The record is not the verdict.
A published contract can look like accountability. It names the project, names the supplier, gives you a value, even shows the procurement method. But a visible contract is not yet a knowable project. To know what the record is worth, you have to ask what it can prove, what it cannot answer, and what someone has chosen to keep locked.
The mistake is not transparency. It is premature certainty.
Nobody serious thinks publication makes a project good. The real error is quieter: we treat a published record as if it has already performed accountability. Here is the inference, and where it breaks.
Each arrow is assumed, not earned. The record has not yet answered who controls the supplier, what changed after award, whether the work was completed, whether the asset was maintained, or whether any of it can be checked.
Publication changes the question. It does not end the inquiry. It moves a project from un-checkable to check-able, and the check still has to be run.
A visible contract is not yet a knowable project.
Run six tests against the record.
Not against the road. Against the record. Each row is one question the public should be able to answer, and its result. The states are defined in the key below; tap any row to read why.
This inspection is built from Zambia's OC4IDS mapping. It characterises what the public record, as currently structured, lets anyone check across the mapped system. It is a representative inspection record, not an accusation against one named road.
Bright at award. Dark after it.
Zambia checked its public record against the OC4IDS standard, phase by phase. Disclosure is strongest at procurement and collapses through delivery, completion and maintenance, where the risk actually lives. The gap is not random. It has a shape.
Coverage is fields mapped against applicable OC4IDS slots per phase, from the CoST Zambia field-level mapping (76 records, FY2023–2024). After award, the record thins. The risk does not. And the gaps are not all the same failure: in four of the seven phases the cause is "collected but not yet disclosed" — the data already sits inside government. Maintenance and decommissioning score zero because no reviewed system collects them at all.
Different gaps need different fixes. Naming the gap names the reform.
The supplier is named. The beneficiary is not.
The record can tell you the contractor's name. It can give you a registration number, an address, a contract value, even a capability grade from one to six.
Those fields answer one kind of question: who is legally present in the procurement record? They do not answer the ownership question: who controls the supplier, who benefits from the contract, and whether that interest is connected to the decision. In Zambia's mapping, the Parties sheet fills 18 of 968 slots, and beneficial ownership is not collected in any reviewed system.
That is why this test is locked, not merely fogged. The missing answer is not a blank field someone forgot to fill. It is a disclosure-regime problem: the cause is recorded as restricted by law or workflow, and the fix is policy, not data entry.
Qualification answers "can they build?" Ownership answers "who benefits if they do?"
This locked test is the reason the ownership x-ray exists. A supplier name is where the record starts, not where accountability ends. The Name on the Certificate Is Not the Owner →
A record lets you inspect a case. Usable data lets you inspect a system.
One disclosed record can tell you what happened here. It cannot tell you whether "here" is normal. The questions that matter most are system-level, and they need more than one record.
Useful. But one record cannot tell you whether this is normal or an outlier.
These need records structured the same way, linked by reliable identifiers, available in bulk. That is usability. Not a nicer portal. The ability to compare.
Zambia shows how a record can be bright at award and dark after it. Uganda shows the same problem at portal scale: procurement visibility exists, but project-level accountability still depends on structured lifecycle data the portal does not yet carry.
The portal can name who won which contract. It cannot yet tell you whether the works were finished on time, within budget, or to the commitments made. Publication makes a record visible. Usability makes records comparable.
Take a single test deeper, on a real record.
Each essay below picks one of the six tests and runs it all the way down on a real disclosure record. Pick the question you want to chase.
The Name on the Certificate Is Not the Owner
The supplier is named, sharp and certain. Who controls it stays locked. The ownership x-ray, in full.
Open the x-ray →After the Award
The record is brightest at the moment of award and thins after it, while the risk runs on through delivery and maintenance.
Follow the record →The Road May Be Complete. The Record Is Not.
Malawi marks a road complete, priced, finished. The proof you would need to check it is missing. Across all 162 projects.
Run the scanner →Methodology, sources and the honest limits
Anchor data. The six tests and the lifecycle bars are built from the CoST Zambia OC4IDS v0.9.5 field-level mapping (model report, 20 May 2026, published under CC BY 4.0). The team analysed a purposive sample of 76 procurement records on the ZPPA e-GP platform for FY2023–2024, spot-checked against live records on 17, 18 and 20 February 2026. Verified figures used here: 35 OC4IDS fields publishable today; 55 within 12 months; roughly 25 fields needing policy action, including beneficial ownership, maintenance and decommissioning. The Parties sheet fills 18 of 968 slots. Lifecycle coverage by phase: Identification 28%, Preparation 12%, Procurement 21%, Implementation about 2%, Completion 8%, Maintenance 0%, Decommissioning 0%.
The subject record. The inspection is a representative inspection record: it characterises what the public record, as currently structured, lets anyone check across the mapped system. It is not a verdict on, or an accusation against, any single named project. Individual test outcomes follow the mapping's phase-level and sheet-level findings.
Supporting data. The portfolio figures are from Uganda's Government Procurement Portal (PPDA): it meets 17.6% of OC4IDS requirements, 26 of 148 data elements, and 4.2% of project-level disclosure paths (CoST Uganda and Shift Media News assessment, 2022).
The three states. Sharp, fogged and locked describe testability, not quality. Sharp: published, structured and usable. Fogged: missing, not collected, held only as unstructured documents, or impossible to link. Locked: present somewhere but restricted by law, policy or workflow. They are deliberately not a good/bad colour scheme; a record can be sharp on every test and still describe a bad road, which is the whole point.
What this essay does not claim. It does not argue that open data prevents corruption, or that more transparency produces better projects. It argues only that disclosure is the precondition for testing a project, and that the test still has to be run. Source-brief references to external standards (BODS, FATF Recommendation 24, Open Ownership) are reflected as general principle, not quoted as findings, because they sit outside the verified materials behind this build.
Transparency does not make a project good. It lets you find out whether it is.
A published contract is a receipt, not an audit. The honest verdict is never "this project is good." It is "this project can now be tested." A smaller claim, and a far harder one to take away from you.
Anchor data: CoST Zambia OC4IDS mapping · CC BY 4.0 | Portfolio data: Uganda GPP / PPDA · CoST Uganda 2022 | Standard: OC4IDS
Analysis, derived framing, visual design and essay: Michael Cengkuru · Published 27 Jun 2026