After processing 100,000 procurement contracts through Uganda's Public Procurement and Disposal of Public Assets Authority, I arrived at a conclusion that changed how I approach this work entirely: disclosure is not transparency. Governments publish thousands of documents, declare the job done, and move on. But a signed contract tells a citizen nothing about whether a school gets built, whether a road holds together after the first rainy season, or whether public money was spent on anything resembling what was promised. Structured, connected data does. That distinction took me 100,000 contracts to fully understand.
The Data Stops at the Signature
Working through Uganda's PPDA dataset, the first pattern was unmistakable. The procurement records were often detailed up to the point of contract award. You could see which company won the tender, the awarded value, and the date of signature. Then the trail went cold. When a contractor walked off a site, the procurement record still showed the project as active and funded. There was no mechanism in the data to surface that gap.
This happens because most governments build their transparency systems around the Open Contracting Data Standard, which is designed to track the purchasing process from planning through to contract award. OCDS does that well. It answers who won the bid and at what price. But infrastructure delivery does not end at signature. It runs through site mobilisation, variation orders, material testing, structural delays, and a final completion certificate. OCDS was never designed to capture that journey, and asking it to do so produces a false picture of accountability.
At CoST, we use OC4IDS, the Open Contracting for Infrastructure Data Standard, precisely because it was built to track public works projects across their full lifecycle, from project identification through to completion. OC4IDS connects the original contract to on-site progress, formal variations that change scope or cost, and physical delivery milestones. Without that connection, you are tracking promises. With it, you are tracking physical reality. Those are not the same thing.
Document Dumping Is Not Data
The second pattern the Uganda dataset exposed was widespread document dumping. The majority of disclosures existed as scanned PDFs. A scanned document is a dead end for accountability. Civil society organisations cannot aggregate figures across a scanned page. A journalist cannot filter a folder of image files by contractor name and contract value. Researchers cannot run any analysis at all. The documents exist, technically satisfying a disclosure obligation, while actively resisting scrutiny.
Some will argue that publishing documents is a meaningful first step, and that is a fair point. Releasing any information is better than releasing none. But a first step that looks like a finish line actively slows progress. When a government points to a repository of uploaded PDFs as evidence of transparency, it creates the political cover to stop investing in the structured systems that would generate real accountability. I have seen that dynamic play out, and it is difficult to reverse once it takes hold.
Structured data under OC4IDS changes the accountability surface entirely. When contract values, contractor identities, project locations, and completion dates exist as machine-readable fields rather than image files, civil society groups can ask substantive questions at scale. They can identify which contractors consistently run over budget, which districts consistently receive incomplete infrastructure, and which project types carry the highest risk of abandonment. That analysis is impossible when the data lives in a scanned document.
Systems That Do Not Talk to Each Other
The third pattern was fragmentation. Uganda's procurement data sat in one system. Financial disbursement data sat in another. Physical progress information, where it existed at all, sat somewhere else entirely. No single actor could see the full picture without manually reconciling records across systems. That reconciliation work is slow, expensive, and rarely done.
Integrated Financial Management Information Systems, known as IFMIS, hold a significant part of the answer. When IFMIS data is connected to OC4IDS project records, the payment trail becomes visible alongside the physical delivery trail. A disbursement that precedes any recorded site activity becomes a flag rather than a footnote. CoST has worked on precisely this kind of IFMIS integration because the accountability value of procurement data multiplies when it is linked to financial and physical progress records rather than standing alone.
The CoST Independent Review mechanism reinforces this integration by bringing trained assessors to examine disclosed data against physical reality on the ground. The review is not a document audit. It is a structured process that tests whether what is recorded matches what was built. That connection between disclosed data and physical verification is what makes CoST's approach distinct from a compliance exercise.
No single actor sees the full picture. Reconciliation is manual, expensive, and rarely done. That is what fragmentation costs in practice.
One Step to Take Now
If you are working on procurement transparency and your current disclosure system produces PDFs, start by mapping where your data actually lives. Identify which fields, contractor name, contract value, project location, completion date, already exist in structured form inside your procurement or financial management system. Those fields are the foundation for OC4IDS adoption. You do not need to rebuild everything at once. You need to stop treating document publication as the destination and start treating structured, connected data as the baseline. One hundred thousand contracts taught me that the gap between those two positions is where accountability goes to disappear.
Found this useful?
I write about open data systems, transparency, and implementation.
Read more articles →