Tinderbox for Information Architecture, Pt. 1
During the recent Asilomar IA retreat, we discussed various tools we use to do our job. Some of us mentioned using Tinderbox, and after one of the sessions a colleague asked to see samples of the sort of stuff I do with this tool. Unfortunately we couldn’t catch up again on this topic during the remainder of the retreat, so I thought I’d post some samples here.
-more->
During the first meetings for a new project, I try to get an impression of the content that will make up the site. Folks will usually talk about this stuff in a very unstructured manner; they will start with the obvious things (such as products or services), and other items will come up randomly as part of the conversation. (I also try to ask pointed questions to help them think in terms of “webby” content.)
I’m a fairly fast typist, so I use Tinderbox to take notes of all the content concepts that come up. Each new concept becomes a new note (this is done very quickly in Tinderbox by typing [return]-title-[return]-[spacebar]-text, a key combination that soon becomes second nature). During the meeting, I make no effort to organize the notes; at this point it’s more important to just listen and type. When an obvious link between two concepts is discussed, I create a link between the two relevant notes; apart from these links and the occasional grouping, the result of these meetings are completely disorganized groups of notes.
In later sessions, we take these notes and start grouping them in Tiderbox-based card-sorting-like exercises. I use decorations in Tinderbox to highlight the content groups as they start emerging. I also start exploring links between notes, and understanding how they relate to each other. This starts to suggest where secondary level pages are needed, in a typical bottom-up fashion.
Up to his point in the process, Tinderbox has been used primarily as a set of digital Post-It notes. However, the real power of the tool comes in when attributes are applied to the notes. Each note can have an arbitrary set of attributes—metadata—which can be easily assigned and modified. I use the attributes to keep track of items such as the content source (who on the customer’s side is responsible for producing the content), whether the content item will be presented statically or dynamically in the site, and all the other typical metadata we use when talking about web content. If sample content has been submitted by the client, I will start organizing the content itself into the relevant notes.
The end result is a rough sketch of the site’s structure and content. The clincher is that Tinderbox can generate entire HTML web sites from these notes, so site prototypes are fairly easy to generate. These prototypes are usually not fit for public consumption; however, they allow us to very quickly test the proposed structure—especially its navigation—without having to fire up a separate development environment.
What I like about this process is that I can start from the roughest of rough sketches (the disorganized content concepts gleamed from the first client meetings) and go to a working prototype of the site relatively quickly, all within one tool, and all done in a very visual manner. All of the information stored in Tinderbox can be exported to any textual format, including XML, using templates you can define.
Although this approach is very useful, it’s also fairly simple. It doesn’t even employ one of the most powerful features of Tinderbox: agents. These are notes that organize other notes automatically based on rules that employ the notes’ attributes. For example, you could create an agent that groups all the notes for whom the content source will be Jane Doe in marketing. I am constantly exploring the capabilities of this extraordinary tool, and am currently working on templates to use for usability testing and requirements gathering. I’ll post more examples as they start to gel.
