Building a unified fraud detection & management system
![](https://images.squarespace-cdn.com/content/v1/5f1cbb90b0a3be4aac9ea7a1/91b5051e-90d9-44e5-988b-48826aa0bf83/Frame+126.png)
Overview
I designed a scalable fraud management tool for tens of thousands of AWS customers.
Organization
Amazon Web Services
Project Type
Internship — Enterprise — Web
My Role
Design lead - emphasis on interaction design
Tools I Used
Balsamiq Mockups, Sketch, Paper + Pencil
Team 🙏
Brendan Lee, Design Product Manager
Tyler Anderson, UX Designer, Mentor
Nick Tostenrude, Product Manager
Idon Shaman, UX Designer
Aditya Tripathi, Business intelligence Engineer
Erin Carlson, Product Marketing Manager
Me
Problem
Picture yourself as the CEO of a well-known startup company that has been growing exponentially in the recent months.
Unfortunately, your company is outgrowing your fraud management solution and you are using too many fraud prevention vendors to protect your clients and gaps in your solution are being taken advantage of by fraudsters.
Their private information, including their identities and credit card information, is in danger. You are in risk of losing your customers and tens of millions of dollars in revenue and bringing damages to your reputation.
Solution
Based on user metrics of existing solutions in the fraud prevention space, I designed a unified fraud risk evaluation experience for professional to perform risk evaluations with Amazon’s data and expertise.
The main design goal is to unify fraud management and allow them to track different types of fraud within one application.
User Benefits
With high adoption rates, customers who use other existing fraud prevention solutions will be much more content when using this solution, which equate to high customer retention and satisfaction rates.
Business Benefits
The released product aims to be established as one of the core AWS offerings. This will help establish its prominence in the fraud detection and prevention space with continually increasing revenue and market share for AWS as a business.
Released Product
As of July 2020, AWS has released the Fraud Detector to the general public. Much has changed after my efforts during the internship, but I am so glad to see that product I worked on has released!
Tens of billions of $ lost
in a $130bn+ industry.
+28%
New Account Fraud in 2019
Damage in countless companies’ reputation
& lost in customer trust.
Number of competitive vendors in the fraud prevention space
Part 1: Research🕵️
At the start of the internship, I jumped into this project after its research phase, where much of the research have already been completed. I was provided the research deliverables by my team.
Instead, I digested the existing research documents and reviewed them with stakeholders to learn about the main pain points that bogged down existing fraud prevention solutions in the market, as well as the types of customers who currently use these solutions.
The stakeholders included engineers, product managers, and business intelligence engineers.
I read through the Press Release FAQ (PRFAQ) and Business Requirements Document (BRD) to understand more about my project.
Personas
From reading through the business documents and talking to stakeholders, I gathered that companies have two main groups of users that need this type of fraud management system.
I also talked to a few of its users since they were also the stakeholders in AWS.
Technical users
Engineers, data scientists
Still prefer to write code when rule building, but a more cohesive user experience would help.
Non-technical users
Administrators, financial analysts, and business owners
Less tech-savvy. Need a less complex user interface for rule management.
Pain points & design opportunities
Pain points
CRUD (Create, read, update, delete) workflows feel disconnected to one another (i.e. creating and editing variables, rules, etc.)
Graphical user interface (GUI) rule builder is difficult to use, especially for non-programmers
Difficulty navigating through application lifecycle (i.e. moving an app version through different versions for testing purposes)
Design opportunities
Connect the entire fraud prevention experience into one seamless, integrated experience (first-time setup, creating and building variables, rules, etc., testing and promoting applications, and monitoring application performance)
Simplify and improve the usability of the rule builder GUI (select from templates, allow users to access the code any any time, etc.)
Simplify the application lifecycle and show the appropriate information during critical decisions (ex: promoting an app version to live)
Part 2: Ideation💡
With this information I synthesized from the existing research, I sketched out key flows of the experience and presented them to the Product Manager for the project.
