Skavio Flow: one batch of supplier PDFs into reviewed line-item spreadsheets

Skavio Flow turns supplier documents into structured exports, including an XLSX workbook with document data and line-item worksheets. The important step is not simply getting PDFs into Excel: it is checking that every row still means what the supplier wrote. For purchasing and bookkeeping teams, a useful workflow starts with a small batch, uses consistent fields and keeps unresolved documents out of the approved accounting handoff.

1. Prepare a batch you can check document by document

For this fictional exercise, prepare three supplier PDFs: a one-page digital invoice in EUR, a two-page scanned invoice in EUR and a one-page digital invoice in USD. This gives you different input types without making the first review unwieldy. Before uploading, open each PDF and check that all pages are present, readable and correctly oriented. Keep an intake list with the filename, expected page count and a reference your team can use during review. Keep the original files unchanged. Flow reads embedded PDF text locally; scanned documents go through OCR before structured extraction. Both routes produce draft data that needs checking. A completed extraction does not establish that the source document was complete.

2. Define document fields separately from item fields

Use custom fields to distinguish information that belongs to the whole invoice from information repeated on each line. The schema below is a recommendation, not a promise about Flow’s default column names. Give the document reference a clear purpose: it should let a reviewer associate an item row with the correct invoice. After export, confirm that the available references preserve that relationship. Do not rely on the invoice number alone to identify a document across different suppliers. Save the field structure as a company workflow for later batches. Reuse the definitions—not values copied from an earlier invoice.

3. Check the estimate, then treat the result as a draft

Review the processing estimate before submitting the batch. The server quote at submission is authoritative for the job price. Check the current interface for limits and any trial eligibility rather than assuming a particular allowance. Once processing finishes, compare the results with your intake list. Account for each document and check whether multi-page invoices have retained all their item rows. Flow provides source evidence and review warnings, including flags for missing or ambiguous fields. Use these to direct attention, but do not treat the absence of a warning as confirmation that every value is correct.

4. Reconcile each invoice before considering batch totals

Review identifiers, dates and currency first. Then compare every extracted item with the original: description, quantity, unit, price and line amount. Pay particular attention to rows split across pages or descriptions that wrap onto several lines. Next, reconcile the document’s totals. OpenPeppol’s Peppol BIS Billing framework distinguishes line net amounts, document-level allowances and charges, VAT, prepayments and rounding. Those distinctions explain why adding line amounts does not necessarily produce the amount payable. Use this as a review framework—not as a claim that Flow performs Peppol validation or automatic reconciliation. In a fictional EUR invoice, line net amounts total €180. A €10 document allowance and a €5 charge produce a net total of €175. Adding the invoice’s stated €35 VAT gives an invoice total of €210; a €50 prepayment leaves €160 payable, assuming no rounding adjustment. This illustrates the arithmetic, not a determination of the correct tax treatment. Comparing €180 directly with €160 would miss the intervening amounts. Keep the USD invoice separate from the EUR invoices when calculating totals. Do not combine their amounts into an unlabeled batch total or invent an exchange rate.

5. Export, then check what the receiving spreadsheet actually shows

Choose XLSX when reviewers need document and line-item worksheets together. CSV is useful for a defined table handoff, while JSON suits structured downstream processing. In every case, inspect the downloaded output rather than assuming that a completed export is ready for import. For CSV, Microsoft documents that opening a file directly in Excel uses default format settings, which can reinterpret dates and identifiers. Use a controlled Text/CSV import and explicitly set invoice numbers and product codes to text where necessary. An identifier such as 000417 should not silently become 417. Confirm the expected documents and item rows, column meanings, currencies and number formats. Retain the original PDFs and the reviewed export under your team’s document policy. The result is a checked data handoff; exporting it does not constitute accounting approval or automatic ERP posting.

Explore the sample, then prepare your own review workflow in Skavio Flow →