How to Extract French Delivery Note
Data into Excel (2026 Guide)

Why the Delivery Note Is the Bottleneck French AP Teams Overlook

The invoice gets the automation budget. OCR tools, AP software, e-invoicing mandates — the entire facture (invoice) ecosystem is being rebuilt for September 2026, when every French business must be capable of receiving electronic invoices through a certified platform. The bon de livraison (delivery note) gets a clipboard and a biro. Yet the delivery note is the document that gatekeeps the entire receiving-to-payment pipeline.

Under standard French procurement practice, payment flows through a three-way verification: the bon de commande (purchase order) sets what was ordered, the bon de livraison (delivery note) confirms what was delivered, and the facture (invoice) states what must be paid. If the delivery note data never enters a structured system, the three-way match stalls before it starts. Accounts payable must either chase the warehouse for delivery confirmations, manually type line items from scanned PDFs, or — most commonly — skip the delivery verification and hope the invoice matches the order.

This is where the extraction approach differs fundamentally from invoice tools. An invoice tool expects a total, a tax breakdown, and a supplier ID — fields that are legally standardized across French invoices under Article L441-9 of the Code de commerce. A delivery note, by contrast, carries whatever the supplier's logistics system generates: sometimes a line-item table with SKUs and quantities, sometimes a product description and a box count, sometimes nothing but a tracking barcode and a driver's handwritten delivery time. The extraction tool needs to handle all of them — and output columns that make sense downstream.

What Makes a French Bon de Livraison Different from a Standard Packing Slip

Unlike an invoice, the bon de livraison (delivery note) is not legally required in France. The Code de la consommation Articles L216-1 through L216-6 and Code civil Article 1610 establish the seller's delivery obligation and the buyer's right to receive conforming goods, but they do not prescribe a specific document format. The bon de livraison is a commercial practice, not a legal requirement — which means it has almost infinite format variability. Yet three characteristics make French delivery notes distinct from generic packing slips in ways that matter for data extraction.

No pricing. Under French commercial convention, pricing belongs on the facture (invoice). The delivery note lists what was delivered — product references, descriptions, quantities, units of measure — but does not show unit prices or totals. For the receiving team, this is correct. For an extraction tool trained on invoices, this is a problem: the tool looks for a "Total" field that does not exist. Columns must be defined around the fields that actually appear on the document, and any price matching happens later — in the spreadsheet, where line items from the delivery note are compared against the purchase order and the invoice.

Company identification through SIREN and RCS. Articles R123-237 and R123-238 of the Code de commerce require every registered French business to display its numéro SIREN (the 9-digit unique company identifier in the Sirene register) and its RCS registration (Trade and Companies Register, followed by the city of the greffe where it is registered) on commercial documents — including delivery notes. Most supplier delivery notes therefore carry a SIREN in the header or footer. Extracting this field correctly links the delivery note to the supplier master data in the ERP, enabling automated matching against open purchase orders from the same supplier.

Signature as legal evidence. While not legally mandatory, a signed delivery note creates strong evidence of receipt. Under Article L133-3 of the Code de commerce, any réserves (reservations — notes about damaged packaging, missing items, or delayed delivery) written on the delivery note at the time of receipt must be confirmed by registered letter within three days. A clean, signed delivery note without reservations is présomption de livraison conforme (presumption of conforming delivery). For extraction purposes, this means your tool needs to preserve not just the structured fields (SKU, quantity) but also the unstructured annotations — the handwritten shortage note, the circled item count, the driver's scribbled delivery time — because those annotations carry legal weight in a dispute.

The Fields That Matter: From Delivery Note to Three-Way Match

Extraction column design for delivery notes starts with a different question than invoice extraction. With an invoice, you're extracting what the supplier demands to be paid. With a delivery note, you're extracting what the supplier actually delivered — and the output needs to feed into a verification process, not a payment instruction. Here is the field structure that supports a French three-way match (rapprochement bon de commande / bon de livraison / facture):

* For goods receipt accounting under the Plan Comptable Général (PCG, the French national chart of accounts): Debit the purchase account (compte 607 — Achats de marchandises for resale goods, or 601/602 for raw materials and consumables) and debit compte 44566 (TVA déductible — deductible VAT) with the VAT amount; credit compte 401 (Fournisseurs — suppliers). If goods are received but the invoice has not yet arrived at month-end, use compte 408 (Fournisseurs — Factures non parvenues) for the provisional accrual. Delivery note data provides the quantities and references that determine which amounts to post.

This field structure produces a spreadsheet where each row is one line item from one delivery note, with header fields — supplier, delivery note number, date, PO reference — repeated on every row. A three-way match in Excel becomes a series of VLOOKUPs or INDEX/MATCH operations: the delivery note quantity is checked against the PO quantity, the invoice quantity is checked against the delivery note quantity, and discrepancies are flagged per line.

