Customs Explorer
⌕

Customs Explorer

Invoice data extraction

CUSTOMS DOCUMENT INTELLIGENCE / 01

Invoice Data Extraction.

From Commercial Invoices to Structured Customs Data.

Commercial invoices contain essential information for international trade. Invoice data extraction transforms that information into structured, traceable fields that can support customs declarations, document checks and digital workflows.

Extraction technologies can identify parties, goods descriptions, quantities, prices, currencies, delivery terms and document references. The extracted values must still be checked for accuracy, completeness and suitability for their intended use.

Important distinction: extracting an invoice amount is not the same as determining the customs value of imported goods.

DOCUMENT INTELLIGENCE
DEMO WORKSPACE
SOURCE DOCUMENT

Commercial Invoice

INVOICE NO. INV-2026-0852
SELLER Northstar Components Ltd. Illustrative exporter
BUYER Meridian Industrial S.r.l. Illustrative importer
DESCRIPTION QTY VALUE
Industrial control modules 100 PCS €5,000.00
INVOICE TOTAL EUR 5,000.00
STRUCTURED OUTPUT

Extracted Fields

SAMPLE OUTPUT
invoice_number INV-2026-0852 CAPTURED
seller_name Northstar Components Ltd. CAPTURED
goods_quantity 100 PCS CAPTURED
invoice_currency EUR CAPTURED
invoice_total 5,000.00 CAPTURED

Illustrative data only. This is not a live extraction result, accuracy assessment or customs compliance check.

01 / CAPTURE

Read the Source

Identify relevant information within commercial invoices.

02 / STRUCTURE

Organise the Data

Convert document content into usable, structured fields.

03 / CONTROL

Preserve the Evidence

Keep extracted values connected to their original source.

02 / INVOICE DATA FIELDS

The Data Behind Every Invoice.

A commercial invoice is more than a document containing a price. It is a structured source of information about a transaction.

The extraction process should identify relevant data at both document and line-item level, preserving its original meaning, currency, units and relationship to the source.

Not every invoice contains every field needed for customs purposes. Missing information must be distinguished from information that has simply not been detected.

DATA EXTRACTION FRAMEWORK

Eight Core Categories of Invoice Data.

A reliable extraction model should organise information according to its operational role rather than treating every value as isolated text.

01 / IDENTIFICATION 01

Document Identification

Information used to identify the invoice and associate it with the correct transaction.

Invoice number Issue date Invoice type Order reference
02 / PARTIES 02

Commercial Parties

Names, addresses and identifiers relating to the parties involved in the transaction.

Seller Buyer Addresses Tax identifiers
03 / GOODS 03

Goods Information

Product descriptions and commercial references associated with individual invoice lines.

Description SKU / Part number Product code Line reference
04 / QUANTITIES 04

Quantities & Units

Numeric quantities and their units of measurement, which must be preserved together.

Line quantity Unit of measure Net weight, if stated Gross weight, if stated
05 / FINANCIAL 05

Prices & Currency

Monetary values associated with each line and with the overall invoice.

Unit price Line amount Invoice total Currency Discounts, if stated
06 / DELIVERY 06

Delivery & Trade Terms

Commercial delivery terms and references relevant to shipment arrangements.

Incoterms rule, if stated Named place Transport reference Delivery date
07 / ORIGIN 07

Origin Statements

Any origin-related information appearing on the invoice, including statements that may require separate assessment.

Country of origin, if stated Origin declaration Exporter reference
08 / REFERENCES 08

Supporting References

Identifiers connecting the invoice with orders, transport records and supporting documents.

Purchase order Packing list Shipment reference Certificate reference
LINE-ITEM INTELLIGENCE

One Invoice Line. Multiple Structured Fields.

Extraction must preserve the relationship between a product description, its quantity, unit price, currency and calculated line amount.

