Customs Broker Digital Tools: Tracking, Document Automation and What to Audit

Whether a customs broker gives you digital tracking and document access is a decision each firm makes, not something the international framework requires of them. What that framework leans towards is the other half of the picture: where the relevant provisions are applied, the customs side is expected to be electronic — the Revised Kyoto Convention has customs permit electronic lodgement of the goods declaration and of supporting documents (Transitional Standards 3.21 and 3.18), and in the EU the Union Customs Code requires exchanges and storage to use electronic data-processing techniques (Art. 6(1)), subject to the derogations in Art. 6(3) and (4). Whether a machine-readable record of your own clearance exists, and whether you may see it, depends on your administration — confirm it with them. The harder question is what a tool shows when nothing is happening: a portal that renders only events makes "waiting on you", "selected for control", "normal dwell" and "nobody has looked" produce the same blank screen. And "automated customs documentation" covers three different things — reading fields off documents you already hold, checking the same fact across documents that should agree, and lodging a declaration with an administration. Only the first two are what a documentation platform does; the third stays with the declarant, who remains responsible for the accuracy of the particulars (RKC Standard 3.8; UCC Art. 15(2)).

Data last updated: 23 August 2026.

• Whether a customs broker gives you a login, a tracking screen and your documents as files is a decision each firm makes. Nothing in the international customs framework obliges them to. Choosing a broker on those terms — what direct and indirect representation change about who owes the duty, what to ask about retention and about getting your archive out — is covered on the customs page, and none of it is repeated here.
• This page starts one step later: you have the tool, or you are being shown a demo of it. What can it actually tell you? Three things go wrong at that point often enough to be worth naming — a screen that renders silence as a blank, a phrase that covers three unrelated operations, and a visibility claim that stops at a border without saying so.

Which kind of quiet is this?

• Which kind of quiet is this? — Four questions about your own file, applied in the order the seven rows below run. It stops at the first class your answers establish, and where they establish none it says so.
• This reads nothing and connects to nothing. It applies the seven rows on this page to what you already know about your own consignment, and returns the class, who would know first, and the one question to put to your broker. Where a decisive answer is unknown it asserts no class. Notification duties, time limits and who they run to are set nationally — confirm those with your customs authority.
• Has your broker named something they are waiting on from you? — A certificate, a corrected invoice line, an instruction on which procedure to use. Answer yes, no, or not sure. Row 1 — the missing item is already a field in the broker's own record, so it can be shown to you with the date it was first needed.
• Have you been told the goods are held or being checked — and by whom? — Answer no; yes, by customs; yes, by another authority or the terminal; or not sure. Rows 3 and 4 — UCC Art. 46(2) has controls other than random checks based primarily on risk analysis, and RKC Transitional Standard 3.35 treats a non-customs inspection as a separate clock to be co-ordinated.
• Where has the declaration got to? — Acceptance and release are defined moments rather than moods, so this is a question with a mechanical answer. Answer not lodged; lodged with acceptance not recorded; accepted; released; or not sure. Rows 2 and 6 — UCC Art. 172(1) has a compliant declaration accepted immediately and Art. 172(2) makes the date of acceptance the date that governs the procedure; UCC Art. 194(1) and RKC Standard 3.40 define release.
• Does your screen show an expected duration next to the elapsed one? — Without it, a file that is fine and a file nobody has opened render identically. Answer yes or no. Rows 5 and 7 — a screen that shows the last event and nothing else renders normal and stuck as the same picture, and a last-opened timestamp has to be built for and nothing obliges it to be.
• Every rule this applies, and what each rests on — Nothing is decided in the interface that is not written here. Each rule names the class it selects and the provision or row it comes from.
• Order of testing — The classes that rest on a recorded fact are tested first, in the order the rows run — a named item outstanding from you (row 1), a declaration lodged and not yet accepted (row 2), a customs check (row 3), a hold from a body that is not customs (row 4), then a recorded release (row 6). Rows 5 and 7 are what is left when none of those holds: they are the two classes with no event at all. That row order runs from the class most cheaply made visible to the class least likely to be made visible, which is also the order in which your leverage as the buyer runs out. — Stated in the body prose immediately before the table on this page.
• Named item outstanding from you — If your broker has named something they are waiting on from you, that is the class whatever else is true: nothing is happening at the border because nothing has been asked of the border yet. — Row 1. RKC Standard 3.17 also allows supporting documents that cannot be lodged with the declaration to be produced within a specified period, so a file can be moving with a document still outstanding.
• A customs check has been notified — If you have been told there is a customs check, that is the class. The fact of it is usually showable; the reason often is not. — UCC Art. 46(2); RKC General Annex Standards 6.3 and 6.4; RKC Standard 9.6 on details of a private or confidential nature.
• The hold is not a customs hold — A veterinary, plant-health, market-surveillance, terminal or berth step outstanding is a different class, and a screen that shows it at all is showing you a relationship rather than an integration. — RKC Transitional Standard 3.35 — customs co-ordinate with other competent authorities' inspections, which tells you these are separate clocks.
• Sent, acceptance not recorded — A declaration sent and not yet accepted is its own class. What decides the worth of that state on a screen is whether it is learned from the filing channel or from a person re-typing it later. — UCC Art. 172(1) and Art. 172(2) — the window is meant to be short and the moment it ends is legally load-bearing rather than cosmetic.
• Release recorded, movement not finished — Once release is recorded the customs stage is finished, and further silence belongs to a transport or commercial stage that may sit in a different system. — UCC Art. 194(1); RKC Standard 3.40, subject to its four conditions.
• An expected duration is on the screen — With an expected duration shown next to the elapsed one, ordinary waiting can be told apart from a file nobody has opened, and the class is normal dwell. — Row 5 — a screen that shows the last event and nothing else renders normal and stuck as the same picture.
• No expected duration: two classes stay merged — With no expected duration on the screen, normal dwell and nobody has looked are the same picture. This device does not choose between them: it returns both, and the question that separates them. — Row 7 — a system has to be built to record when a file was last opened, and nothing obliges it to; this is the only class whose visibility runs against the interest of the party who would have to build it.
• An unknown answer establishes nothing — Where a decisive answer is not sure, no class is asserted. The unanswered questions come back as questions to put to your broker. — The device applies the rows on this page; it does not model your file and has no access to it.
• Nothing here is a statement about your own administration's practice, and no output of this device is advice about your consignment. Retention periods, who may act as declarant, notification duties and the form records must be kept in are set nationally — confirm all of it with your customs authority.