But building that spreadsheet manually — typing 12 line items from a Sage supplier's delivery note, then another 8 from a different-looking Geodis form, then transcribing handwritten shortage notes from a courier's carbon copy — is what makes delivery note processing a bottleneck. The extraction step is what turns this from a data entry task into a data verification task. And for that to work, the extraction tool needs to understand what each column means, not just where it sits on a page.

Step-by-Step: Extracting Delivery Note Data into Structured Excel Columns

Template-based OCR tools treat a delivery note like a grid. You upload one example from supplier A, draw rectangles around the quantity field and the SKU field, label them, and save a template. The next time supplier A sends a delivery note with the same layout, extraction works. When supplier A updates their ERP and the layout shifts — or when supplier B's delivery note arrives in a completely different format — the template breaks. For a French warehouse receiving goods from 20, 40, or 100 different suppliers, template maintenance becomes a full-time job. The approach that works across all formats is Custom Column Extraction: instead of telling the tool where a field sits, you tell it what the field means, and the AI locates the value by understanding the document's content — by reading it.

Here is the workflow, from supplier delivery note to structured Excel:

This extraction workflow removes the template-maintenance burden entirely. The same column definition — "Supplier SIREN," "PO Number," "Quantity Delivered," "Receipt Notes" — produces consistent output regardless of whether the delivery note came from a Sage-printed PDF, a Divalto WMS-generated form, or a driver's carbon-copy book. The AI reads each document fresh, locating fields by semantic content rather than coordinates. For logistics teams processing delivery notes from suppliers who use formats that change between shipments, this is the difference between a tool that works today and one that needs constant maintenance.

Connecting Extracted Data to French ERP and Accounting Systems

Once delivery note data is structured in a spreadsheet, the integration path depends on the ERP your team uses. French logistics and accounting software has well-established import pathways for structured data — the bottleneck has always been getting the data into a structured format in the first place.

Sage 100 ERP and Cegid. Both support Excel and CSV import for goods receipt records. For Sage, the receiving entry maps to the Gestion Commerciale (commercial management) module, where the delivery note data triggers the stock movement and generates the provisional accounting entry. Cegid's Gestion des Stocks (inventory management) module accepts structured imports for entrées de stock (stock receipts), linking each line to the purchase order for automatic quantity matching.

Divalto infinity. With its native WMS module, Divalto accepts structured receipt data that maps directly to warehouse locations. The delivery note SKU, quantity, and unit fields populate the réception (receipt) screen, and the WMS engine allocates stock to the optimal picking location based on predefined storage policies. The key integration point is the PO reference — when the delivery note carries the purchase order number, Divalto automatically matches the receipt against the open order and updates the inventory in real time.

EBP and Pennylane. For smaller structures, EBP Comptabilité imports Excel files directly into the journal des achats (purchase journal). Pennylane, widely adopted by French experts-comptables (chartered accountants) for its compliance with the upcoming e-invoicing mandate, accepts structured data imports that feed into both the accounting and the supplier invoice reconciliation workflow.

At the accounting level, a delivery note triggers the provisional facture non parvenue (FNP — invoice not yet received) entry when goods arrive at month-end before the corresponding invoice. The entry debits compte 607 (Achats de marchandises) and credits compte 408 (Fournisseurs — Factures non parvenues) for the estimated value. When the invoice arrives, the FNP is reversed. The delivery note provides the quantity data needed to estimate the accrual — and without it, the month-end cutoff requires calling the warehouse or guessing.

For teams already handling French invoice (facture) data extraction, adding delivery note extraction creates a complete digital trail: the purchase order defines what was ordered, the delivery note confirms what arrived, and the invoice states what to pay. Each document feeds into the next, and when all three are structured in spreadsheets, the three-way match that used to take hours per supplier becomes a set of Excel formulas that run in minutes. For year-end processing, the combination of delivery note and invoice extraction produces a document trail that satisfies both the clôture comptable (year-end closing) requirements and the 10-year retention obligation under Article L123-22 of the Code de commerce.

FAQ

Can AI extract data from handwritten French delivery notes?

Yes, with an important qualifier. ImageToTable.ai reads handwritten text using vision model AI — it can capture handwritten quantities, product descriptions, delivery dates, and even scribbled shortage notes in the margin. Accuracy on handwriting is lower than on printed text. A delivery note with clear block-capital handwriting in the line-item table will extract reliably. A completely handwritten carbon copy in cursive French script will produce lower fidelity — the AI will capture the fields it can read with confidence, but some characters may be misread. The tool is not a substitute for requesting legible forms from suppliers. For that specific use case, see handwritten delivery note extraction for field-by-field expectations, or how handwritten proof-of-delivery automation works in practice.

