When I am conducting IA training, I like to deconstruct and critique the BBC Food site. It’s a fascinating exercise, illustrating lots of UX and IA principles, mostly excellent with enough you’re kidding moments to enrich the learning.
We start with a quick walk through the site. It’s busy! Twelve thousand recipes, richly connecting to ingredients, chefs, shows. Really busy. Structurally busy.
But it is busy in the same sense that a mall is busy, and this gives us some useful insights.
In each case, we are building places with lots of offerings of different types, where users go to meet their specific needs. Just as we don’t go into every store in a mall, users don’t look at every page in a site like BBC Food.
Of course, there is no such thing as “users”. Instead, there are people with goals. Some visitors to a mall have a specific shopping goal, for example to get a new headset for their iPod at the Apple store. Others may want to find that perfect Christmas present for their partner. Some may just be getting out of the cold.
It’s the same with BBC Food. Some visitors may want a lobster stroganoff recipe because they enjoyed it on a recent trip; others may want to cook an ostentatious Christmas dinner for their in-laws. Some may just have got there by accident.
Designers have to help all types of user meet their goals.
Navigation comes to mind right away. Both the mall and the BBC Food site, as well as individual stores and sub-sites, have their own navigation schemes: floor plans, directories, lists, menus, signage, and headings for page sections. Navigation is pretty familiar.
But navigation by itself is always a compromise, pointing us in the right direction but seldom getting us exactly or uniquely there.
The floor plan tells us that the Apple store is at the far end of the next corridor on the right, but you still have to walk down that corridor, and scan store fronts until you see it. And once you get in the store, you have to find the iPod area. As it happens, there is a big sign for iPod (a different navigational scheme than the mall as a whole). But at the end of the day, you have to scan the packages of earbuds one by one.
This interplay of navigating and scanning is typical, but combinations vary. In the Apple-iPod-earbud example, there was a lot of navigating and a tolerably small amount of scanning. In the perfect-Christmas-present example, there will be a lot of scanning, some would say intolerable.
Intolerable as the scanning may be, it is certainly efficient. We can assess in an instant whether an unfamiliar store is likely to contain that perfect Christmas gift. We walk along, judging yes-no-maybe without breaking step. We don’t even go into some stores.
How on earth do we do this? Well, it is based on brand and signage and numerous display cues, and on our goals. The window dresser has sent us a myriad of signals that they design with skill. We respond to the set of signals, even if we can’t recognize and articulate them individually. “Too expensive, too dull, nothing new, worth a look, she’d kill me, let’s go in” we summarize, ruling things in and ruling things out in the blink of an eye.
Designers in the online world are the counterpart of skilled window dressers. They provide skillfully designed signals about content; the user can decide whether the content is useful without having to read the content itself. We call these signals “information scent”. And just like real-world scents, they are complex and elicit an immediate yes-no-maybe response.
Information scent is used extensively throughout BBC Food. Some content areas are handled in a straightforward and systematic way.
- Chefs: we know an area of a page refers to a chef because there is a head and shoulders picture of the chef, captioned by the text “By” followed by a name, styled as a hyperlink.
- Programme: we know an area of a page refers to a BBC programme because there is slice of life picture of the host(s) in their show setting, captioned by the text “From” followed by a programme name, styled as a hyperlink, and followed by a date.
Is this scent effective? It depends. If I am an avid viewer of a show, this scent is very effective. It will be used in conjunction with my goals to come up with the yes-no-maybe assessment: “they always do good stuff”, “I can’t stand them”. If I don’t know any of the shows, then it triggers no response.
Other content areas are less consistent, and need considerable design skill to provide effective information scent. Let’s take a look at Ingredients as a great example; they range from highly recognizable vegetables to obscure spices and enigmatic bowls of stuff.
BBC Food cues up an ingredient by its name and an image of the ingredient in a kitchen setting. Here are some examples of information scent and its effectiveness in triggering responses in users.
- The ingredients “Dover Sole” and “Sea Bass” both have a thumbnail of a fish. The thumbnails do not help me distinguish the fish; however, if ever I need a medium-sized fish, these would be candidates, and I would know to examine each ingredient in detail.
- The ingredient “asafoetida” is completely unknown to me. It is illustrated by a small mound of yellow particles, perhaps a grain. It turns out to be a spice and I later figure out that a small mound is a common representation for a spice. The thumbnail is useful to me as I am currently looking for exotic meats, so I can rule this ingredient out.
- The ingredient “gooseberry” is illustrated by glasses of white gooseberry fool; they look delicious. Would a thumbnail of a gooseberry have been better? More literal perhaps, but not as enticing.
Providing useful information scent is a multi-faceted, nuanced activity.
So where does all this leave us? Let’s summarize the whole exercise this way. In large complex information spaces (malls, BBC Food), we can and should help visitors by providing both navigation and information scent.
Navigation is the better known. It involves building classification schemes, taxonomies, labelling systems, as we have learnt from the Polar Bear book and others. It is mainly left brain work, but needs an ability to understand user goals and behaviours, and to engage users in testing your constructs.
Information scent is less mainstream. There are research articles, illustrative articles (on labelling, glosses, icons, providing previews), but there is no equivalent of the Polar Bear book.
We should introduce information scent into our conceptual framework and vocabulary, and ask how it can be incorporated into our solutions.
Until the definitive book is written, we should watch for examples of information scent and see what lessons we can bring back into our solutions.
For an excellent introduction to user search behaviours, information scent, and with comprehensive references, see the book “Designing The Search Experience” by Tony Russell-Rose and Tyler Tate, reviewed here http://theinformationartichoke.com/designing-the-search-experience-book/.
For a detailed application of information scent to help users process search results, see http://theinformationartichoke.com/information-scent/.