LedgerLens Field Notes: The Best Reconciliation Tool Is a Spreadsheet

Short Field Note — LedgerLens Recon

---

I built LedgerLens Recon after watching a finance team struggle with a "modern" reconciliation platform. It had dashboards, real-time updates, and a React frontend. The finance team exported everything to Excel anyway.

Three lessons:

1. The output format determines the tool's adoption.

A JSON API is useless to a finance manager. A terminal table is slightly better but still requires a developer to interpret. An Excel file with color-coded rows is immediately usable. The finance team doesn't need training. They open the file, see red rows, and start investigating. The tool's success is determined by its output, not its architecture.

2. Mock modes are essential for trust.

Every external dependency — Stripe, the database, even the CSV parser — has a mock mode. This means the tool is testable and demoable with zero setup. A prospect can run it in 30 seconds. A developer can debug it without API keys. A CI pipeline can validate it without network access. Mock modes feel like overhead until they're the reason you can ship confidently.

3. Deterministic output makes debugging possible.

We enforce stable ordering and consistent formatting so that reruns are diffable. If a bug changes the match result for transaction #12345, a diff of the two Excel files shows exactly what changed. Without deterministic output, debugging becomes a forensic exercise — comparing two reports that look similar but aren't directly comparable.

---

LedgerLens Recon is open source at [github.com/Gwerdonatus/Ledgerlens-recon](https://github.com/Gwerdonatus/Ledgerlens-recon)