When a customs tracking screen has not changed for three days, what is it telling you?

• A tracking tool gets judged on the events it shows. But much of a customs file's life is not events — it is the gaps between them, and the gaps are where the cost and the temper go. Several genuinely different situations produce exactly the same picture on a portal: the last event, and then nothing.
• That is not a rendering problem, it is a modelling one. Events have an obvious home in a record; non-events do not. Somebody has to decide that "no change since Tuesday" is itself a fact worth showing, name what kind of quiet it is, and say who is in a position to know. A screen either makes that decision or it does not — which is why "we have a tracking portal" and "I can tell whether my consignment is in trouble" are not the same sentence.
• The table below sets out the kinds of quiet: what each one is underneath, who is first to know, whether it can be put on a customer screen at all, and the question that separates a firm that has thought about this from one that has bought a screen.
• The rows are in a deliberate order, and the order is part of the argument. They run from the class most cheaply made visible to the class least likely to ever be made visible — which is also the order in which your leverage as the buyer runs out. The first row costs a broker nothing to show you, because the missing item is already named inside their own system. By the last row you are asking for something no one has a commercial reason to build.
• Two rows are worth reading twice.
• Normal dwell is the class that makes a tracking screen misleading by omission rather than by error. Without an expected duration shown next to the elapsed one, a file that is fine and a file that has been forgotten render identically, and the reader supplies the difference from their own anxiety.
• Nobody has looked is the only class whose visibility runs against the interest of the party who would have to build it. A system has to be built to record when a file was last opened, and nothing obliges it to. Asking for a last-opened timestamp rather than a last-status-changed timestamp is one question, and it tests something a demo cannot show you.
• One caution on the whole table: it describes what can be known, not what any particular administration will tell you, or when. Notification duties, time limits and who they run to are set nationally — confirm those with your customs authority.

