At this point you may be thinking: ‘But what navigational context is there, apart from keywords?’ Of course, for many simple (aka web) search experiences that’s all there is: a handful of keywords in the opening game that are then echoed in the middle game. But many professional search applications (such as those used by lawyers, scientists, information professionals, etc.) make a virtue of offering a relatively complex opening game in which the user is invited to articulate the full extent of their information need in the form of a complex, pre-coordinated query. In these cases, the full detail of that navigational context needs to be propagated to the middle game in a manner that makes its presence transparent and its effects easily editable.
This complexity represents a design challenge that is not always adequately addressed. However, the irony is that many of the most instructive examples of professional search applications are available on a subscription-only basis, and hence not open to public scrutiny and review. In their stead therefore, we will refer to simpler but analogous instances from the open web to illustrate the key concepts and principles. With that caveat in mind, let’s look at some examples.
We can see an example of the role of navigational context (or rather, its absence) when we enter an ambiguous query such as ‘golf’ into eBay. In this instance, we are presented with a slightly surreal juxtaposition of search results for golfing apparel combined with facet values for cars and vehicles:
It isn’t until the user provides context in the form of a product category (e.g. by using the scope selector in the opening game or the facets in the middle game) that the search experience becomes coherent. We see an example of a slightly more complex opening game at The National Archives, which invites the user to articulate two further aspects of navigational context via the associated checkboxes:
These are then propagated through to the middle game via the use of pre-selected facet refinements that are displayed in the breadbox:
By using a conventional breadbox approach, this approach supports the principle of making the navigational context transparent and easily editable. But it isn’t the only way to achieve such an end. At the British Library, for example, navigational context is articulated in the opening game via the use of a search box featuring multiple tabs and a scope selector (which disappears when the second or third tab is selected):
These choices are then propagated through to the middle game by propagating not only the navigational context but the actual opening game control itself:
If the user now wants to edit their navigational context, they can adopt the same approach as earlier. This offers a degree of consistency, but refreshing the search results as the user switches tabs feels slightly unexpected, particularly as the tabs give the impression of being scoped to the search box only, and applied retrospectively rather than instantaneously.
It’s instructive also to note that it’s not just the results that change as the context switches: the facets themselves do too. Under ‘Main catalogue’ we see a facet for Access Options, but no such choice exists under either of the other tabs. This is of course quite understandable, since this option only makes sense within the context of a main catalogue item.
We see a slightly more elaborate opening game at Portland State University Library, which offers three search variants presented as vertically stacked tabs:
Adopting an approach that is something of a hybrid of the previous two, the navigational context is then propagated to the middle game via the scope selector, which is pre-populated with the chosen category:
But not all search journeys are concerned with the transition from opening game to middle game. For some, it is the transitions within the middle game itself where the subtlety lies. Tate, for example, offers a relatively simple opening game:
This then segues into an ostensibly conventional middle game, with tabs serving as convenient category selectors:
But beneath those tabs lies a further layer of navigational sophistication, as illustrated by the ‘Artworks’ tab:
Now we see options that are specific to that content type, but within the context of all the other available content types. This 2-level approach provides an elegantly scalable solution to the problem of providing the right navigational choices at the right time, and neatly avoids the earlier issue of displaying search results alongside irrelevant facets (or vice versa).
Despite the clear benefits, the search (and/or UX) community has yet to establish any consistent terminology for this approach or indeed to the principles that govern its behavior. In my time at Endeca (aka Oracle), we used to call these precedence rules. In other words, in the above example there would be a precedence rule of the form:
IF content_type=”ArtWorks” THEN display facets [Attributes, Location, Era, Type, Date]
In some professional search applications with multiple content types (and/or categories), these precedence rules can be quite complex, and become the focus of considerable manual review and maintenance.
We see this principle taken one step further at Crate & Barrel, where the user has the option not just to apply secondary facets specific to each content type, but to actually select which secondary facets they wish to see. For example, a search for ‘lights’ returns a results page in which the facets are displayed in a horizontal configuration, and closed by default:
So far, so conventional. But what is unusual here is the ability of the user to select additional facets, and have them displayed alongside the previous ones:
A somewhat unfortunate side effect of this is that it is possible to add facets that seem to offer implausible choices (can a light really have a ‘capacity’?), but the degree of flexibility this offers is unprecedented in most web search experiences, and suggests intriguing possibilities for professional search applications where fine control and navigational sophistication is required.
The transitions between opening, middle and end game guide and shape a users’ information journey, and reliably propagating their navigational context from one stage to the next is a crucial part of the search experience. In this post, we have briefly examined some of the ways in which that context can be articulated, and the different techniques for making its presence transparent and its effects easily editable. All of these concerns require detailed consideration of the intersection between the user, their search journey and their context. In our next post, we’ll examine a simple spreadsheet-based approach for capturing these considerations in a principled and systematic manner.