Top-Down Boxes or Bottom-Up Piles?

A chat with Ryan Singer after the next08 has been bouncing around in my head for a few days.

Photo by A30_Tsitika

Photo by A30_Tsitika

We got onto some basic info architecture assumptions which define almost all sites my company makes, and most sites in the web. The more we explored the idea, the more both of us were surprised at how these assumptions often keep users from doing what they’re used to doing in the web — finding things fast.

Top-Down Boxes in Boxes

A new site structure usually starts with a number of assumptions. Big Fat Assumption Number One is that the content chunks will be stuffed into a hierarchy of boxes. The first group of boxes form the 1st level navigation, and each of those boxes usually has more boxes inside it. Often the 2nd level boxes have more boxes inside them, and I’ve seen this boxes in boxes structure repeated down to the 5th level.

A user can only try to guess which box what he’s looking for might be in. And instead of getting the content chunk he’s looking for when he selects a box, he gets a new set of boxes and starts guessing again. When he finally reaches real content (after guessing, guessing, guessing, etc.) the chances are relatively good that he’s not found what he wants, and the guessing game can start from the top again.

If individual chunks of content (or user goals) are the bottom of our structure, we’ve just built an info architecture from the top down.

Hierarchical boxes are also made of “conceptual steel” and separate chunks from each other. This works for a book: the boxes are chapters, which have boxes in them called paragraphs, which are full of chunks called sentences. But once a book’s printed it doesn’t change. Web sites change constantly, and boxes are change-resistant.

Bottom Up Piles & Lenses

Instead of imposing a structure on content chunks from the top down, why not look at the chunks themselves first; i.e. bottom up? If we find common attributes for the chunks, e.g. colour, and label each chunk either blue, green, yellow or red, then we have a labeled pile of chunks.

To find something in a colour labeled pile, users could use “lenses”. A red lens would make all blue, green and yellow chunks disappear, leaving only the red chunks visible. If our chunks also had an attribute size–with labels big, medium and small–they could then combine size and colour lenses to quickly find large/red chunks, or small/green ones.

Old Hat?

Sure, none of this is particularly revolutionary–it’s the way a great deal of those sites we insist on calling “web 2.0” work. Flickr is a gigantic pile of images whose labels are tags, but also technical details such as which camera made the image, Creative Commons license, interestingness, etc. Most of Flickr’s navigation doesn’t throw the user into a box, it provides them with a lens through which they can look at the chunks they’re interested in, and ignore the rest of the pile.

Where The Hell am I? Who cares?

The assumption that drives us to make top-down box architectures is that without a structured, categorised series of boxes in boxes, the poor user will lose his orientation. I think this is seriously outdated thinking, which comes out of the interface thoughts of the pre-internet software design era. It might make sense in the focused, daily-use context of an application, but a user who can find any page in your site from google, and jump directly to it, doesn’t give a crap where he is as long as he finds what he wants. When I’ve asked a number of daily-use but less than cutting-edge users “where” in the web they are right now, the answer surprised me: “What do you mean? I’m in Google.” So much for that carefully thought out color-coded box hierarchy.

Hunters & Gatherers

Am I recommending that all of my corporate clients throw their hierarchies out the window? Hell no. But I would like to see them experiment a little more with piles & lenses to supplement their box hierarchies. Traditional boxes-in-boxes navigation is fine for gatherers, but google is teaching users hunter strategies: select a goal, focus on it, jump on it as quickly as possible. The next time you start a concept with Big Fat Assumption Number One, and begin stuffing content into boxes-in-boxes, take a step back, look at the chunks at the bottom, and see if you can offer the hunters piles and lenses to speed their hunt.

Comments

  1. Craig says:

    Good article Matt, it really got me thinking about the different ways we use navigation. When I’m starting a new website, I always try to avoid making a navigation that goes any deeper than one level, because I just think it confuses people. More often than not a website isn’t big enough to justify such a large navigational structure, but people naturally think we need them.

    In reality, when have you ever bothered using a site that you can’t find what you are looking for in the top navigation? If I can’t see it from the start, I click off and visit somewhere else.

  2. Mb. says:

    Well, one level navigation’s fine if you don’t have very much content, but if you’ve got a huge pile of chunks, then you uually need more boxes in boxes to organise it, otherwise your user’s going to be overwhelmed after his first click.

    As for how long I’ll click through the navigation, it depends a lot on how clear and determined I am about my goals, and if I can find what I need on another site. For example, I don’t have any choice but to keep clicking if I’m looking for the service hotline number for my ISP on their site.