30 Albaranes a Day, One Log
Batch Processing Spanish Delivery Notes
Spanish warehouse management software maker Mecalux — a Barcelona-based intralogistics group operating in 70+ countries, named in the 2026 Gartner Magic Quadrant for WMS — reports that its Easy WMS platform achieves 99% error elimination and 60% productivity improvement for its warehouse clients. Those numbers describe what happens when the WMS receives clean data. The gap between that promise and reality is the albarán (delivery note) stack sitting on the receiving desk every morning: 30 pages from 20 different suppliers printed by 20 different ERP systems, some on thermal paper, some carbon-copy, most carrying handwritten quantity corrections from the receiver. This article is about closing that gap at batch scale.
Key Takeaways
- A morning stack of 30 albaranes from 20 suppliers costs 2 hours of typing — and that's the cheap part.
- Mistype one albarán number by a single digit and the three-way match for that supplier's entire monthly invoices breaks — but you won't find out for three weeks.
- ImageToTable.ai reads albarán numbers, NIFs, and quantities from 20 different supplier formats with one column definition — because the AI looks for what the field means, not where it sits on the page.
A single albarán takes minutes to key in. The real problem starts when there are 30 of them from 20 different suppliers every morning.
Single-document extraction solves a 3-minute problem. Batch processing solves a structural workflow problem where 30 minutes of data entry isn't the cost — it's the 30 independent decisions about field placement, the 30 opportunities to misread a handwritten quantity, and the 30 manual keystrokes into the WMS that each carry a 1-3% error rate.
A single Spanish delivery note (albarán) is manageable. The receiving clerk reads the printed fields — albarán number, supplier NIF, product codes, shipped quantities — and types them into the WMS or ERP. At 3 minutes per page, a stack of 5 takes 15 minutes. That's tedious but survivable. The problem is that warehouses receiving from a typical Spanish supplier base don't see 5 albaranes a day. A mid-size distribution center serving regional retail, or a manufacturing plant receiving raw materials from a mix of large and small Spanish suppliers, handles 20-50 albaranes daily. At that volume, the 3-minute-per-page math produces 1-2.5 hours of pure data entry. But the real cost isn't the typing time — it's what happens when those 50 manual transcriptions meet the three-way match.
If you need a primer on what makes Spanish albaranes structurally different from generic delivery notes — including the NIF tax ID format, the carbon-copy paper tradition, and the legal weight they carry under the Código de Comercio — start with the single-document albarán extraction guide. This article assumes you already know what an albarán is and focuses on what changes when you process them by the dozen instead of one at a time.
The scale cliff isn't about speed — it's about decision density. Processing one albarán requires one set of field-location decisions. Processing 30 from different suppliers requires 30 independent sets. The human brain can handle 5 without fatigue. At 30, error rates compound. At 50, the three-way match — albarán → factura → purchase order — starts breaking not because any single entry was wrong, but because the cumulative probability of a mismatch across 150 reference numbers (50 albaranes × 3 match keys) becomes near-certain.
Batch processing doesn't amplify format fragmentation — it reveals that template-based extraction was never a real solution.
When you process one albarán, you can tolerate its quirks. When you process 30 from 20 suppliers, the format differences between a Mercadona logistics center albarán, a local Andalusian manufacturer's hand-filled albarán book, and a Basque industrial supplier's Sage Murano PDF are not quirks — they are a structural barrier that template-based extraction cannot cross.
The Spanish albarán format landscape is permanently heterogeneous because the economics of standardization are misaligned. The AECOC (Asociación de Fabricantes y Distribuidores, Spain's GS1 standards body with over 35,000 member companies) has developed electronic delivery note standards through its AECOC EDI and AECOC TRANSP platforms. Large retailers like Mercadona, Carrefour Spain, and El Corte Inglés mandate EDI for their Tier 1 suppliers. But for the mid-market — regional distributors, industrial suppliers, local manufacturers — paper albaranes printed from whatever ERP the supplier runs remain the norm.
UNO Logística, Spain's main logistics trade association representing over 350 logistics operators and transport companies, identifies document format fragmentation as a structural barrier to supply chain digitalization. The Centro Español de Logística (CEL), Spain's largest logistics professional community since 1978, has reached the same conclusion in its digitalization working groups: standardization will reach high-volume corridors first, but the long tail of mid-market suppliers will continue producing heterogeneous paper and PDF albaranes for years.
At single-document scale, you can build one template per supplier. At 20 suppliers, that's 20 templates to create and maintain — and templates break when a supplier changes their ERP output format without telling you. The receiving clerk doesn't control supplier formats. They receive what arrives at the dock. The only viable approach at batch scale is Custom Column Extraction: a method where you define the field names you want — "Albarán Number," "Supplier NIF," "Qty Shipped" — and the AI reads each document to locate those values by understanding what the field name means, not by matching a fixed pixel position on a specific supplier's layout.