What if my suppliers all use different delivery note formats?

The tool works by understanding what data a column name represents, not by matching a template. When you define a column called "Supplier SIREN," the AI searches every uploaded delivery note for a 9-digit number near the company identification section — whether the document labels it "SIREN," "N° SIREN," "SIRET," or prints it without a label in the footer. When you define "Quantity Delivered," the AI finds the numeric value associated with each line item in the delivery note table, regardless of whether the column header reads "Qté livrée," "Quantité," "Qty," or has no header at all. The same column definition works across all formats because the AI reads the document to understand which value corresponds to which concept. You do not build templates per supplier.

Can I process delivery notes in batches — say 50 at once?

Yes. Upload multiple files in a single batch and the tool processes all of them against the same column definition. Each delivery note generates its rows in the output table. For example, a batch of 30 delivery notes, each with an average of 8 line items, produces a 240-row spreadsheet — one row per line item, with the delivery note number, supplier, date, and PO reference repeated on each row for each document. Processing time scales with the number of pages, not the number of different formats. A batch of 50 single-page delivery notes completes in minutes. For high-volume environments, this replaces the daily routine of typing delivery data into a receiving spreadsheet and frees the receiving clerk to verify quantities rather than transcribe them.

What if the delivery note does not reference a purchase order number?

This is common with smaller French suppliers and courier services, and it is the single biggest obstacle to automated three-way matching. Without a PO reference on the delivery note, the extracted data can still be matched by supplier name plus delivery date plus product reference — but this requires more manual verification. The extraction tool retrieves whatever identification the delivery note does carry: supplier name, delivery date, product descriptions, quantities. The subsequent matching step in your spreadsheet relies on those fields. To reduce the frequency of missing PO references, include a requirement in your supplier agreements that all delivery notes carry the purchase order number — this is a contractual term, not a legal obligation, and many French suppliers will comply when asked.

Does the delivery note extraction handle TVA or tax data?

No — by design. A French delivery note, unlike an invoice, does not display VAT (taxe sur la valeur ajoutée — TVA) rates, TVA amounts, or unit prices. The bon de livraison confirms the physical transfer of goods; the facture carries the financial and tax data. TVA handling belongs in the invoice extraction step, where columns like "TVA 20%," "TVA 10%," "TVA 5.5%," and "Total TTC" map to the accounting entries for compte 44566 (TVA déductible). For the full invoice-to-accounting workflow, including multi-rate TVA extraction, SIREN capture, and CA3 declaration preparation, see the French invoice extraction guide.

Can the extracted spreadsheet import directly into our existing ERP?

If your ERP supports Excel or CSV import for receiving records — which Cegid, Sage 100, Divalto, EBP, and most French logistics ERPs do — the answer is yes. The extracted spreadsheet uses consistent column names and data formats across all delivery notes, which is the prerequisite for a clean import. For automated integration, the same data can be exported as CSV or JSON and fed into systems through their API or file-watch directories. The extraction layer produces structured, consistently formatted data; the integration layer is where you connect that data to your specific ERP. If your ERP's import tool requires a specific column naming convention, rename the columns in the spreadsheet before import — the extraction tool gives you clean data; the ERP determines the import format.

Does extracting delivery notes digitally satisfy the 10-year retention requirement?

Under Article L123-22 of the Code de commerce, commercial documents — including delivery notes — must be retained for 10 years. Digital copies are legally acceptable in France, provided the digitization process guarantees valeur probante (evidentiary value): the digital copy must faithfully reproduce the original, and the process must ensure the copy cannot be altered. A PDF delivery note stored in your document management system satisfies this requirement. The extracted Excel spreadsheet is a data file derived from the delivery note — it does not replace the original document for legal retention. Keep both: the original PDF in your document archive for legal compliance, and the extracted spreadsheet for operational use.

A Receiving Spreadsheet That Writes Itself

The receiving clerk's daily routine — open the delivery note, find the PO number, type the quantities, note the shortages, repeat — has not changed since paper delivery notes became standard commercial practice. What has changed is that the typing step is no longer necessary. The same AI that reads a French invoice and understands multi-rate TVA can read a delivery note and understand what "Qté livrée" means in a table, what "manquant" scribbled in the margin implies, and which 9-digit number near the header is the supplier's SIREN.

The output is a spreadsheet whose columns match the fields your three-way match, your goods receipt entry, and your ERP import expect — built by defining columns once and processing any delivery note, from any supplier, in any format. Try it on a delivery note from your next receiving shift. See if the spreadsheet writes itself.