ILLUSTRATIVE DATA
SOURCE INVOICE / LINE 001 INV-2026-0852
PRODUCT DESCRIPTION Industrial control modules
PART NUMBER ICM-240
QUANTITY 100 PCS
UNIT PRICE EUR 50.00
LINE TOTAL EUR 5,000.00

Fictional commercial invoice line. Values are shown for illustration only.

STRUCTURED EXTRACTION SAMPLE
line_number 001
goods_description Industrial control modules
part_number ICM-240
quantity 100
unit_of_measure PCS
unit_price 50.00
currency EUR
line_amount 5000.00
CUSTOMS DATA RELEVANCE

Extracted Data vs. Customs Requirements.

A field can be useful for a customs declaration without being sufficient to establish the relevant customs fact.

Invoice data Potential customs use Additional consideration
Seller and buyer Transaction and party identification May not identify every legally relevant customs party
Goods description Supporting product identification May be insufficient for tariff classification
Quantity and units Declaration data and document reconciliation Required customs units may differ from commercial units
Price and currency Starting information for valuation analysis Invoice total is not automatically the customs value
Incoterms and delivery place Understanding delivery and cost arrangements Actual costs and relevant valuation adjustments still require assessment
Origin statement Supporting origin-related review Origin eligibility and evidence require separate verification
Document references Linking supporting records Matching references does not establish document authenticity
THE DATA PRINCIPLE

Capture the Value. Preserve the Context.

High-quality invoice extraction preserves the source document, the original field values and the relationships between them. Validation, interpretation and customs-specific decisions remain distinct subsequent steps.

03 / EXTRACTION ARCHITECTURE

From Document to Trusted Workflow.

Invoice extraction is not a single OCR operation. It is a controlled sequence of document processing, field identification, normalisation, verification and review.

The objective is to produce structured data with sufficient source evidence and quality information to support downstream customs processes.

A technically successful extraction does not establish that the information is correct for a particular customs declaration.

THE EXTRACTION PIPELINE

Six Stages. One Controlled Data Flow.

Each stage has a different responsibility. Separating them makes errors easier to identify and review.

01 INPUT

Document Intake

Receive the invoice and establish its source, document identity, file type and transaction context.

OUTPUT Document record + source reference
02 RECOGNITION

Text & Layout Recognition

Read text and identify document structure, including headings, tables, labels and line items.

OUTPUT Recognised text + layout evidence
03 EXTRACTION

Field Identification

Associate recognised content with invoice fields such as seller, invoice number, quantity, currency and total.

OUTPUT Candidate fields + source locations
04 STRUCTURE

Data Normalisation

Convert identified values into consistent data types while preserving original text, units and currency.

OUTPUT Structured data + original values
05 QUALITY

Quality Assessment

Apply confidence indicators, format checks and consistency rules to identify uncertain or conflicting values.

OUTPUT Quality findings + review flags
06 CONTROL

Human Review & Release

Present exceptions to a reviewer and record decisions before releasing data to an authorised downstream workflow.

OUTPUT Reviewed dataset + audit record
ILLUSTRATIVE EXTRACTION ENGINE

What a Controlled Extraction Result Looks Like.

A useful output contains more than field names and values. It also records where the data came from and whether it requires further attention.

DEMO / NOT LIVE
EXTRACTION CONTROL WORKSPACE
ILLUSTRATIVE PIPELINE v1
DOCUMENT INV-2026-0852
TRANSACTION TX-2026-0842
DOCUMENT TYPE Commercial Invoice
RELEASE STATUS REVIEW REQUIRED
FIELD EXTRACTED VALUE SOURCE QUALITY STATUS
invoice_number INV-2026-0852 Header / Ref. 01 NO FLAG
seller_name Northstar Components Ltd. Parties / Seller NO FLAG
goods_quantity 100 PCS Line 001 / Qty NO FLAG
unit_price EUR 50.00 Line 001 / Price NO FLAG
invoice_total EUR 5,000.00 Totals / Amount NO FLAG
delivery_terms Not detected Source location unavailable REVIEW
EXTRACTION EXCEPTION / EXT-006

Delivery Terms Not Detected

The illustrative extraction process did not identify a delivery term in the source invoice. A reviewer must determine whether the term is absent, appears elsewhere or was not recognised.

