Overview
Detecting Transaction Errors Before the Audit
Designed a 0-1 product to automate a 3-day manual process that helped accountants detect transaction errors before auditors did, reducing compliance risk and human error.
ROLE
Senior Product Designer
IMPACT
50% Adoption
Automated 3 Days of Manual Review
Prevented Customer Churn
TEAM
Product Manager
Engineering Manager and Engineers
Senior Product Designer
COMPANY
FloQast
B2B SaaS for Accountants
Problem
Accountants spend days manually reviewing high volumes of transactions. Auditors still found mistakes months later.
Meanwhile, competitors had just released a similar product, putting customer retention at risk.
Systems Thinking
I influenced the team to build the product within the user’s existing workflow instead of a new dashboard, shipping faster and driving adoption.
Stakeholders initially wanted a new dashboard.
I pushed back with these insights:
Engineering effort for dashboard would delay launch.
User interviews confirmed they needed the solution inside their existing workflow, not a new page to navigate to.
After launch, users confirmed they did not want a new dashboard.
AI Interactions
I evolved the initial prototype with more approachable AI interactions for users with low tech fluency.
The initial prototype required users to detect anomalies only through Natural Language.
My research revealed
Some users have never used ChatGPT before
29% of users preferred dropdowns over Natural Language
Low adoption from other teams who released a Natural Language only approach
Solution
Integrated ERP Transactions, Detection Rules, and Anomaly Insights
Detect anomalies with suggested rules, natural language, and easy to use dropdowns
Filter through large datasets to streamline the review process
Drill into transaction details for anomaly insights and suggested fixes
Impact
Automated a 3-day manual process that helped accountants detect transaction errors before auditors did, reducing compliance risk and human error
50% Adoption
Customers reported increased confidence in their financial data
Prevented competitor churn
Reflection
What I'd Do Differently
Although I made the right call to descope the dashboard and prioritize approachable AI interactions, I could've improved the discoverability of the product. While some users were able to find product embedded in their existing workflow, others didn't. Making the entry point to this product more visible could've further improved adoption.