Filtering By Message Direction
Role:
Senior UX Designer
Collaborators:
UI Designer
Product Manager
Product:
Global Relay Archive
​Background
One of the initiatives that followed on from the False Positives Research was improving the message filtering capabilities, in particular, allowing users to filter by message direction (e.g., incoming, outgoing.)​
​
The brief
While this feature is already available, it was limited to email only. Modern financial institutions organisations have many communication channels they need to monitor, such as messaging clients, call recordings and chat rooms. In order to reduce the number of false positives compliance officers need to review, users need to be able to filter by incoming/outgoing/internal/external.
For example, a regulatory requirement might be to monitor messages sent from an internal employee to clients outside the organisation.​
Challenges
Challenge 1: Technical constraints​​
Not all message types could be filtered in the same way. For example, emails can be classified as incoming, outgoing, internal or external. Chat messages can only be classified as internal or external.
​​
​
Challenge 2: Communicating to users
The second challenge was how to communicate that not all directions applied to all communication channels. The design team were concerned about how users would understand this constraint and be able to combine the filters effectively to find what they were looking for.
Explorations V1
User testing: Round 1
After some quick design iterations we quickly moved to user testing to investigate the following questions:
-
Do these inputs address the user's needs? Can users filter their search in a way that is valuable?
-
Do users understand:
-
The naming of the options?
-
How can they be combined for different communication channels?
-
We tested with 6 users. Each test comprised of 2 tasks, testing two versions of the design.
Findings:
Overall users were able to complete the task and find the correct messages. However there were some nuances that caused confusion:
-
There were some questions about how the two inputs would interact and if there would be an any/all option similar to other filters.
-
There was some confusion about the unclassified option, but it was understood once users were given an example of what it would cover.
Almost all users read the helper text below the options and found it useful.
Design Review
Explorations V2
User testing: Round 2
Since there was uncertainty around how the inputs functioned, we chose to reevaluate the way the options were displayed to ensure we had a clear and understandable design. Another round of user testing was needed to compare the new design with the previous version and validate if it was more understandable to users. For the test, we compared the first and second versions, once again testing:
-
If users could complete the task and find the correct messages
-
If they understood the options presented and how they could be combined
For this and subsequent rounds of user testing we used a quicker, more guerrilla-style test, using members of other teams in the office (who were somewhat familiar with the software but not this specific project) so we could test and iterate the design quickly.
Findings: Inconclusive
Single input: As it was similar to the current feature, users acted fast with the single input, but this led to incorrect results. They liked having everything in one place, but the information was cluttered.
Two inputs: Users found this easier to digest but again questioned the relationship between the two inputs and if they would miss an option going between the two.
There was still some confusion about the unclassified option and how it should be used.
Design Review 2
From here we explored moving the options to a dialog to build on the findings of the user testing:
-
keep all the options in one place
-
allow for more space for helper text and to show what options applied to specific communication channels.
At this time, we moved to UI explorations to better understand how this would fit with the interface and current components in the design system.
Dialog explorations
User testing: Round 3
The aim of this round of user testing was to validate the dialog approach and ensure users understood the unclassified option.
Findings
Users were able to complete the task and understood what message would be returned with their selected options. The rephrasing of the ‘unclassified’ option and helper text was more understandable.
Subsequent rounds of testing
After a review with stakeholders, there were two further rounds of quick testing to improve how the options were displayed and make some minor UI improvements.
Outcome
Reflecting on the process as a whole, it was clear that the multiple rounds of user testing were instrumental in gaining deep insights into how customers interacted with and understood the filters. This iterative process uncovered critical usability issues, clarified areas of confusion, and revealed improvement opportunities that were not previously identified internally. It enabled us to uncover and eliminate potential user errors that, if left unaddressed, could have led to serious consequences. The feature was first released to a limited group of customers to validate the design and value to customers. After a thorough validation phase and positive user feedback, we were able to confidently roll out the feature to all customers, allowing them to achieve their goal of eliminating false positives and creating a more targeted set of results.
