Detect structure and map columns
Transparent by design
How local bank CSV analysis works
The application performs preflight detection, asks for confirmation when needed, analyzes in a Web Worker and renders results without sending transaction contents to Sparkpond.
Local processingThis file is processed in your browser. Its contents are not sent to Sparkpond.
Normalize dates and exact decimal amounts
Find duplicates, merchants and categories
Score recurring-payment hypotheses and explain them
Bank CSV↓Browser File API↓Local parsing worker
Local analysis engine↓Dashboard↓Optional IndexedDB / export
Financial contents do not leave this local flow
File → local preflight → mapping → parsing worker → canonical transactions → local analysis → dashboard → optional local save or export.
How local bank CSV analysis works
Why can mapping be required?
Bank exports vary widely. When headers, date order or debit direction remain ambiguous, asking you is safer than silently guessing.
Why are next dates shown as windows?
Weekends and billing behavior cause variation. An estimated window is more honest than an exact-date claim.
Start a private analysis
No bank connection. No account. Your transaction data stays on your device.
Open a bank CSV