OPEN REVIEW

Fictional document and illustrative statuses. “No flag” means that no issue is shown by the example quality controls; it is not proof of extraction accuracy or legal validity.

CONFIDENCE & QUALITY

Confidence Is a Signal, Not a Guarantee.

Model confidence, rule-based checks and human verification serve different purposes.

01

Recognition Confidence

A recognition model may assign a confidence indicator to text or extracted fields.

Important: a high score does not necessarily mean the value is correct.
02

Consistency Controls

Rules can compare quantities, prices, totals, currencies and references.

Important: a mathematically consistent invoice may still contain incorrect information.
03

Human Verification

A reviewer can inspect the source and determine how to resolve uncertain or conflicting fields.

Important: review decisions should be traceable and supported by available evidence.
EVIDENCE & GOVERNANCE

Keep the Source. Keep the Decision Trail.

Structured output should remain connected to the evidence from which it was produced. This supports correction, investigation and auditability.

A /

Source Provenance

Document identity, file version, relevant page and field location.

B /

Extraction Trace

Original text, normalised value, processing version and quality findings.

C /

Review Record

Reviewer identity, corrections, supporting evidence and decision history.

D /

Controlled Handover

Authorised release of reviewed fields to the intended downstream system.

ARCHITECTURAL PRINCIPLE

Automate the Capture. Control the Consequences.

An effective invoice extraction architecture does not hide uncertainty. It identifies questionable data, preserves evidence and provides a clear path to human review before information is used in sensitive customs processes.

Explore Operational Applications
04 / OPERATIONAL APPLICATIONS

Extract Once. Use with Control.

Structured invoice data can reduce repetitive entry and support more consistent customs workflows.

Its operational value depends on the quality of the source, the extraction process and the controls applied before the data reaches another system.

Extraction is therefore an input to customs operations, not an automatic authorisation to declare, classify or value goods.

PRACTICAL APPLICATIONS

Five Ways Invoice Data Supports Operations.

The same invoice fields may support several processes, provided that each process applies its own validation and approval requirements.

01 CUSTOMS OPERATIONS

Import Declaration Preparation

Extracted invoice details can pre-populate a declaration preparation workspace and reduce manual transcription.

DATA THAT MAY BE REUSED
  • Commercial parties and invoice references
  • Goods descriptions and quantities
  • Invoice amounts and currencies
  • Delivery terms, where stated
Control requirement

Classification, origin, valuation and procedure-specific declaration data require separate assessment.

02 SYSTEM INTEGRATION

ERP & Business System Integration

Invoice data can be mapped to purchase orders, goods receipts, shipment records and other business-system entities.

DATA THAT MAY BE REUSED
  • Supplier and customer references
  • Purchase order identifiers
  • Product and line-item information
  • Quantities and commercial values
Control requirement

Matching records must not silently overwrite authoritative ERP master data.

03 DOCUMENT INTELLIGENCE

Cross-Document Matching

Extracted invoice fields can be compared with packing lists, transport documents and supporting certificates.

DATA THAT MAY BE COMPARED
  • Invoice and shipment references
  • Goods descriptions and identifiers
  • Quantities and units
  • Parties and document dates
Control requirement

Differences require context. A mismatch does not always establish an error.

04 WORKFLOW AUTOMATION

Automated Exception Routing

Data quality findings can direct documents to the appropriate review queue before downstream processing.

EXAMPLE AUTOMATION
  • Flag missing invoice identifiers
  • Detect unrecognised currencies
  • Identify inconsistent line totals
  • Route uncertain fields for review
Control requirement

Routing rules and release permissions must be defined for each workflow.

