Web UI

Case Studies: Allstate Financial


In 2004 Allstate Financial had begun development of a new call center application that would consolidate access to information needed for answering customer questions pertaining to life insurance policies.

During the initial development phase, usability issues had surfaced, despite the existence of comprehensive style guidelines and visual design templates.

Web UI was hired to find solutions for all usability- and interaction design-related issues and to serve as the voice for user advocacy.

Responsibilities included:

Design Activities

We settled on an approach that utilized usage-centered design techniques. This involved the following user research activities, with the ultimate goal of understanding the annuity call center reps’ main goals, responsibilities, informational needs and work flow patterns.

  1. Field observations: inventory of call types corresponding to the main categories of questions from customers; different experience levels; personal and business goals
  2. Brainstorming sessions: Initially this involved further refinement of the inventory of call types; determining task level — an important component of this activity was making it clear to participants at what level we wanted our task modeling to take place. From among the various choices for task level (cloud, kite, sea, fish, clam), we wanted to think in terms of sea-level tasks in order to have the most effective and useful level of granularity.
  3. Stakeholder interviews: individuals at management level with deep understanding of call center reps’ range of roles and tasks — used to better understand role variations and to refine task inventory
  4. User role modeling: Involved enumeration and descriptions of the roles for which the app was being built. We determined that there were two main roles: experienced rep and new rep.
  5. Task modeling workshops: We held two user workshops for the purpose of creating a task model that used as its input data the completed task inventory. In these workshops, users sorted tasks into three categories based on their usage frequency: high, medium and low. In the next stage, users prioritized the high-frequency tasks to determine those tasks most frequently performed.

Once a first draft of the task model was completed, we conducted another modeling session with a subject matter expert who grouped tasks based solely on their relationships to one another, without regard for frequency. These logical groupings were generally based on tasks that were performed in sequence.


The completed task model served as the basis for a conceptual design that allowed us to understand the natural partitions and task contexts within the app that made sense to users based on how they worked. This, in turn, allowed us to begin the visualization of the design, and the process of creating wireframes.

We were also able to provide input to the development process regarding which features had the most business value based on usage frequency and context. This information was used to determine which features should be developed in the early iterations versus which could be postponed until later ones.

The task model was seen by development as a type of roadmap or logical description of the application, which helped developers conceptualize what it was they were building.