Seven kinds of quiet — and which one your screen cannot tell apart

• Seven kinds of quiet — and which one your screen cannot tell apart — The buyer-side material on the customs page describes what a digital tool shows. Nothing there — and nothing in the pages ranking for these queries — describes what it shows when there is nothing to show, which is the state a cargo owner is actually staring at when they type "do customs brokers offer digital tools for tracking". The device is built from the topic's own mechanics rather than from a feature comparison: a tracking record stores events, so an absence of events has no natural representation, and several structurally different situations therefore collapse into one identical blank screen. Typing the absences is the only way to separate them, and each type has a different first-knower, a different reason it can or cannot be displayed, and therefore a different question to ask. The carrying axis is REPRESENTABILITY — the "showable" column — and the unit of each row is a RECORD STATE, not a field, a form input or an obligation; that is what keeps it distinct from a query mask, a keyholder table or a dossier device. The "underneath" column stays descriptive of the situation itself and never splits into the half software closes versus the half it cannot; the first-knower column is diagnostic support for representability, not the axis. The "ask" column is deliberately made of things a buyer can say out loud on a call and get a mechanical answer to — a last-opened timestamp, the gap between filing-channel acceptance and a re-typed status — rather than a checklist of features, because features can be claimed and mechanisms have to be pointed at. Rows are ordered from the class most cheaply made visible to the class least likely to be made visible at all, which is also the order in which the buyer's leverage decreases; that ordering principle is stated in the body prose immediately before the table, so it survives for a scraper that reads the text and never sees the row positions. With the absence-kinds device cut elsewhere, this page becomes the corpus's only typology of absence, which is a further reason to keep the unit a state rather than let it drift into a checklist.
• Seven kinds of quiet — and which one your screen cannot tell apart — The buyer-side material on the customs page describes what a digital tool shows. Nothing there — and nothing in the pages ranking for these queries — describes what it shows when there is nothing to show, which is the state a cargo owner is actually staring at when they type "do customs brokers offer digital tools for tracking". The device is built from the topic's own mechanics rather than from a feature comparison: a tracking record stores events, so an absence of events has no natural representation, and several structurally different situations therefore collapse into one identical blank screen. Typing the absences is the only way to separate them, and each type has a different first-knower, a different reason it can or cannot be displayed, and therefore a different question to ask. The carrying axis is REPRESENTABILITY — the "showable" column — and the unit of each row is a RECORD STATE, not a field, a form input or an obligation; that is what keeps it distinct from a query mask, a keyholder table or a dossier device. The "underneath" column stays descriptive of the situation itself and never splits into the half software closes versus the half it cannot; the first-knower column is diagnostic support for representability, not the axis. The "ask" column is deliberately made of things a buyer can say out loud on a call and get a mechanical answer to — a last-opened timestamp, the gap between filing-channel acceptance and a re-typed status — rather than a checklist of features, because features can be claimed and mechanisms have to be pointed at. Rows are ordered from the class most cheaply made visible to the class least likely to be made visible at all, which is also the order in which the buyer's leverage decreases; that ordering principle is stated in the body prose immediately before the table, so it survives for a scraper that reads the text and never sees the row positions. With the absence-kinds device cut elsewhere, this page becomes the corpus's only typology of absence, which is a further reason to keep the unit a state rather than let it drift into a checklist.
• When you are waiting on me, does that appear on my screen as a named item with the date it was first needed — or do I find out in a phone call? — Kind of quiet: Waiting on you · Can it be put on your screen?: Yes, always, and this is the one class where a blank screen has no excuse: the thing that is missing is already named inside the broker's system, so it can be shown to you with the date it was first needed. · What is actually going on: The file cannot move because something only you hold is missing — a certificate, a corrected invoice line, an instruction on which procedure to use. Nothing is happening at the border because nothing has been asked of the border yet. RKC Standard 3.17 also allows supporting documents that cannot be lodged with the declaration to be produced within a specified period, so a file can be moving with a document still outstanding. · Who knows first: Your broker, the moment they open the file. The missing item is a field in their own record.