05 AUDIT & TRACEABILITY

Evidence & Audit Support

A controlled extraction record can document which invoice fields were captured, corrected, reviewed and transferred.

RECOMMENDED RECORD ELEMENTS
  • Original document and version
  • Source location of each extracted field
  • Original and normalised values
  • Extraction and ruleset versions
  • Reviewer decisions and corrections
  • Downstream transfer and approval history

Auditability supports investigation and accountability; it does not independently establish the legal correctness of a declaration.

ILLUSTRATIVE OPERATIONAL WORKFLOW

One Invoice. Several Controlled Destinations.

The following example shows how a single extracted dataset could support different business and customs functions.

FICTIONAL EXAMPLE
SOURCE DOCUMENT

Commercial Invoice

INV-2026-0852
GoodsIndustrial control modules
Quantity100 PCS
Invoice totalEUR 5,000.00
Delivery termsNot detected
DATA QUALITY GATE REVIEW

Extraction Control

Check the extracted fields, identify exceptions and determine what may be released.

✓ Invoice reference captured
✓ Quantity and amount captured
! Delivery terms require review
A /
ERP Purchase Record

Controlled mapping to an existing transaction

PENDING
B /
Document Matching

Comparison with packing and shipment records

PENDING
C /
Customs Preparation

Draft data for subsequent customs assessment

PENDING

Illustrative workflow only. The example does not represent a live extraction system or an actual declaration. All downstream destinations remain pending until the relevant review and release conditions are satisfied.

AUTOMATION BOUNDARIES

What Can Be Automated — and What Cannot Be Assumed.

Technical automation should be limited by the information available and the controls required for the intended action.

POTENTIALLY AUTOMATABLE

Data Processing Tasks

  • Capturing invoice references and values
  • Normalising dates, quantities and currencies
  • Checking configured arithmetic relationships
  • Linking records through supported identifiers
  • Routing incomplete or uncertain fields
REQUIRES SEPARATE ASSESSMENT

Customs Decisions

  • Determining the appropriate tariff classification
  • Assessing preferential or non-preferential origin
  • Determining the customs value
  • Confirming procedure-specific requirements
  • Authorising a declaration or other legal action

Some customs decisions may be supported by specialised rules or automated systems. However, invoice extraction alone does not establish those decisions or remove the applicable responsibilities of the parties involved.

OPERATIONAL PRINCIPLE

Reuse the Data. Retain the Controls.

The value of invoice data extraction comes from reducing repetitive work while improving the visibility of exceptions. The objective is not unrestricted automation, but a reliable handover between document processing, business systems and customs decision-making.

Explore Risks & Regulatory Considerations
05 / RISKS & REGULATORY CONSIDERATIONS

Accurate Extraction. Responsible Decisions.

Extracting the value printed on an invoice is not the same as determining the information legally required for a customs declaration.

Extraction systems can misread documents, misinterpret context or reproduce information that is incomplete or commercially correct but unsuitable for a particular customs purpose.

Reliable operations require technical controls, regulatory assessment and clearly assigned human responsibilities.

A CRITICAL DISTINCTION

Three Questions. Three Different Controls.

A successful technical check must not be presented as confirmation of customs compliance.

01 / CAPTURE

Was the Invoice Read Correctly?

Does the extracted field match the text, number or table entry appearing in the source document?

CONTROL Source comparison and extraction review
02 / CONSISTENCY

Is the Information Coherent?

Are quantities, prices, totals, currencies and references consistent within and across the relevant documents?

CONTROL Arithmetic and document consistency checks
03 / CUSTOMS ASSESSMENT

Is the Data Appropriate for Declaration?

Does the information support the correct customs value, classification, origin and procedure-specific requirements?

CONTROL Separate customs and regulatory assessment
RISK LANDSCAPE

Six Failure Modes That Matter.

Effective controls must address both technical extraction failures and incorrect use of the extracted information.

