

    happy new year!

  30. maadonna Says:

    Good point Ricardo.

    In this situation I had plenty of other user research – I wasn’t going to be assuming I knew enough about them.

    Something I did know was that individual users were very focused only on their topic of interest – a very small proportion of the whole information set. In a card sort they wouldn’t care enough (or know enough) about the rest of the content to do a decent sort.

    So sure, I may have learned something. I always learn something. But I didn’t think I would learn enough.

    And it isn’t a matter of a 20 minute card sort – I would have had to spend at least half a day contacting people, arranging for them to do the sort, then another half analysing the results. I still don’t think I would have got enough good information from a card sort to justify the planning & analysis. Especially as I had other good user research already.

Leave a Reply