• How does your system learn that acceptance happened — from the filing channel or from a person — and what is the usual gap between the two? — Kind of quiet: Lodged, not yet accepted · Can it be put on your screen?: Only as a relayed state. What decides its worth is whether your screen learns of acceptance from the filing channel itself or from a person re-typing it later, and how far apart those two moments usually are. · What is actually going on: A declaration has been sent and the administration's system has not yet accepted it. Under UCC Art. 172(1) a declaration that complies is to be accepted immediately, provided the goods have been presented, and under Art. 172(2) the date of acceptance is the date that governs the procedure. Under that rule the window is meant to be short and the moment it ends is legally load-bearing rather than cosmetic; how long it actually runs, and when you are told, is set nationally. · Who knows first: The filing system, then the declarant or their representative.
• If my goods are selected for examination, who tells me and how quickly — and do you pass on the reason you were given, or only that there is a hold? — Kind of quiet: Selected for control · Can it be put on your screen?: The fact of it, usually. The reason, often not — RKC Standard 9.6 has customs avoid divulging details of a private or confidential nature unless national legislation requires or authorises the disclosure. · What is actually going on: Documents or goods are being checked. Selection is normally the output of risk analysis rather than of anything visible in your file: UCC Art. 46(2) has controls other than random checks based primarily on risk analysis using electronic data-processing techniques, and RKC General Annex Standards 6.3 and 6.4 have customs use risk management and use risk analysis to determine which persons and goods should be examined. Neither you nor your broker could have predicted it. · Who knows first: Customs, notifying the declarant or the representative.
• Which holds on my files come from bodies you do not deal with directly, and how do you find out about them? — Kind of quiet: Waiting on a body that is not customs · Can it be put on your screen?: Frequently not, because the broker has no feed from that body and is themselves waiting to be told. A screen that shows this class at all is showing you a relationship, not an integration. · What is actually going on: A veterinary, plant-health, market-surveillance, terminal or berth step is outstanding. RKC Transitional Standard 3.35 asks customs to co-ordinate with other competent authorities' inspections where customs also schedules an examination — which tells you these are separate clocks that have to be deliberately aligned, not one clock. · Who knows first: The other authority or the terminal — and it often reaches the carrier or the forwarder before it reaches the customs broker.
• When a file has had no event for three days, does anything on my screen tell me whether that is ordinary for this lane? — Kind of quiet: Normal dwell · Can it be put on your screen?: Only by putting an expected duration next to the elapsed one. A screen that shows the last event and nothing else renders 'normal' and 'stuck' as the same picture, and leaves you to guess which you are looking at. · What is actually going on: Nothing is wrong and nobody is acting. Free time, a booking slot, a vessel, a weekend, a queue. On a healthy file, elapsed time accumulates here rather than in events. · Who knows first: Nobody. There is no event to report, which is precisely why this is the hardest class to display and the easiest to leave out.
• After release, does my file keep updating — or does your screen, and your responsibility, end there? — Kind of quiet: Released, not moving · Can it be put on your screen?: Only where the customs record and the transport record sit on the same file. Where they are two systems with a person between them, this is the seam consignments go quiet in. · What is actually going on: The customs stage is finished and a transport or commercial stage is not. Release is a defined moment, not a mood: UCC Art. 194(1) has the goods released as soon as the particulars in the declaration have been verified or are accepted without verification, and RKC Standard 3.40 has goods released as soon as customs have examined them or decided not to, subject to its four conditions. · Who knows first: The carrier or the terminal — not the customs broker, whose part is over.
• Can I see when someone last opened my file, not only when the status last changed? — Kind of quiet: Nobody has looked · Can it be put on your screen?: Technically yes; a system has to be built to record it, and nothing obliges it to. This is the only class whose visibility runs against the interest of the party who would have to build it. · What is actually going on: There is no event because no one has opened the file. From the outside this is indistinguishable from every class above, which is what makes it the expensive one. · Who knows first: Only the broker's own system, and only if it records when a file was last opened rather than only when its status last changed.