RISK 01 DATA CAPTURE

OCR & Recognition Errors

Similar characters, poor scans and complex invoice layouts can cause incorrect values to be extracted.

EXAMPLE

A quantity of 1,000 units is recognised as 100 units.

CONTROL RESPONSE Field-level evidence, quality flags and targeted source verification.
RISK 02 INTERPRETATION

Incorrect Field Association

A system may read the text accurately but associate it with the wrong invoice field.

EXAMPLE

A subtotal is identified as the final invoice total.

CONTROL RESPONSE Layout-aware extraction and semantic field verification.
RISK 03 NORMALISATION

Currency & Unit Errors

Incorrect conversion or interpretation of currencies, decimal separators and units can change the meaning of extracted values.

EXAMPLE

EUR 1.250,00 is incorrectly interpreted as EUR 1.25.

CONTROL RESPONSE Preserve original values and apply locale-aware normalisation.
RISK 04 CUSTOMS VALUE

Invoice Value Misused as Customs Value

The commercial invoice amount may not equal the customs value determined under the applicable valuation rules.

EXAMPLE

A transaction may require adjustments relating to transport or other relevant valuation elements.

CONTROL RESPONSE Separate invoice extraction from customs valuation logic.
RISK 05 COMPLIANCE

Unsupported Regulatory Assumptions

Extracted descriptions, origin statements or delivery terms may be insufficient to establish a customs decision.

EXAMPLE

A country stated on the invoice is treated as proof of preferential origin.

CONTROL RESPONSE Require appropriate supporting evidence and procedural assessment.
RISK 06 GOVERNANCE

Uncontrolled Data Release

Extracted fields may be transferred into business or customs systems without adequate review or traceability.

EXAMPLE

An uncertain delivery term is silently replaced with a default value.

CONTROL RESPONSE Approval gates, access controls and recorded overrides.
WORKED EXAMPLE / CUSTOMS VALUATION

The Invoice Total Is Correct. The Customs Value Still Needs Assessment.

A correctly extracted commercial value is only one input into the customs valuation process.

ILLUSTRATIVE ONLY
COMMERCIAL INVOICE INV-2026-0852
Goods Industrial control modules
Quantity 100 PCS
Unit price EUR 50.00
Invoice total EUR 5,000.00
Delivery terms Not detected
EXTRACTION Invoice amount captured and arithmetic relationship confirmed.
SEPARATE CUSTOMS ASSESSMENT

What Still Needs to Be Established?

01

Whether the transaction value method applies under the relevant customs valuation rules.

02

Which transport, insurance or other cost elements are included in the price.

03

Whether additions or exclusions are required under the applicable valuation provisions.

04

Whether further evidence is needed before determining the declarable customs value.

OUTCOME CUSTOMS VALUE NOT YET DETERMINED

The example intentionally does not calculate a customs value: the information required for that determination is incomplete.

LEGAL CONTEXT

Union Customs Code, Regulation (EU) No 952/2013, Articles 70–74 (customs valuation), including Articles 71–72 on additions and elements not included in customs value.

RISK CONTROL MATRIX

Match Each Risk to a Defined Control.

Controls should be documented, assigned and tested against realistic document variations.

RISK AREA CONTROL MECHANISM HUMAN RESPONSIBILITY
Recognition errors Source-linked field checks and quality flags Verify uncertain values against the document
Incorrect field mapping Schema validation and contextual checks Confirm disputed field interpretation
Currency or unit errors Original-value retention and normalisation controls Approve material conversions or corrections
Customs valuation Separate valuation assessment and supporting evidence Determine the applicable valuation treatment
Regulatory assumptions Procedure-specific rules and evidence requirements Assess classification, origin and legal applicability
Uncontrolled release Role-based permissions and recorded approval gates Authorise or reject downstream transfer
EU REGULATORY FRAMEWORK

The Legal Framework Behind the Data.

