When we were discussing how to redesign an email client, it became soon obvious that we all wanted to do something completely new. Inspired by the BumpTop desktop environment our aim was to create a virtual environment, where you could effectively, yet jolly place your emails.

Clustermail makes
emails physical

Clustermail makes emails physical. They have a size, need space and can be stacked and sorted. The touching, moving and placing process helps you to deal with the mail. First it forces you to decide wether to keep it (assess the mail), second where to put it (sort the mail) and third it allows you to set up deliberate email cluster (make use of your intuition). Like on your desk you will know what your heaps are there for and hence get overlook the situation. Once you placed email, it will be easy to find your way back.

Since Clustermail relies on the spatial sense and you could actually touch your email on an iPad obliged us to choose this platform.

During this first stage of designing Clustermail we came to a junction, when we entered the screen design. To the left you could solve the visualization strictly graphically (»Clustermail-2D«). To the right you could reuse Apple visual patterns (»Clustermail-iOS«). We decided to go both ways, since both offer different insights.

We consider the Clustermail approach as feasible

Our Findings:

Clustermail is the attempt to actually visualize email workflows. Although the approach is challenged with the organization of (basically) endless virtual space that remains intuitive, we consider it feasible. The increasing amount and – simultaniously – importance of emails might render our approach more attractive. Eventually Clustermail applies the human spatial intuition to the classification of information.

Our Foresight:

The Evolution paragraph outlines a number of tasks that need to be tackled to actually build a programatic prototype. Some are conceptual question (organization of space), but most of them are about visualization (interaction and tools). Hopefully the constantly improving interaction capabilites of devices answers some of those questions in the future at ease.

The Boxmaps are the playground to organize your emails as you like

Clustermail is a prototype for a new concept. It is developed to show its potential – it is not ready-to-use (please see the Evolution paragraph). Concerning the workflow, we adapted some ideas from David Allen's GTD: Emails get assigned to projects and be processed later.

Clustermail comprises four major elements: There is the virtual desk that can be divided into sub-segments, the acquainted email-inbox which collects all incoming mail, an icon bar to use multiple desks (e.g. projects), there are email icons giving you hints about their contend.

The application relies heavily on drag and drop interaction and rarely employs popup-menus.

The Evolution paragraph below explains more on the derivation of the concept.

After proving that the concept is feasible we needed to reinvent basic interactions to handle the virtual space

Clustermail challenged us in two aspects: In the beginning we needed to prove the concept was feasible. As we became confident we were confronted with reinventing basic interaction patterns. At this point managing the design process meant to reduce complexity. We ignored lots of sidestories to focus on one usecase to demonstrate the concept's potential. Eventually the reducing part has been far more challenging than the inventing part. It often meant to build on unknown ground and – as Professor Müller put it – just trust our intuition.
The design process started with sofa discussions, continued with a paper prototype, wireframes and ended with screen design.

Bridging the gap between your finger and the virtual space

BumpTop joyful access to digital work has been very inspring. Meanwhile it raised our awareness for the giant gap between real an virtual world. If you print all your important mails, your desk will never be sufficiently big to place them properly. How can we use the advantages of the virtual desk to solve this problem without losing the intuitiveness of real world desk? In addition the paper prototype (real world!) revealed the limits to our memory: What did this heap mean, again?

Introducing segments helps you organizing your mails but also raises the complexity

It became clear that we had to move a way from the plain desk to something more subdividing. Our answer, the Boxmap, brought up further questions: How do you handle placed emails in a segment, which is reduced to make up a new segment? Zoom them out until they fit into the smaller segment? The recursive idea of the Treemap suggested to put a further Boxmap into a segment of a (superior) Boxmap. This is technically simple and would immediately solve the limitation of space. But how do you visualize a segment which contains further Boxmaps? How do you navigate through the Boxmap-levels? We developed some solutions – and grew confident that there we can solve the rest.
But then, with the wireframes we had to waive the recursive Boxmaps again and arranged with space limitation. The scene design further challenged the use of distinguishing colors, typography, interaction toolset and visualization model (i.e. 2D/3D-resemblance, "visual story") and platform limitations (finger size).

We spared further interessting aides, such as a chrono-filter, to finish the prototype

The interaction toolset has been cut down to its essential parts. The paper prototype indicated the need of a collecting folder for semi-important mails: Keeping yes, but not worth placing. We dropped that feature with the wireframes.
The input through touching limits the accomodation of context-sensitive functions severly. You don't want micro-buttons or cascading menues. We ended up using the classic iOS popup. It's not a beauty, but it works.
The scrollable Project-Bar and the position of the Global-Inbox symbol, easy in Clustermail-iOS, became difficult to visualize in the straight 2D model of Clustermail-2D. We eventually went for an animated Global Inbox.
A chronological filter was planed in the beginning, but was abandoned for limitation of scope. This filter was designed as double slider on a timeline, where you could also overview the distribution of mails over time. You'll still find this feature in the wireframes.