One column definition works across 20 different supplier formats because the AI reads by meaning, not by position. The column names you type become the headers of your output spreadsheet — no template maintenance required.
The batch workflow for Spanish albaranes: upload the day's stack, define columns once, export a unified receiving log.
The entire morning's albaranes — PDFs from the supplier portal, scanned carbon copies from the dock, photos taken at delivery — go into one upload. One column definition. One output spreadsheet.
Batch processing at ImageToTable.ai means uploading multiple files simultaneously, defining your extraction columns once, and receiving a single consolidated output where each document becomes one row in the table. The fundamental advantage is that you don't repeat the column definition step per document — you do it once for the entire batch, and the AI applies the same semantic understanding across every supplier's layout. For Google Sheets users, the sidebar add-on extracts results directly into the active sheet without leaving the spreadsheet.
Upload the day's albaranes — all formats together
Drop in PDF albaranes from supplier portals, scanned carbon-copy pages from the receiving dock, or phone photos taken during delivery checks. Digital PDFs, scanned paper, and photos can be mixed in the same batch upload. If your drivers or remote warehouse staff need to submit albaranes directly, the Collection Link generates a shareable upload page — the sender opens the link, enters a verification code, and uploads files. No login or registration required for the sender. Albaranes land directly in your processing queue, bypassing email attachments and WhatsApp photos entirely.
Define your receiving log columns once — they work across every supplier
Enter the field names your receiving workflow and ERP need: Albarán Number | PO Reference | Supplier Name | Supplier NIF | Delivery Date | Product Code | Item Description | Qty Shipped | Qty Received | Exception Notes | Receiver Signature. The AI finds each value on every supplier's albarán by understanding what the column name means — not by matching a remembered pixel position. Add a computed column like Qty Discrepancy (Qty Shipped − Qty Received) and the AI calculates the difference during extraction, flagging mismatches before data reaches your WMS. Add an inferred column like Delivery Condition (options: Complete/Partial/Damaged) and the AI evaluates the albarán content to assign the correct status across every document in the batch.
Export the consolidated receiving log
Download to XLSX, CSV, or JSON. Each albarán becomes one row. Every field — albarán number, supplier NIF, product codes, shipped and received quantities, exception notes, signature status — appears in its own column. The spreadsheet is ready for WMS goods receipt posting in SAP Business One, Sage 200 / Sage Murano, Microsoft Dynamics NAV / Business Central, Holded, or Mecalux Easy WMS. The albarán number column is the key that links each row to the subsequent factura for three-way matching. Processing runs at 5–10 seconds per page.
Try the extraction flow below with a sample delivery note. The demo loads a preset — a pre-configured set of column names matching the standard delivery note and packing slip structure — so you can see extraction results immediately without typing column names.
Files are processed securely and not stored.
Three-way matching doesn't scale linearly — one mistyped albarán number in a batch of 50 can break reconciliation for an entire supplier's monthly invoices.
The Spanish Plan General de Contabilidad (PGC) assigns a specific accounting path to every albarán: debit Grupo 3 Existencias (cuenta 600 for merchandise purchases or 601 for raw materials) and credit cuenta 400 Proveedores. The albarán number is the reference that bridges the physical receipt to this accounting entry — before the factura (invoice) arrives.
In a single-document workflow, if you mistype one albarán number, you catch it when the corresponding factura arrives and doesn't match. You spend 5 minutes tracing the paper trail and fix the entry. Annoying but contained. In a batch workflow where you're keying 30-50 albaranes into the ERP before any facturas arrive, the damage compounds silently: incorrect albarán numbers sit in the system until the supplier sends their invoice, at which point the three-way match — purchase order → albarán (goods received) → factura (invoice) — fails not for one document but potentially for an entire month's deliveries from that supplier. The accounting team now has to reverse-engineer which entries are wrong across dozens of transactions.
This is where extraction that pulls the albarán number directly from the document — rather than relying on human transcription — changes the risk profile. A typed albarán number has a 1-3% error rate per field. A machine-read albarán number taken from the document image itself has effectively zero transcription error on printed text. For scanned carbon copies where print may be faded, the AI flags low-confidence values rather than silently producing a wrong result — which means the receiving clerk verifies the 2-3 uncertain reads instead of proofreading all 50 entries.
The three-way match breaks not at the invoice stage but at the albarán entry stage — weeks before anyone notices. Fixing it at the point of data capture costs seconds. Fixing it after facturas arrive costs hours per supplier.
A Spanish albarán carries two data layers: what the supplier shipped and what the receiver actually counted. At batch scale, reading both layers manually means reading identical corrections 30 times over.
The albarán's round-trip nature — supplier prints it, goods travel with it, receiver annotates it, signed copy returns — creates a two-layer document problem that is manageable in single-document processing but becomes the dominant bottleneck at batch scale.
