connecting*the*dots

I’m currently working to prep the project environment of one of my clients — a domain that relies on ad sales for survival. The stakeholders have hired me to lead the redesign of their site, which includes the information architecture. Knowing the degree of “get it” in the domain, I need to provide easily digestible “IA” education before I can move forward with my design methodology to improve the tactical findability of their most valuable content.

Yes, it’s a typical IA consulting gig, but I’d like to establish a reusable approach; not for creating explicit architectural solutions across different project types, but with a presentation of explicit, findability techniques.

I’m looking for feedback of my current progress, so if you’d like to participate, feel free to comment on this post or contact me at spcoon [at] seancoon [dot] org. Please feel free to point me to any existing work available online as well. Once I’ve pulled together my findings, I’ll iterate my work and release it into the ’sphere for anyone to use.

The Baseline

Humor me for a moment and try to forget everything you know about classification, structure and order. Instead, imagine that the only element of a web site (we care about) is the most holistic/granular information object:

• On flickr, it would be the image page
• On Amazon, it would be the product page
• On Wired, it would be the article page

Now imagine that your goal is to increase stickiness across this entire object level. Remember, the revenue model is ad sales, so the more content explored, the better for my client.

Each of the previously mentioned domains have crafted specific information architectures to accomplish this goal of “stickiness.” They also have extremely different revenue models, so the “value” of findability is relative to the nature of the product, the domain’s degree of advertising/marketing and the bottom line.

For example, flickr image pages aren’t weighed down with contextual recommendations of similar images from other users (similar to how products are displayed on an Amazon product page) but the inclusion of a simple globe icon next to an image’s tag does expose index pages of similarly tagged photos from other users. This increases discovery, which both entertains me (the user) and increases page views for potential ad clickthroughs.

Different context; similar goals. Expose avenues of findability in the interface to increase domain stickiness.

I’m currently illustrating technique similarities (that are not domain specific) for optimizing information architectures to expose valuable content. Again, consider this exercise an effort to describe a baseline standard, or best practices for findability that can be reused in one way or another across project types.

Along these lines, I’m clarifying techniques by using non-specific terminology (i.e. contextual and relational are generic terms, as compared to collaborative filtering). Secondly, I plan on augmenting the illustration for this particular client by labeling specific values (a 1 to 10 scale, possibly) to the various avenues of findability, distinguishing the value proposition (ROI) between focusing on, say, relational discovery compared to categorical browsing. I won’t be able to complete this second part until user research has been finalized.

Here’s my current list of best practices:

• Object level contextual discovery: Hyperlinks to contextual content, embeded within the primary object of the page (i.e. hyperlinks within an article to other articles, linked notes on a flickr image, etc.)
• Object level relational discovery: Accessible related objects, determined via appropriateness (i.e. as simple as “Related Articles” or as complex as “Other shoppers purchased…”)
• Object to index level relational discovery: Using tags to move from the object level to the index level (i.e. flickr globe icons, del.icio.us tags, etc.)
• Index level relational discovery: Related tags presented from a mass sample of tagged objects (i.e. a tag search on Technorati creates a list of related tags to the original query on the index page)
• Tag/Meta-data search: Optimizing tagging to improve the results when searching objects that have been explicitly tagged (i.e. Gmail labels, Technorati tags, flickr tags, etc.)
• Full-text search: Optimizing objects and result pages to increase precision and to manage recall into precise, secondary, relational options in the presentation layer
• Categorical navigation: Traditional top-down navigation, with a focus on keeping categories both shallow and non-cascading, while keeping the breadth of topical choices as narrow as possible

The diagram (177kb .pdf) displays another element — Third party relational discovery, which is specific to partner deals with external domains.

Ideas, feedback, critique; all appreciated.