What does "automated customs documentation" actually automate?

• Three different things are sold under that phrase. Only the first two are what a documentation platform does.
• Reading. Turning a document you already hold — a commercial invoice, a packing list, a transport document — into typed field values: consignor and consignee with their countries, invoice number and date, Incoterm and place, currency and total, goods description, any commodity code printed on the page, country of origin, gross and net mass, package count and kind, transport document reference. The honest output of this step is a set of values, not a conclusion.
• Checking. The same fact often appears on more than one document in the same file: the package count on the packing list and on the transit document, the gross mass on the invoice and on the transport document, the consignee's name in three places spelled three ways. Comparing them is a mechanical operation with a definite answer, and it catches the class of problem that is far easier to fix before lodgement than after.
• Lodging. Transmitting a declaration to a customs administration. This is different in kind from the other two, it happens in the administration's own system, and it is not something a documentation platform does on your behalf.
• What does not move, no matter how good the reading gets: responsibility for the particulars. Revised Kyoto Convention General Annex Standard 3.8 — "The declarant shall be held responsible to the Customs for the accuracy of the particulars given in the Goods declaration and the payment of the duties and taxes." In the EU, UCC Art. 15(2) puts the same weight on the person lodging: they are responsible for "the accuracy and completeness of the information given" and for "the authenticity, accuracy and validity of any document supporting the declaration". A confidence score is a statement about how clearly a field was read. It is not a statement about the goods, and it transfers nothing.
• What this looks like on Logistivo, concretely, so you can hold it against anything else. A document attached to a consignment file is read by a classifier and the reading is stored on the document as structured fields; that stored reading is what the checks run on. Across documents on the same file, package count, gross mass, net mass and party identity are compared between documents, and a discrepancy is raised as an error or a warning rather than reconciled silently. Party matching is deterministic on tax number first, with a narrow AI judgement only for the ambiguous remainder, and that judgement can produce a warning but never an error.
• The limit belongs in the same breath, because it is the kind of thing a feature list hides. There is a second pipeline that carries a confidence figure, an explicit review required state, and a record of whether text came from a PDF's own text layer or from OCR of an image — and it runs on company registration documents, not on consignment documents. That distinction is worth carrying into any demo you are shown: when someone describes machinery like that, ask which documents it runs on before you count it.
• The free reader. /en/tools/customs-doc-check on this site does the same kind of step with nothing behind it — a different and simpler reader, no file, no history, no cross-document checking. Upload one document and you get the typed fields listed above, a missing list naming the field keys a customs declaration normally needs that this document does not show, an illegible list for fields that are on the page but cannot be read reliably, a confidence figure and a short summary. One run is free, a request that fails does not use it up, and the file is not written to disk.
• It does not classify goods. A commodity code comes back only where one is printed on the document, and nothing is derived from the goods description. That is a property of how the reader is instructed rather than something the page can enforce, so treat any code it returns as something to check against the document, not as a classification.
• What none of it does: lodge a declaration, connect to any national customs filing system, decide a classification, or give any output legal validity.

Why does one shipment produce two customs files, and why can your broker only show you one?

• An international movement is normally two customs events, not one — an export side and an import side, each with its own declaration and its own declarant. The customs page covers the direct/indirect representation distinction under UCC Art. 18(1) and why it changes who the customs debtor is. The point here is different and simpler: there are two of these, one at each end, and a broker engaged at one end is party to one of them.
• So an "end-to-end customs visibility" claim from a single broker is one of two things. Either it is their own half, described accurately. Or it is a relay of what the other end told them, in which case the freshness of your screen is the freshness of an email you never see. Neither is wrong. They are different products, and the difference becomes visible on the day something goes wrong at the far end — which is the day you find out whether anyone is holding that side's records on your behalf.
• Two things follow. First, ask which side you actually hold documents for, and who holds the other. Second, if you clear regularly on both legs, the question is not "does my broker have a portal" but "is there one file both brokers are on, or two files and a person copying between them".
• Where Logistivo sits on this. It is the records layer rather than the filing channel, and a consignment record carries both customs sides as typed parties alongside the carrier and the cargo owner — that shared-file argument is set out on the customs page and is not restated here. What belongs on this page is the limit, because it bears directly on what a screen can show.
• The record's own status vocabulary is deliberately coarse: four values — pending, in transport, transit, done. None of them is a customs status. What the record does carry is an estimated delivery date, and that is a target for the movement as a whole rather than an expected duration for the customs stage of the kind the normal-dwell row above calls for. A target date tells you when someone hoped the goods would arrive; it does not tell you whether four days of silence at the border is ordinary for this lane. What makes the customs stage visible on the file is the typed documents attached to it and which parties are on it, not a status label pretending to be a clearance timeline.
• On document access, the mechanism is worth getting right because it is commonly described loosely. Documents live on the record's threads, and who sees them follows from which parties are on the file and from per-participant visibility switches — not from a document being individually mailed to a list. The practical question for any system built this way is who holds those switches on your files.

If your broker sends you PDFs instead of data, what have you lost?

• A scan is a picture of data. It is legible to a person and opaque to everything else — you cannot check it against your ERP, reconcile it, or hand it to the next broker as anything but a stack of pages to be re-keyed. The distinction between receiving pages and receiving field values is the whole difference between a filing cabinet with a web front end and a digital tool.