Every Spanish albarán that reaches the data entry desk carries three types of handwritten content: quantity corrections ("solo 8 uds recibidas" next to a printed "10"), condition notes ("caja dañada," "embalaje roto"), and the receiver's signature confirming receipt. In a single-document workflow, the receiving clerk reads these annotations and types the adjusted quantity into the WMS. When you have 30 albaranes, the clerk reads 30 handwritten corrections — many using similar phrasing across different suppliers — and must correctly associate each correction with its corresponding printed line item without transposing data between documents.
This is not a speed problem. It's a cognitive-load problem. After 10 albaranes, the brain stops distinguishing between "8 uds recibidas" on the Mercadona albarán and "8 uds" scrawled on the RS Components España delivery note. Template-based extraction tools compound this by either ignoring the handwritten layer entirely (producing shipment quantities that don't reflect what was actually received) or blending handwritten and printed text into the same output cell (producing nonsense data).
A semantic approach solves this by treating the printed and handwritten layers as separate extraction targets. Define Qty Shipped and the AI reads the supplier's printed quantity from the line-item table. Define Qty Received in the same column setup and the AI reads the handwritten correction. These land in adjacent columns in the output — printed quantity and received quantity side by side, discrepancy immediately visible. Define Exception Notes and handwritten margin comments are extracted as structured text.
Standard block handwriting extracts reliably across the batch. Heavily rushed cursive — common in driver annotations scribbled at the dock — may require spot verification for full-text transcription, but structured detection (signature present: yes/no, quantity corrections) remains reliable even on rushed handwriting. This is the honest boundary: the tool reads what's legible, flags what's ambiguous, and never silently produces a wrong value.
Frequently Asked Questions
How many albaranes can I process in one batch?
The tool processes files sequentially within a batch and handles uploads of any practical size — 30, 50, or more albaranes in one go. Each page takes approximately 5–10 seconds to process, so a batch of 30 albaranes completes in roughly 3–5 minutes of processing time. You don't watch the progress bar — you set the batch running and return to a completed spreadsheet. The practical constraint is not the tool's capacity but the receiving dock's intake rhythm: most warehouses batch-process the morning's albarán stack in one session and handle afternoon arrivals in a second run.
What happens if an albarán has no printed albarán number — only a handwritten one?
The AI reads handwritten text regardless of language. If the albarán number is handwritten — which happens with some small suppliers using generic albarán books — the extraction still attempts to locate it based on the column name you defined. The accuracy is lower than for printed numbers (legible block handwriting extracts reliably; rushed cursive may require manual verification). If the AI cannot confidently find a value, the cell is left empty rather than filled with a guess. For quality control, you can add a column named Albarán Number Confidence to surface which entries need manual review.
Can I process albaranes that mix Spanish, Catalan, and English field labels in the same batch?
Yes. The AI reads field meanings, not label language. Whether a supplier prints "Albarán N.º," "Albarà Núm.," or "Delivery Note Ref," the column you defined as Albarán Number finds the correct value because the AI understands what a delivery note number is semantically, not what language it's labeled in. This is particularly useful for Spanish warehouses receiving from international suppliers where albaranes may carry a mix of Spanish, English, and regional language labels — and from domestic suppliers in bilingual regions (Catalonia, Basque Country, Galicia, Valencia) where field labels may appear in the regional language.
Does batch processing handle the NIF/CIF validation for tax ID fields?
The AI extracts the NIF as it appears on the document but does not perform algorithmic NIF validation (check-letter verification against the modulo-23 algorithm). The 8-digit-plus-check-letter format helps the AI distinguish the NIF from the albarán number during extraction, but formal validation should be done in your ERP or via the Agencia Tributaria's online verification service. For quality control, NIF fields benefit from adding a computed column like NIF Format Check (valid/invalid) to flag entries that don't match the expected pattern — but this is a format check, not a legal validation.
How do I get albaranes from my drivers or remote warehouses into the batch without email?
The Collection Link feature generates a shareable upload page. Send the link to delivery drivers, remote warehouse staff, or third-party logistics partners. They open the link, enter a verification code, and upload albaranes directly — as photos taken at delivery or as PDF scans. No registration or login required for the sender. Files land in your processing queue automatically. This is particularly useful for Spanish logistics operations where deliveries happen at multiple sites — a central office in Madrid, a satellite warehouse in Valencia, a distribution hub in Barcelona — and receiving documents need to reach a centralized AP or inventory team without relying on drivers managing email attachments.
The Spanish albarán stack on the receiving desk every morning is not a data entry problem — it's a format translation problem multiplied by the number of suppliers you receive from. When one column definition works across every supplier's layout, the stack becomes a single upload, a single processing run, and a single spreadsheet that feeds your WMS the clean data it was designed to receive. That's the difference between a WMS that reports 99% error elimination on paper and one that actually achieves it on your dock.
Process Your Albaranes in Batch
Related: How to extract data from Spanish delivery notes (albaranes) into Excel · Batch extract packing slips and delivery notes to Excel · Extract packing slip fields from any format into Excel · Delivery note to Excel converter