What formats can document data be extracted into (JSON, CSV, database fields)?

JSON is by far the most common output format for modern document data extraction, and for good reason. It's a structured, human-readable format that maps naturally onto the kind of labeled, nested data an extraction result typically produces, a set of named fields (name, date of birth, document number) potentially grouped into categories (personal details, document details), along with metadata like confidence scores or validation flags for each field.

Most APIs and SDKs in this space return JSON by default, since it's straightforward for developers to parse and integrate into almost any programming language or backend system.

CSV is less commonly used as a direct API response format, since its flat, tabular structure doesn't handle nested or variable-field data as gracefully as JSON does. Different document types extract different sets of fields, which is awkward to represent consistently in a fixed-column CSV format.

That said, CSV export is sometimes offered for batch processing scenarios or reporting purposes, where a business wants to review extracted data for many scans at once in a spreadsheet-friendly format, even if the underlying API communicates in JSON.

Direct database field mapping isn't really a separate "format" so much as an integration pattern. Once a business receives structured JSON data from an extraction API, its own backend code maps those JSON fields directly into the corresponding columns of its customer database, CRM, or compliance system.

This mapping step is typically something the business's own developers handle, informed by clear API documentation describing exactly what each returned field represents and how it's formatted.

Some vendors also support XML as an alternative structured format for businesses whose existing systems are built around it, though this has become less common as JSON has become the dominant standard across the software industry more broadly.

For most practical purposes, the format question comes down to how well an extraction result maps into whatever a business's downstream system expects, and how clearly that mapping is documented, more than any inherent superiority of one format over another for this particular use case.

ScanDoc returns extracted document data as structured JSON by default, with clearly labeled fields and metadata like confidence scores, designed to integrate directly into a business's existing backend systems, CRM, or compliance workflow without requiring additional reformatting on the business's end.