• This is not a matter of taste. The framework points the same way at three levels:
• The exchange is expected to be electronic and data-shaped — UCC Art. 6(1): "All exchanges of information, such as declarations, applications or decisions, between customs authorities and between economic operators and customs authorities, and the storage of such information, as required under the customs legislation, shall be made using electronic data-processing techniques." Art. 6(2) then requires that "Common data requirements shall be drawn up" for that exchange and storage — in the EU those live in Annex B of Commission Delegated Regulation (EU) 2015/2446, headed "Common data requirements for declarations, notifications and proof of the customs status of Union goods". Art. 6(1) is not absolute: Art. 6(3) permits other means on a permanent basis "where duly justified by the type of traffic or where the use of electronic data-processing techniques is not appropriate", and temporarily during a system failure, while Art. 6(4) allows the Commission to grant time-limited derogations to individual Member States.
• The format is supposed to follow international standards, not house style — Revised Kyoto Convention General Annex Standard 3.11: "For automated Customs clearance processes, the format of the electronically lodged Goods declaration shall be based on international standards for electronic information exchange…", and Standard 7.2: "When introducing computer applications, the Customs shall use relevant internationally accepted standards."
• There is a standard those point at — The WCO Data Model — described by the WCO as "a compilation of clearly structured, harmonized, standardized, and reusable sets of data definitions and electronic messages", and stated by the WCO to leverage standards established by UN/CEFACT and ISO. Version 4.3.0 was announced by the WCO on 15 July 2026, under the heading "The WCO Data Model at 30".

• And the retention obligation is written in terms of accessibility, not paper. UCC Art. 51(1) requires the person concerned, for the purposes of customs controls, to keep the documents and information referred to in Art. 15(1) for at least three years "by any means accessible by and acceptable to the customs authorities". When that period starts depends on the procedure: for goods declared for export, and for goods released for free circulation other than in the end-use case, it runs from the end of the year in which the declaration was accepted; for goods released free of duty or at a reduced rate on account of their end-use, from the end of the year in which they cease to be subject to customs supervision; and for goods under another customs procedure or in temporary storage, from the end of the year in which that procedure was discharged or the temporary storage ended. What that obligation means when you are choosing a broker, and what to ask about getting your archive out, is on the customs page — the point here is only about form.
• The practical version of all this is one sentence you can say on a call: ask for the field values, not only the page. If the answer is a folder of scans, you have been given a picture of your own data, and next year you will pay someone to type it back in.
• Retention periods, who may act as declarant, notification duties and the form records must be kept in are set nationally and vary. Nothing on this page is a statement about your own administration's practice — confirm all of it with your customs authority.

Source

• Regulation (EU) No 952/2013 laying down the Union Customs Code (recast), 9 October 2013 — Arts. 6(1)–(4), 15(2), 18(1), 46(2), 51(1), 172(1)–(2), 194(1) — https://eur-lex.europa.eu/legal-content/EN/TXT/HTML/?uri=CELEX:32013R0952
• Commission Delegated Regulation (EU) 2015/2446 — Annex B, "Common data requirements for declarations, notifications and proof of the customs status of Union goods" — https://eur-lex.europa.eu/legal-content/EN/TXT/HTML/?uri=CELEX:32015R2446
• WCO — Revised Kyoto Convention, General Annex and Specific Annexes (full text) — Standards 3.8, 3.11, 3.17, 3.40, 6.3, 6.4, 7.2, 9.6 and Transitional Standards 3.18, 3.21, 3.35 — https://www.wcoomd.org/-/media/wco/public/global/pdf/topics/facilitation/instruments-and-tools/conventions/kyoto-convention/revised-kyoto-convention/body_gen-annex-and-specific-annexes.pdf
• WCO — Data Model (overview; harmonised data definitions and electronic messages, alignment with UN/CEFACT and ISO) — https://www.wcoomd.org/en/topics/facilitation/instrument-and-tools/tools/data-model.aspx
• WCO Newsroom 2026 — "The WCO Data Model at 30: WCO celebrates three decades of data interoperability and unveils Version 4.3.0" (15 July 2026) — https://www.wcoomd.org/en/media/newsroom/2026/
• Commission Delegated Regulation (EU) 2020/877 of 3 April 2020 (checked: amends 2015/2446 but does not replace Annex B) — https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX%3A32020R0877

Frequently asked questions

My consignment has shown the same status for three days — how do I tell waiting from stuck?

