top of page
Amazon Revolutionizes Financial Close with MARS
Amazon redefines the financial close experience by automating account reconciliations through AI and LLM, enabling a faster financial close and improving accuracy in consolidated financial statements
To comply with my non-disclosure agreement, I have omitted and obfuscated confidential information in this case study. The information in this case study is my own and does not necessarily reflect the views of Amazon.
Results
After launching the MLP of product, we received significant positive feedback from our financial and accounting customers. The AI-powered account reconciliation product MARS automated a significant part of tedious manual work. It has reduced 67% of their manual work time just for the MLP launch, and automated 2 times of accounts reconciliations.
2X
Automation of reconciliation process
65%
Reduction of working time on monthly reconciliation
80%+
Reduction of manual work
90%+
Positive feedback from customers
Context
Challenge
The Amazon finance and accounting team manages trillions of transactions monthly across various countries and regions such as NA, EMEA, APEC, LATAM, ensuring accurate and healthy month-end close through reconciliations and other accounting activities
Prior to implementing the product, account reconciliations at Amazon were critical but highly manual, time-intensive, and operationally unsustainable. Finance and accounting teams spent an astonishing hundreds of thousands of hours annually on this laborious process. Additionally, due to the diverse nature of Amazon's businesses, the reconciliation process was complex and required strict adherence to tight timelines. So the goal of this product is to design a scalable, seamless account reconciliation experience.
My Role
I led the end-to-end user research and design. I collaborated with cross-org teams including engineering teams, data scientists, product managers, and program managers. The project started from July, 2023 and we launched the MLP in June, 2024.
User Research & Discovery. I conducted the user research, pictured the persona, user journey map and research report and information architecture.
Product Strategy. I facilitated three cross-org workshops to discuss and align on the UX and product strategies between teams, and came up with the long-term and short-term strategies.
Design & Validation. I created frameworks and prototypes to share the vision, design principles and content strategy. This helped to evangelise ideas, gain alignment and drive decision making.
User research
Key Findings
I conducted 8 in-depth interviews spanning 4 countries and regions to collect the initial insights from end-users. Here are some key findings from the research.
-
Users have to fetch data from up to 30+ tools, which is manual and time consuming
-
Users have to manually prepare the data format to ensure compatibility for matching
-
Users have to manually match the two datasets in Excel and identify unmatched items
-
Users need to switch between tools to download and upload files for approval
-
Users need to communicate and collaborate with others on multiple communication tools regarding unresolved items
Product Strategy
I faciliated three cross-org workshops together with product managers, engineering teams and key stakeholders to define the product strategy together.
-
How might we simplify the data fetching process for finance and accounting users
-
How might we simplify the process of preparing account reconciliation
-
How might we help users to detect the variance between two datasets
-
How might we help users to investigate and explain the variance
-
How might we provide users a seamless approval and review experience
.jpg)
End-to-end User Journey
After the user research and workshops, I draw the current state end-to-end user journey and the future state user journey.
Vision
The goal of MARS is to accelerate and automate the book closing process. To integrate all upstream data and automatically ingest financial transactions, and flagging variances for manual review. By analyzing past reconciliation data, MARS attempts to explain and recommend the required rectification. This eliminates the need for accountants to manually assemble datasets, create matching rules to investigate variances, prepare reconciliation documentation and manually certify transactions.
Design solution
Personalized Dashboard
Persona based personalized dashboard for the different roles. The dashboard provide users a starting point with relevant tasks and actionable insights, the view of accounts they're responsible for, their reconciliation status, auto-reconciled accounts with supporting documents by system, and steps taken for account that couldn't be auto-reconciled, along with recommended next steps.
AI Generated Insights and Recommendations
Instead of users manually preparing the process, MARS will automate most of the reconciliations systematically. It will highlight unmatched reconciliations for users' attention and provide insights into potential issues, along with recommendations for them to take action.
AI Chatbot
AI chatbot that answer questions about the financial data being reconciled and provide month-over-month trends and anomalous transactions
© 2024 BY KRISTEN
bottom of page