While observing customers using the product, I quickly identified a behavioral pattern that frequently preceded complaints about information findability. In their minds, customers maintained a list of attributes they deemed critical for initial evaluation; if a prospective date failed to match this mental list of attributes they were immediately dismissed. To make things more interesting, some attributes were about who this individual was, while other attributes pertained to who this individual sought. The former was represented by people seeking dates that did not smoke or drank casually, while the latter exposed personal insecurities like acceptance of one's body type or one's height.

The solution involved three separate tasks:

• The definition of a taxonomy for the categorization of the structured data sections
• A new layout that made it faster to contrast an individual against what they seek in a date
• The creation of a module that allows for the customized display of user-prioritized topics

Card Sort Exercise
After taking an initial pass at categorization of topics, I validated the foundation of my design with a card-sort exercise. The card sort provided valuable information about the priorities of online daters and allowed me to finalize the taxonomy of the dating profile's structured data.

The diagram shown here is a representation of where "relationship seekers" place importance. This is important because while relationship seekers and casual daters both find appearance attributes important, casual daters place a higher importance on "interest" and "lifestyle" categories whereas relationship seekers care more about "background & values." This was important because it helped me understand that my design would have to serve two distinct types of users.

I created and used diagrams like this to understand and define the information architecture of the dating profile. The abstraction of the information architecture from the actual design helps communicate design priorities without the coloring expectations with an early design sketch.

Layout Explorations
The content, layout, and tone of the dating profile was iterated several times before being tested with users. The final wireframe shown on the left was the product of over 20 explorations and feedback from user testing.

I bagan with exploring layouts at a very high level on a whiteboard and on paper. These explorations often used massing diagrams to explore the relationship between parts of the interface. Once a confidence about the UI elements was reached, I moved into more detailed sketches and iterated them based on in-house heuristic evaluations. The design taken into usability testing was well accepted and required only minor changes.

The final design provided ours customers with an interaction model that improved information findability by clustering attributes into topics and displaying values for who the individual was and what they sought in a side-by-side arrangement.

First Impressions Module
As mentioned earlier, the natural behavior for most users while reviewing a dating profile is to quickly check for deal breakers and must-have attributes. If the dating profile does not pass this check, the searcher immediately begins reviewing a different profile.

The user's need to quickly review profiles is actually grounded on their desire to not build any kind of emotional attachment to a person until they know that they meet a bare minimum set of requirements. Though one is able to screen-out people as part of a search query, many searchers opt for broad search queries in order to retain control over who is rejected and who is reviewed. I addressed this user need by providing a customizable module at the top of the interface so that users can see what they deem to be important.

The flow diagram displayed here shows how this module can display a default set of attributes if it has not been customized, or can show a customized set if the user is signed in and has defined such a set. To define the contents of the default attribute set, I analyzed search logs to isolate the most common search topics. The set was then compared against the card sort exercise results and prior user observations to validate their accuracy.

The display of these important topics is identical to the way other profile sections are displayed. Values for who the profile owner is and what they seek are displayed side-by-side to allow for quick comparisons and evaluation. By working closely with a visual designer early in the design process, you can see how the wireframe and final design versions are very similar.

Customization of the module, on the other hand, was iterated during usability testing and its final incarnation differs from the wireframe. User testing revelaled a strong user desire to specify the order of selected attributes. Though this change happened very late in the design process, it was carried out because of my involvement in the project from start to finish.

In both the wireframe and final design, a modal dialog box is displayed when the user chooses to customize the set of attributes. Users choose the topics they feel are the most important for their initial review and save their settings. The modal dialog is dismissed and the content of the module is updated.

When discussing this finding with Product Managers, the knee-jerk reaction was to add more structured data topics. Product Managers hoped that the addition of witty and interesting topics would offset the need for depth since more breadth would be available. In reality, however, the expansion of breadth would only exacerbate the problem. To verify this, findings from the open card sort exercise were used to explain our customers' mental model for profile creation. Participants were asked to selected and group the topics they thought were most important toward evaluating a person's dating profile. Without exception, all participants added open-ended questions targeted at expanding on key topics.

I designed a system that allowed users to add both depth and breadth to their profiles. During validation, however, it became apparent that two new data points had to be considered -- time and fear. While answering structured data topics is simple and quick, answering an open-ended question is very time consuming; more importantly, it can also be emotionally draining. It would be very unlikely that most users could devote the time, or have the stamina, to complete a profile in one sitting. The second datapoint, fear, was manifested in the blank stare and hesitation users had when confronted by an open-ended question. Even though one may have just answered a structured question pertaining to their favorite movie genres, the follow-up question "What is your favorite movie?" resulted in unprecedented self reflection. Users worried that a straight answer would be seen as too vague, but too much information could reduce conversation on a first date, or could be used to stereotype them.

I iterated the design a few more times and ultimately arrived at a solution that embraced the need for users to take breaks throughout profile creation. The final design also gives users the option to answer a single open-ended question or answer more granular questions. Unlike the first iteration, granular questions are presented with the option to select a pre-written answer based on structured data input, or the option to answer the question in one's own words. The pre-written answers that users could select were purposely dry and shallow in tone; their purpose in the interface was to provide a starting point and gentle nudge about how to answer a question.

The first phase of this project addressed the consumption of dating profiles. Since its launch in December 2006, the number of profiles viewed has increased 34%, subscriptions increased by 15%, and communication between users increased nearly 25%.

The second phase of this project is currently underway and will address the profile creation process.

As design lead for this project I worked with two visual designers, one user experience researcher, and another interaction designer. Together we conducted field studies, reviewd designs, and build visual design mood boards. I oversaw many elements of this project's design and worked in a highly collaborative environment to craft an experience that properly addressed our customers' needs.