Bank Statement Parsing

This article explains how Account Reconciliation parses bank statement files to extract transaction data for reconciliation matching.

What is Parsing?

Parsing is the process of reading and extracting structured data from uploaded bank statement files. When you upload a bank statement, the system automatically:

1. Identifies the file format (PDF, CSV, or Excel)
2. Locates transaction tables within the file
3. Extracts transaction details (dates, descriptions, amounts)
4. Extracts the closing balance (when available)
5. Prepares the data for AI-powered matching against GL transactions

The parsing process runs automatically when you upload bank statement files to a reconciliation. No manual configuration is required.

Supported File Types

Account Reconciliation supports the following bank statement formats:

Parsing Rules

When you upload bank statement files, the system follows these simple rules:

Rule 1: CSV/Excel files are always preferred over PDF files

CSV and Excel files contain structured data that the system can read directly and accurately. PDF files require additional processing to extract data from visual layouts.

Rule 2: Transactions are extracted from CSV/Excel when available

If you upload a CSV or Excel file, the system extracts transaction data from that file. PDF is used for transactions only when no CSV/Excel file is uploaded.

Rule 3: Closing balance source depends on what's available

• If CSV/Excel contains a closing balance → System uses it
• If CSV/Excel does not contain a closing balance but PDF is uploaded → System extracts closing balance from PDF
• If only CSV/Excel is uploaded without closing balance → You must enter closing balance manually

Rule 4: PDF serves different roles based on what else is uploaded

• PDF only → Primary source for all data
• PDF with CSV/Excel (CSV has closing balance) → PDF attached for reference only
• PDF with CSV/Excel (CSV has no closing balance) → PDF used for closing balance only

Parsing Workflow Scenarios

The following table explains how the system handles different file upload combinations:

Multi-Page PDF Parsing

Multi-page PDF parsing is the system's ability to process bank statements that span multiple pages within a single PDF document. Many bank statements, especially for accounts with high transaction volumes, extend across several pages. Account Reconciliation reads and processes all pages in a PDF file as a unified document, extracting transaction data regardless of which page it appears on.

When processing multi-page PDF files, the system:

• Analyzes all pages within the document
• Identifies transaction tables across pages
• Extracts transaction dates, descriptions, and amounts
• Consolidates data from all pages into a single transaction set
• Triggers AI matching against GL transactions

Best Practices for PDF Files

To ensure optimal PDF parsing results:

• Use electronically generated PDFs — These contain embedded text data that the system can read directly. Scanned documents may require OCR which can introduce errors.
• Avoid password-protected files — The system cannot access protected files and will return an error status.
• Match the statement period to the reconciliation period — Period mismatches can cause incorrect data alignment and parsing failures.
• Use original bank-generated PDFs — Third-party modified PDFs or statements exported from other applications may have non-standard formatting.