Invoice extraction technology operates within existing customs obligations. The rules governing declarations and customs value are distinct from the technology used to capture commercial data.

01 / PRIMARY LEGISLATION EU

Union Customs Code

Regulation (EU) No 952/2013

Establishes the core legal framework for Union customs procedures, declarations, responsibilities and valuation.

RELEVANT PROVISIONS Article 15 — Information and supporting documents Article 77 — Customs debt on release for free circulation Articles 70–74 — Customs valuation Articles 162–163 — Declaration particulars and supporting documents
Official Regulation
02 / DELEGATED RULES EU

UCC Delegated Act

Commission Delegated Regulation (EU) 2015/2446

Supplements the Union Customs Code with detailed provisions relevant to customs procedures and the treatment of information.

OPERATIONAL RELEVANCE Detailed rules supplementing the UCC Customs valuation provisions Procedure-specific requirements Applicable data and documentary conditions
Official Regulation
03 / IMPLEMENTING RULES EU

UCC Implementing Act

Commission Implementing Regulation (EU) 2015/2447

Provides implementing rules for applying the UCC, including relevant customs procedural and data-handling requirements.

OPERATIONAL RELEVANCE Customs declaration processes Implementation of UCC requirements Supporting procedural arrangements Rules applicable to customs information
Official Regulation
DATA PROTECTION

Personal Data in Commercial Documents

Invoices may contain names, contact details or other information relating to identifiable natural persons. Where personal data are processed, Regulation (EU) 2016/679 (GDPR) may apply, including requirements concerning lawful processing, security and data minimisation.

Consult the GDPR on EUR-Lex

Legal references are provided for orientation. The applicable consolidated texts, amendments, procedural context and national implementation must be checked for the operation concerned. These regulations do not, by themselves, certify or approve any particular invoice extraction software.

HUMAN OVERSIGHT

Automation Needs Clear Decision Ownership.

Review effort can be risk-based, but responsibility must not disappear when processing becomes automated.

LEVEL 01

Routine Processing

Apply configured extraction and technical quality checks to clearly defined fields.

EXPECTED CONTROL Documented rules, monitoring and defined exceptions.
LEVEL 02

Exception Review

Investigate missing, ambiguous or conflicting information before it is reused.

EXPECTED CONTROL Source evidence, correction records and review decisions.
LEVEL 03

Customs Assessment

Evaluate regulatory questions that cannot be resolved through extraction alone.

EXPECTED CONTROL Appropriate expertise, documented assessment and authorised action.
RISK MANAGEMENT PRINCIPLE

Preserve the Evidence. Make the Limits Visible.

Trustworthy invoice extraction makes data easier to use without disguising uncertainty. It preserves the original source, distinguishes technical quality from customs correctness and ensures that consequential decisions remain subject to appropriate controls.

Explore Implementation & Best Practices
06 / IMPLEMENTATION & RESOURCES

From Invoice Capture to Controlled Operations.

A reliable extraction workflow is built around data quality, traceability and clear operational responsibilities.

The objective is not simply to turn a commercial invoice into structured data. It is to ensure that the resulting information can be reviewed, understood and reused safely.

A phased implementation helps organisations identify limitations before extraction becomes part of business-critical customs processes.

IMPLEMENTATION ROADMAP

Five Stages of a Controlled Deployment.

Start with a clearly defined document scope and expand only when extraction performance, exception handling and downstream controls are understood.

01
FOUNDATION DEFINE

Define the Invoice Scope

Identify the invoice types, languages, layouts, commercial fields and business processes that the system must support.

DELIVERABLE Approved document inventory, field catalogue and explicit exclusions.
02
DATA ENGINEERING STRUCTURE

Establish the Data Model

Define standard fields, currencies, units, date formats, line-item structures and source-document references.

