- ROLE
Head Researcher
- FOR
Global businesses and international banks.
- DATE
2018-2019
Feedzai is a company that uses machine learning algorithms to detect and prevent fraud. Three products work together in order to make this happen.
The initial design team working on improving these products by doing research and creating improvements consisted of myself as the Lead Researcher, two UX Designers, and 1 UI Designer.
BACKGROUND
The project shown here consists of a contextual inquiry (for all three products that I am not allowed to show fully due to an NDA) and a semi-structured interview (on a product called: Genome, where you can see the potential improvements I’ve created further below) in order to figure out the following:
What are the users (currently) doing? What are their processes and workflows?
What do they expect to see in the new features and products? Why?
Does the navigation make sense? Why or not?
Are the current and potential features/products on the path to being intuitive? Why or why not?
Fraud Detection Trio refers to the three products this company sells that protects and detects fraudulent user data.
THE RESEARCH PLAN
One of my first assignments was to set a baseline with Nike as our main customer. We were testing our old UI called Fraud Detect (a simple case management tool that was heavily customized by the user in order to find patterns) against our new UI called Genome (a link analysis tool that uses machine learning algorithms to find patterns for the user).
“QUAL” stands for qualitative data. The others are quantitative.
The personas that we discovered. This is a way – in UX – to gain empathy for the user. Users buy stories – not products.
RESEARCH RESULTS
Fraud Detect (the old UI) usability test results versus Genome (the new UI) test results below.
.
SOLUTION
Below are some screenshots of what the new product looks like based on our first round of testing.
The cluster of nodes are all attributes of any transaction. Name, credit card, IP address, etc.
Genome takes this information and finds patterns in terms of when the user signed into Nike’s system, when the user used a credit card, or where they accessed the account from.
If something seems suspicious (such as: the same credit card being used in random places throughout the world or credit cards being used in dangerous geographical zones) then Genome would let the user know.
FINAL THOUGHTS
Below you can see what we learned and what we thought we could improve on.
TAKEAWAY
The product was so well-received that other organizations bought and used the product heavily because our platform and algorithm was helpful. We did multiple rounds of maintenance testing after this in order to improve the UI.