Choose the right document type

When you create a parser, you pick a Document Processing Type (e.g. Invoice, Purchase order, Bank statement, Custom). Use the type that best matches your documents. That tells DigiParser what to look for and improves accuracy.

If you use Custom document, DigiParser will try to detect fields automatically. For invoices, receipts, or bank statements, the specific types usually work better.

Choosing Document Type goes into this in more detail.

Add clear, specific fields

In Fields & Tables, use field names that describe what you want (e.g. “Invoice Number”, “Total Amount”, “Vendor Name”) rather than vague labels. Names that match how the data appears in your documents tend to work best.

Use the right field type (Plain text, Number, Date, Table, etc.) for each piece of data. For repeating information like line items, use a table with clear column names.

For each field, write a short, crystal-clear description that explains exactly what should be extracted and where it usually appears in the document. The clearer and more specific these descriptions are, the easier it is for the AI to find the right data and avoid confusion between similar fields.

Choose the right extraction mode

In Parser Settings → Parsing Configuration, choose the Data extraction mode that matches how critical accuracy is for your workflow:

• Fast – Prioritizes speed; good for quick checks or less critical use cases.
• Accurate – Balances speed and accuracy; a good default for most production workflows.
• Critical – Uses more credits per page for maximum accuracy when precision is essential.

Picking the most relevant extraction mode for your documents can significantly improve extraction quality, especially for complex layouts or high-stakes data.

Use good-quality documents

• Clear scans or PDFs: Avoid blurry or low-resolution files. Text should be readable.
• Straight, legible layout: Crooked or heavily annotated pages can slow extraction or reduce accuracy.
• Consistent layout: When similar documents look alike (e.g. same invoice layout), results are more consistent.

Review and correct when it matters

For important workflows, review extracted data before you export or send it to Xero, Google Sheets, etc. Use the document view to fix mistakes, and Reviews & Approvals if you work in a team.

Re-process documents after you change fields or settings so the data matches your current setup.

Use Post Processing for consistency

Post Processing can clean and standardize data after extraction (e.g. format dates, trim spaces, match vendor names to IDs with lookup tables). That improves consistency even when the raw extraction varies a bit.

Use one parser per document “type”

If you process several kinds of documents (e.g. invoices vs. receipts vs. bank statements), use separate parsers for each. Each parser can have the right document type and fields, which keeps accuracy higher than one parser for everything.