Skavio Flow: choose XLSX, CSV or JSON after review

Skavio Flow turns invoices, orders, scans and text into structured outputs that teams can review before exporting as XLSX, CSV or JSON. The useful choice depends on what happens next: a finance colleague checking a workbook, an operations team importing rows, or an application consuming structured data. Start with the recipient’s requirements, check the extracted values against the source, then test the exported file in its destination.

1. Agree on what the receiving team needs

Ask the recipient for a sample import template or a short description of how they will use the file. “We use Excel” is not enough: they may want a workbook for manual review or a CSV that must match an accounting system’s import rules. Establish whether the handoff needs document-level fields, individual line items, or both. An invoice’s total and its item amounts serve different purposes; the recipient needs to know how those records relate. Flow supports custom fields and saved company workflows. Use those capabilities to make recurring extraction requirements consistent, while checking the actual export against the destination’s requirements. An export option does not imply a built-in ERP connector or automatic accounting posting.

2. Review the values before exporting

A valid file can still contain an incorrect invoice date, supplier reference or total. Flow provides source evidence and review warnings; use them to check important extracted fields against the original document before export. Review dates, currencies, totals and line items together. Compare item amounts with the document’s stated subtotal, accounting for any discounts, charges or rounding shown in the source. Check tax and gross totals separately rather than assuming that one correct figure validates the others. Treat an unresolved warning as a reason to investigate. If a due date or reference is absent from the source, agree with the recipient how to represent that absence rather than guessing. Changing the output format cannot resolve uncertainty in the underlying data.

3. Choose the format by its next use

| Format | Choose it when… | Check before handoff | | --- | --- | --- | | **XLSX** | Finance or operations needs a spreadsheet-based handoff. Flow’s XLSX output includes document and line-item worksheets. | Confirm that the recipient can use both sets of records and that identifiers, dates and amounts behave as expected. | | **CSV** | The destination requests a tabular import. | Inspect the actual columns and record layout, then test delimiter, encoding, quoting and identifier handling. | | **JSON** | An application expects structured data. | Inspect an actual Flow export and agree field names, structure, identifier strings, missing values and monetary representation. | For XLSX, distinguish the workbook from a later conversion to CSV. Microsoft documents that saving a workbook as CSV in Excel retains only the active worksheet’s displayed text and values—not the complete workbook or its formatting. That is [Excel’s conversion behavior](https://support.microsoft.com/en-us/excel/excel-formatting-and-features-that-are-not-transferred-to-other-file-formats), not a description of Flow’s CSV exporter. CSV also needs a compatibility test. [RFC 4180](https://www.rfc-editor.org/info/rfc4180/) documents consistent field counts and quoting conventions for commas, quotes and line breaks, but it is informational rather than a guarantee that every importer behaves identically. JSON supports objects, arrays and explicit value types. However, [RFC 8259](https://www.rfc-editor.org/info/rfc8259/) notes numeric interoperability limits. Agree how amounts and identifiers should be represented; do not assume that choosing JSON alone settles those decisions.

4. Test one invoice through the receiving software

Consider this illustrative invoice—not a customer result or a sample Flow schema: supplier code **00123**, invoice number **INV-2026-0042**, invoice date **2026-10-02**, currency **EUR**, and two line-item amounts of **120.00** and **80.00**. The document states a net total of **200.00**, tax of **40.00** and a gross total of **240.00**. After reviewing those values against the source, export the intended format and load it through the recipient’s normal process. Do not stop at checking the downloaded file: verify what the destination actually received. The supplier code is a useful test. Excel can remove leading zeros and limits numeric precision to 15 significant digits. For CSV imports, [Microsoft’s documented approach](https://support.microsoft.com/en-us/excel/keeping-leading-zeros-and-large-numbers) is to use Data → From Text/CSV and set identifier columns to Text before loading. This is a manual handoff test, not an automatic Flow validation feature.

5. Release the file with context, not assumptions

Record who reviewed the data, which documents the export covers and any unresolved exceptions. Make the source documents and relevant review evidence available alongside the deliverable through your agreed sharing process. Flow provides source evidence and review warnings, but do not assume they are embedded in every XLSX, CSV or JSON file. Add a separate security check when externally supplied text will be opened in a spreadsheet. Untrusted CSV content can be interpreted as formulas. [OWASP warns](https://community.owasp.org/attacks/CSV_Injection) that ordinary quoting or escaping is not universally sufficient, particularly after saving and reopening in Excel. Follow the receiving organisation’s formula-handling policy; a successful syntax check is not a security clearance. The handoff is complete when the recipient can use the reviewed values as intended—not simply when an export downloads successfully.

Explore Skavio Flow’s review and export workflow