At the same time, I created simple wireframes to visualize the flow of the actual experience and reviewed them with my design team.
GUI Rule Builder - sketching out possible options when adding another condition
Initial wireframe of rule builder page on Basalmiq
Edit mode for rules - created to receive feedback regarding order controls and resizable columns
Initial wireframe of rule list page
Possible views for different application versions
Initial wireframe of application versions page
Part 3: High-fidelity designs ✏️
Following more feedback with the project team and my design team, I moved onto Sketch to create a mid-fidelity / hi-fidelity set of mockups of the key user flows of the service designed using AWS design guidelines.
Below are the some of the flows I can show. All mentions of Amazon and Amazon-related terminology are removed.
Pain point #1 resolved
Old: Difficulty navigating through application lifecycle
New: All four parts of the lifecycle is visible as tabs and the user can easily switch between the tabs to access any version.
Pain point resolved
Old: CRUD workflows feel disconnected to one another
New: CRUD workflows are now streamlined and follow design conventions at Amazon (i.e. wizards).
Pain point resolved
Old: GUI rule builder is difficult to use, especially for non-programmers
New: Rule builder is now more intuitive with a simple layout for all types of user
s
Handoff 👌
The biggest constraint that I faced was the limited availability of stakeholders for my project, which made it very challenging for me given an exceptionally complex project.
To make their lives easier, I created a site map of all of my screens with page numbers, so they can easily go through any updates I made, especially for after the internship, which saves them valuable time during their busy work day.
Prior to these design reviews, I send a copy of my mockups with annotations to everyone present in the meeting. The annotations discussed possible technical constraints, conditional items, and questions I may have for a specific page.
This is to give my stakeholders some time to look over my designs before going into the meeting.
Unfortunately, I was unable to test my designs with external users from other companies. I was told that this was outside of the scope of my internship, but I definitely would have preferred completing some tests myself.
Part 4: User Study for Key AWS Product
Because of the lack of research opportunities during the internship, I took an initiative to complete a user journey project for a key product in AWS to improve its user experience as a side project.
I completed a cognitive walkthrough of the current product, then interviewed five of the product's customers.
After analyzing and cleaning the notes I took, I created a user journey that details the key user flows of the product and customers' pain points, and opportunities for design improvements.
Part 5: Reflection 🧠
Understand context before designing.
This may sound obvious, but it was imperative for me to fully understand my project details before diving deep into design. The context of my project had so many more moving parts that I did not fully comprehend until weeks into my project. Understanding context from multiple perspectives (reading existing project documents, setting up meetings with stakeholders to better understand project details, etc.) made it easier for me to design and made my mockups more focused on solving customers’ pain points.
Set the roadmap of the project for myself.
As I approached the midpoint of my internship, I realized that I needed to determine the flows that I needed to design and ones that were in “Phase Two” of the project before my internship was over.
Given that my project was in multiple phases, I had to set clear expectations to my project manager in terms of the amount of designs and flows that I can create given the limited amount of my work hours. In addition, many of my stakeholders were busy during most of the day, so I had to take full advantage of each meeting and earn their trust throughout the internship.
Always question my design decisions.
After completing this internship, I realized that asking myself "Why does this need to exist?" or "What purpose does this serve?" is a valuable skill to have. Sometimes my design could be instantly improved by removing the unnecessary parts and simplifying the existing design. This made it easier for me to disagree and commit to my design decisions in front of key stakeholders.
User experience applies to all types of products.
No matter what the product is, or the audience of the product is, I must think about users' experience from their perspective. Before I mainly thought of users as everyday consumers, but in the case of AWS, the customers can be highly specialized and technical individuals whom I do not always think of as your typical users. Having empathy for the customer, therefore, was extremely important here.
Special Thanks 🙏
Brendan Lee, Design Product Manager
Tyler Anderson, UX Designer, Mentor
Nick Tostenrude, Internship Project Manager
Idon Shaman, UX Designer
Erin Carlson, Product Marketing Manager