DELIVERABLE Versioned extraction schema with original-value retention and field provenance.
03
QUALITY ASSURANCE VALIDATE

Build Quality & Exception Controls

Test extraction accuracy, arithmetic consistency, field completeness and handling of uncertain or conflicting values.

DELIVERABLE Tested validation rules, review queues and documented exception outcomes.
04
INTEGRATION CONNECT

Control Downstream Data Transfers

Map approved fields to ERP, document matching or customs preparation systems without silently replacing missing information with assumptions.

DELIVERABLE Field mappings, transfer permissions, reconciliation checks and release records.
05
OPERATIONS IMPROVE

Pilot, Monitor & Refine

Run controlled pilots using representative invoices, track corrections and measure the impact on review workload and downstream errors.

DELIVERABLE Pilot results, acceptance criteria, issue register and ongoing monitoring plan.
OPERATIONAL READINESS

Six Questions Before Production Use.

A technically functioning extraction model is not necessarily an operationally ready solution.

01

Is the Document Scope Defined?

Supported invoice formats, languages and exclusions are documented.

DOCUMENT COVERAGE
02

Can Every Field Be Traced?

Users can identify the source of each material extracted value.

DATA PROVENANCE
03

Are Exceptions Visible?

Missing, ambiguous and inconsistent values trigger appropriate review.

QUALITY CONTROLS
04

Are Release Rights Controlled?

The workflow identifies who can approve, correct or transfer data.

ACCESS & APPROVAL
05

Is Customs Assessment Separate?

Invoice extraction does not automatically establish valuation, origin or classification decisions.

REGULATORY BOUNDARY
06

Has the Workflow Been Tested?

Representative invoices and difficult cases have been evaluated before production expansion.

OPERATIONAL TESTING
FREQUENTLY ASKED QUESTIONS

Invoice Data Extraction — FAQ.

Practical answers to common questions about extraction accuracy, customs use, automation and compliance.

01 What is invoice data extraction?

Invoice data extraction is the process of identifying and converting information from commercial invoices into structured fields. Typical fields include invoice numbers, dates, parties, goods descriptions, quantities, prices, currencies and totals.

The extracted information may support downstream workflows, but its accuracy and suitability must be evaluated for the intended use.

02 Is OCR the same as invoice data extraction?

No. Optical Character Recognition (OCR) converts visible text into machine-readable characters. Invoice data extraction also identifies the meaning and structure of that information.

For example, recognising “EUR 5,000.00” is different from determining whether it represents a subtotal, tax amount or invoice total.

03 Can extracted invoice data be used in a customs declaration?

Extracted data can support preparation of a customs declaration, including transcription of relevant commercial information.

However, the information must be checked against the requirements of the specific customs procedure. Extraction does not itself determine classification, customs value, origin or the completeness of a declaration.

04 Does the invoice total always equal the customs value?

No. The customs value must be determined under the applicable customs valuation rules, including Articles 70–74 of Regulation (EU) No 952/2013.

Depending on the circumstances, the determination may involve additions, exclusions or alternative valuation methods. A correctly extracted invoice amount is not sufficient to establish the customs value.

05 Can invoice extraction be fully automated?

Many extraction and normalisation tasks can be automated within a defined document scope.

The appropriate level of automation depends on data quality, system reliability, the consequences of errors and the controls required for downstream actions. Exceptions and regulatory decisions may still require human review.

06 What records should an extraction system retain?

A well-controlled workflow should retain appropriate evidence of the source document, extracted fields, normalisation, corrections, review decisions and downstream transfers.

Retention periods and access arrangements must be determined under the applicable legal and operational requirements, including customs record-keeping obligations and data protection rules where relevant.

THE CUSTOMS EXPLORER PERSPECTIVE

Better Data Capture. Better-Controlled Customs Workflows.

Invoice data extraction can improve efficiency, consistency and traceability. Its greatest value comes when reliable extraction is combined with transparent quality controls, appropriate customs assessment and accountable operational decisions.

Back to Page Introduction