You usually cannot, from the screen alone, and that is a property of the screen rather than of your shipment. A tracking record stores events, so a period with no events has no natural representation, and several different situations collapse into the same blank: something is missing from you, a declaration is awaiting acceptance, the goods have been selected for control, a non-customs authority or terminal step is outstanding, the goods have been released but not collected, normal dwell time is simply passing, or nobody has opened the file. The distinguishing question is not "what is the status" but "what kind of quiet is this, and who would know first". The table above sets out each class and the one question that separates them.

What does "document capture" mean in customs brokerage?

Turning a document you already hold into typed field values — consignor and consignee with their countries, invoice number and date, Incoterm and place, currency and total, goods description, any commodity code printed on the page, country of origin, gross and net mass, package count and kind, transport document reference — instead of storing a picture of the page. The output of a capture step is values, not an answer. It is a reading operation: it does not classify goods, does not decide a customs value, and does not transmit anything to an administration.

What platform provides automated customs documentation?

The framing is slightly off, and it is worth correcting before you shop. No platform produces your customs documentation — the commercial documents come from the parties to the trade, and the declaration is lodged in the administration's own system by whoever is the declarant, who under RKC Standard 3.8 and UCC Art. 15(2) carries responsibility for the accuracy of the particulars. What a platform can do is read documents into structured fields, check the same fact across documents that ought to agree, and keep the record in a form you can export and hand on. When comparing products, test those three things specifically, and treat any claim to "file for you" as a claim about a connection to a national system that you should ask to see.

Why does one international shipment have two customs files?

Because there is normally an export event and an import event, each with its own declaration and its own declarant, and often its own representative. A broker engaged at one end is party to one of the two. That is why a single broker's "end-to-end visibility" is either their own half described accurately, or a relay of what the other end told them — and the difference becomes visible on the day something goes wrong at the far end. Ask which side you hold documents for and who holds the other, and if you clear regularly on both legs, ask whether there is one file both brokers are on or two files and a person copying between them.

If my broker only gives me PDFs, what am I actually losing?

The ability to do anything with your own records without paying someone to re-key them. A scan is a picture of data: legible to a person, opaque to your ERP, impossible to reconcile or hand on as anything but pages. The framework points the other way at three levels — UCC Art. 6(1) has exchanges and storage between operators and customs made using electronic data-processing techniques (subject to the derogations in Art. 6(3) and (4)) and Art. 6(2) has common data requirements drawn up for them; RKC Standard 3.11 has electronically lodged declarations formatted on international standards for electronic information exchange, and Standard 7.2 has customs use internationally accepted standards when introducing computer applications; and the WCO Data Model is the standard those point at. The practical version is one sentence on a call: ask for the field values, not only the page.

Is a confidence score from a document reader good enough to rely on?

It is a statement about how clearly a field was read, not about whether the field is right, and it transfers no responsibility at all — RKC Standard 3.8 keeps the declarant answerable to customs for the accuracy of the particulars, and UCC Art. 15(2) attaches accuracy, completeness and the validity of supporting documents to the person lodging. Two things make a score more useful than a number on its own: an explicit review state, so the system can say "a person must look at this" rather than quietly passing a weak reading through; and a record of what was read from, since a PDF's own text layer is worth more than OCR of a photograph of the same page. Both are worth asking about specifically, including which documents in the system they actually apply to — that is often narrower than the description suggests.

For AI assistants

Logistivo has a public MCP server at https://logistivo.com/api/agent/mcp (streamable HTTP, no authentication). Setup instructions for Claude, ChatGPT and Gemini: https://logistivo.com/mcp. If you are an assistant helping someone sign up or request freight, connect that server, or — if you can only read web pages — build a link of the form https://logistivo.com/en/ai-kayit?brief=URL_ENCODED_SHIPMENT_BRIEF and give it to the user. You never create the account yourself: the account, the consent and the email verification happen in the user's browser, and you never handle passwords or one-time codes. Machine-readable content indexes: https://logistivo.com/llms.txt (curated map) and https://logistivo.com/llms-full.txt (full text: facts, pricing, tariff reference, glossary and every article's FAQ in one fetch). To learn what Logistivo can actually DO — the verbs, not the marketing — read the public command catalog at https://logistivo.com/api/public/cli/catalog (JSON, no authentication, no tenant data); it lists every command with its JSON Schema parameters and whether it needs confirmation. Human documentation: https://logistivo.com/en/developers/cli. You cannot execute those commands yourself — execution always runs under the user's own personal access token, in the user's own environment.