Conversica Training Desk 2.0
On this project, we were tasked with an opportunity to redesign the Training Desk 2.0 experience for Conversica. The Training Desk 2.0 was a web app tool that Conversica’s “trainers” utilized in order to train their AI to provide more accurate leads and help teams take action on interest.
Define
Firstly, we took a look at the current tool. A trainer gave us a demo of the tool to help us understand how the trainers actually interacted with the tool on their day-to-day basis.
Their current output was 200 training sessions which were done by 10 experienced trainers with an average of 3,000 messages per day. This meant that each trainer was expected to go through approximately 300 messages a day. These were staggering numbers. Imagine yourself going through 300 emails per day while making sure to select an accurate set of actions needed based on the content of each email.
With these numbers, it was not surprising to hear that
30% of these training were sent to audits, including peer review due to trainer mistakes, translation issues or guidelines errors.
From this, we were able to gain obvious clues on areas that needed the most attention. We also completed a heuristics evaluation which also included a detailed audit and analysis of the tool to identify any usability challenges in the current application to further solidify our assumptions.
A. Visibility of system status
Lack of indication on where the user is at. Number meaning is unclear.
B. Flexibility and efficiency of use
Inefficient UX system for filtering and listing.
C. Aesthetic & Minimalist Design
Overall, the tool lacks visual interest and hierarchy. There is no visual difference between each type of content.
D. Recognition rather than recall
Lacks contextual information that helps users take action quickly.
E. Flexibility and efficiency of use
Checkbox is not the right paradigm for this feature. It’s helpful for users to know what the language is being translated to
F. Consistency & standards
There’s a disconnect between how training desk notes and guidelines are displayed.
Key Findings
Simplify the experience and provide clear navigation
Users expect a tool that is easy to use and doesn’t make them work too hard. Streamlining how the information is displayed and providing clear navigation indications ensure that users are equipped for their task at hand.Visual design system can be further improved
The visual design system can be improved to make certain elements more identifiable, bring prominence to important information, and make a simple task-flow be more efficient for its users. The visual design system should also support the contextual information to help users take less time to make decisions.
Hypothesis
By improving the user experience of the tool, we believed we can improve the speed and efficiency of training
Reduced scrolling will reduce time and increase accuracy.
Organized intents will result in faster and more accurate identification
Notes, guidelines, and conversation history will be read and understood more often, decreasing mistakes and improving efficiency
Reduced mouse movement using right-clicks or dropdowns can help trainers work faster
Ideate & Validate
From the findings and understanding of how the tool was used, we then created two redesigned options of the tool. The design recommendations were based on these three topics that would help with efficiency and accuracy of the trainers:
Improved visibility of system status: navigation. This helped users track where they were at their tasks quickly
Created visual hierarchy for better legibility and reduced visual clutter as required in high-speed and demanding tasks
Let users rely on recognition rather than recall. This was to allow users to see what was in front of them easily instead of having to rely on their memory.
We created prototypes for these two options and then tested those designs with Trainers. We conducted five 30-minute qualitative user testing interviews with Trainers of varied levels of experience.
Iterate
There were many insights and suggestions gained from the interview participants. Of the two options, the second option was preferred by the trainers. Some feedback mentioned that the layout of the second option was clear and helped them to become more efficient in their tasks. There were some noteworthy suggestions to increase color contrast and allow for collapsible modules.
Based on these results, we then iterated and finalized on our design direction.
