Data Access Request
Redesigned the data access workflow to streamline and simplify access to data assets for data scientists at the Bank of Montreal.
Agile
UX Design
Enterprise Software
Partnered With
Data Lifecycle Lead
Product Owner
ServiceNow Developer
Software Developers
QA Analyst
I Led
Concept Testing
Prototyping
User Interviews
Wireframing
Tools
Figma
FigJam​
Timeline
4 Months
Context - Mary's Journey through the Data Ecosystem
This case study is on the "Access" phase which consisted of Requesting Access and Tracking Requests.
Approach
UX Design Process used by DnA Marketplace team
I started with the users and then worked my way backwards
Conducted 7 in-depth, 60-minute user interviews with Data Scientists to explore how they request data access and the obstacles they face.
User Interview in progress
Notes in FigJam
What Data Scientists (Users) Shared?
Tracking request is painful
70% said that the current request tracking experience lacked visibility into key information and status granularity.
Requesting access is time-consuming
The request form consumed a lot of time due to repeated questions, an overwhelming layout, and a lack of intuitive interactions.
User satisfaction is lacking
The overall SEQ score of 3/5 indicated that data scientists found the current experience neither easy nor difficult.
Need for additional actions
There was a desire to duplicate, edit, and cancel a request once submitted.
Problem Statement
"How might we make the data access request process efficient, easily learnable, and keep the data scientists informed so that they can seamlessly access the data assets?"
Early Design Decisions
Single vs Multi Step Form?
Divided the process into multiple steps to help data scientists focus and execute with confidence.
Vertical vs Horizontal Stepper?
Utilized the horizontal stepper wizard to optimize the space and followed the company's design guidelines.
One vs Two Column Layout?
Decided to use the one column layout given that there were maximum 5 input fields per step.
Screen from the data access request form
Adapted When Needed
Limited requests made for others
Analysis revealed minimal use of “submitted for me,” leading me to deprioritize it after team consultation.
Dropped the idea of progress bar
The Data Lifecycle Lead confirmed that, due to status granularity, a progress bar was not technically feasible.
My first proposal for Request Tracker which was tested with users and reviewed with the partnering teams
Behind The Scenes - Backend Process
Each request has three states/stages
Tested Proposed Solution
Ran concept testing with 5 Data Scientists to test the initial proposal.
Testing the initial version with a participant
Key Steps In Data Access Request Journey
Participant
Role Details
ID Details
Participant #1
Participant #2
Participant #3
Project Details
Justification
Tracking
Participant #4
Participant #5
Legends
User Delight
Concerns Expressed
Design Evolution - Final Output
Flow 1: Data Access Request Form
One step at a time
Created a multi-step form starting with the most critical inputs.
In the moment critical information
Added key role details (description, included data assets, owner) to help data scientists make informed, confident selections.
Access for everyone
Introduced a "Business as Usual" option for day-to-day data needs.
Reduced Redundancy
Streamlined justification inputs by merging them into one field.
Flow 2: Request Tracker
Enhanced Visibility
Displayed asset (schema) names in the tracker to help data scientists quickly locate and follow their requests.
.png)
Access to key information
Surfaced key information, such as Current Assignee, Current State, and Estimated Delivery, to keep data scientists informed and in control.
Impact
20%
Increase in Product Adoption, driven from the increase in unique visitors
28%
Increase in User Satisfaction, derived from the increase in the SEQ score
100%
User Acceptance, derived from the Office Hours conversations and Demos after release
Future Considerations
-
Introduce additional actions like edit, cancel, and duplicate the requests
-
Create a summary view of the request form
-
Add all desired filters to the request tracker landing page
Challenges
-
Navigating the limited data ecosystem knowledge
-
Advocating for proper participant selection
-
Interaction limitations with the Stepper Wizard in the design systems
-
Funding cuts significantly impacted the overall project scope
-
High complexity in the backend request approval processes
-
Difficulties integrating with ServiceNow workflows
Learnings
-
SEQ rating is generally calculated on a scale of 7
-
When and why to choose a multi-step form over a single-page form
-
​Always test, the first idea isn’t always the right one