Somewhere in Africa, a truck waits at a border while a declaration already in the system is retyped. The customs system behind it knows the cargo, the consignee and the duty paid. The system 50 meters away, across the line, does not know of it. What one knows about a consignment rarely follows it to the next. The real database still travels in a folder on the dashboard.

This is Africa in 2026. In February 2025, the African Union adopted eight annexes to the AfCFTA Protocol on Digital Trade, covering cross-border data transfers, digital identities, digital payments and emerging technologies. Customs cooperation already has a legal footing in the AfCFTA’s Protocol on Trade in Goods, whose annexes on customs cooperation and mutual administrative assistance and on trade facilitation bind State Parties to work together at the border. Much of this still has to be ratified and written into national law. The rules are largely drafted. The wires are not.

The gap sits between the rulebook and the wire. Protocols are negotiated in conference rooms, but clearance happens at border posts with systems that were never designed to talk to each other. Closing this gap is the real test of the digital trade era.

It’s easy to blame technology, but interoperability fails more for lack of legal frameworks than technical tools: agreements to recognise and accept each other’s data, mutual recognition of Authorised Economic Operator status, and data protection adequacy. Rules of origin show how wide the gap between agreement and application can be. In February 2026 the AU Assembly adopted the last outstanding rules, for automotive products, clothing and textiles, completing the AfCFTA’s origin rules on paper. A rule helps a trader only once it sits in national tariff schedules, customs systems and certificate checks, and across much of the continent that work is still under way. Technology follows the legal mandate; once it exists, tech determines if the rulebook delivers cleared containers or just communiqués.

Each country has its own procedures, data standards and systems, built at different times to different specifications. The company I represent built several, giving me direct insight into the issue: a system optimised for one administration’s revenue and compliance needs rarely works for its neighbour. Traders crossing several borders pay the price in duplicate declarations and delays.

The potential is huge: UNCTAD estimates that tackling non-tariff barriers could generate about $20 billion a year for African economies, more than five times the $3.6 billion expected from eliminating tariffs.

Why single windows stop at the border

Many countries have built national single windows, achieving real domestic gains. But without an outbound layer to pass declaration data to the next customs administration, the process breaks down at the border, reverting to manual checks.

A second limit receives less attention. A Single Window is not a customs window. Much of the delay sits with agencies that are outside Customs: standards bodies, food and drug regulators, phytosanitary inspection, and port health. And this is significant. The World Customs Organization (WCO) Time Release Study shows where delays occur; every interoperability programme should start and end with a TRS.

The bilateral trap

It would be wrong to suggest Africa has not solved cross-border data exchange. It has, bilaterally, and repeatedly. The ECOWAS SIGMAT transit system was operating in nine member states by May 2025, when the Benin-Nigeria interconnection launched at Sèmè-Kraké. On 11 September 2026, the same two administrations went further. Their Customs Declaration Exchange System pilot, linking Benin’s Customs Webb with Nigeria’s B’Odogwu, lets officers in the destination country see a declaration lodged in the departure country, along with manifests, transit data and risk alerts. Each country keeps its own duty calculation, risk assessment and release. That is federation in practice, on the Abidjan-Lagos corridor that Nigerian Customs says carries about 70 per cent of West Africa’s transit trade. The East African Community’s Single Customs Territory shares declaration data and cargo tracking along the Northern and Central Corridors. COMESA’s transit guarantee moves goods under a single bond across multiple states. These are working systems, built by African institutions.

The problem is scale. Sèmè-Kraké took years of work to connect one pair of administrations. Fifty-four signatories generate 1,431 possible bilateral links, each separately negotiated, mapped and maintained, and re-tested whenever either side upgrades. Continental success requires one shared data definition, implemented once per country; one integration each, not one per pair.

A common language, and what technology can do for it

The WCO Data Model provides the vocabulary, code lists and message structures almost every African administration aims for. The difficulty is that national implementations diverge. Fields are added for local requirements, legacy structures go undocumented, and the knowledge is often lost. Mapping national systems to the Data Model is laborious and must be redone with every upgrade.

This is where machine learning helps: discovering candidate mappings between undocumented legacy systems and the Data Model, inferring field meanings, and keeping those mappings current as systems change. Today, this work is manual and slow, and the highest cost in interoperability programmes.

But AI should not rewrite declarations in flight. Declarations are legally binding and must be transformed in ways that are deterministic, versioned, logged and auditable. Use machine learning to build and maintain mappings, but let published, deterministic mappings execute them. Anything else asks a revenue authority to accept an unauditable component inside its assessment chain. The same discipline applies to exceptions, because mismatched records and disputed classifications are where the operational burdens lie.

Federation, and who owns it

Africa should not centralise its Customs data. Federation is the right architecture. Each country keeps its systems and data, connects through agreed standards and secure interfaces, and shares only what’s needed for each transaction. The reasons are legal as well as political. Customs data is fiscal data, subject to statutory secrecy and regulations in most jurisdictions.

Federation, however, is an architecture, not a governance model, and this is where the debate is thinnest. Someone needs the mandate to define “only what is necessary.” The AfCFTA Secretariat, with the WCO and the Regional Economic Communities (RECs), is the obvious home for a continental exchange standard and the dispute mechanism behind it.

Until empowered, every implementation is a bilateral negotiation with added technical complexity. Security must also be addressed as each new interconnection widens the attack surface of critical revenue infrastructure.

What to do next

The cautionary example is TradeLens. Backed by Maersk and IBM, blockchain-enabled and presented as an open, neutral industry platform, it closed in 2023. The technology worked, but participation never reached critical mass. Interoperability offers little value to the first mover: country A gains nothing until B, C and D connect. So, four priorities require an owner, a date and a measure:

1. Designate the standard. The AfCFTA Secretariat, WCO and RECs should publish a single continental profile of the WCO Data Model as the mandatory exchange format, with a conformance test. Until one profile is named, every country and vendor builds to a different target.

2. Build the legal gateway. A model bilateral data-exchange agreement and framework for mutual recognition of AEO status would remove the largest source of delay in these projects, which is legal rather than technical.

3. Fund and measure two corridors. Baseline two live corridors with a Time Release Study, connect them to the continental profile, and publish the results. A target such as data reused across the border in 60 per cent of consignments, and release time down by a third, is what attracts financing from the AfCFTA Adjustment Fund and the development banks. Pair it with PAPSS, so settlement matches the speed of the paperwork.

4. Bring the traders in, including the smallest. Broadband, power and current hardware at border posts are the prerequisites and must be costed as multi-year infrastructure. Much intra-African trade is informal, often by women small traders. Without simplified regimes and a low-value threshold, interoperability  will serve large shippers and formalise nobody.

The technology exists. Most standards are set, and the working examples are African. What’s missing is a named owner for the standard, cross-border legal instruments and proper funding for two corridors. That’s the code to crack. It is a governance problem in a technical disguise, and why Africa’s goods still move faster than their data.

​