Eliminating Bank Disconnections: A technical UX solution that cut costs by 66%

How smart duplicate prevention solved customer churn while saving $1,000+ monthly
  • 87% reduction in duplicate connection attempts (8 → 1 monthly events)

  • 66% reduction in total Plaid costs ($1,493 → $510 monthly)

  • 93% reduction in balance API calls (12,918 → 800 monthly)

The Crisis

Fincent faced a critical retention problem: 56.2 monthly bank disconnections were driving customer churn. Customer Success identified frequent disconnections as a primary reason customers left the platform.

Root cause analysis revealed:

  • 1,128 duplicate bank connections cluttering the system

  • 2.79 average duplicates per company

  • Major institutions (Chase, Bank of America, American Express) most affected

  • $1,493 monthly Plaid API costs from unnecessary balance calls

The Technical Challenge
Plaid API Constraints:
  • Duplicate connections had to be managed at Plaid's end, not just our database

  • Users could inadvertently create multiple connections for the same institution

  • No native duplicate prevention in existing flow

  • Balance API calls ($0.08 each) triggered on every transaction regardless of user activity

Business Impact:
  • Customer frustration with cluttered account lists

  • Support overhead from connection issues

  • Unsustainable operational costs scaling with user growth

The Solution
Dual-pronged approach addressing UX and technical efficiency:

1. Smart Duplicate Prevention

  • Added connection detection before account setup

  • Implemented warning dialogs to guide users to existing connections

  • Created account deletion flow with proper Plaid cleanup

2. API Cost Optimization

  • Changed balance fetching from "every transaction" to "user login only"

  • Reduced API calls from 12,918/month to ~800/month

  • Added bookkeeper-specific balance refresh controls

Implementation
Timeline: Sept 2024 → Dec 2024 (3 months)

Key Features:

  • Account selection interface preventing duplicates

  • Confirmation dialogs for account disconnection

  • Optimized balance fetching triggered by user activity, not transactions

Result
Customer Experience Impact
  • 87% reduction in duplicate connection attempts (8 → 1 monthly events)

  • Eliminated customer churn from disconnection issues

  • Streamlined bank account management experience

Cost Impact
  • 66% reduction in total Plaid costs ($1,493 → $510 monthly)

  • 93% reduction in balance API calls (12,918 → 800 monthly)

  • $12,000+ annual savings in operational costs

Technical Impact
  • 1,128 duplicate connections cleaned up

  • Sustainable scaling of Plaid integration costs

  • Improved system reliability with proper connection management

Key Insights
  • Low-effort, high-impact intervention: Simple UX friction points prevented expensive technical problems downstream.

  • User education over restriction: Warning dialogs informed users without blocking legitimate use cases.

  • Operational efficiency: Solving the UX problem simultaneously addressed a major cost center.

Conclusion

This project demonstrates how thoughtful UX intervention can solve multiple business problems simultaneously. By understanding the root cause of customer frustration, we eliminated a major churn driver while cutting operational costs by two-thirds.

The result: Happier customers, cleaner data, and sustainable unit economics.

Project completed Dec 2024 | Impact measured through May 2025