Infrastructure transparency · Kampala, 8-9 April 2026
A portal can go live and still not land.
143 procurement officers from 59 Ugandan public bodies sat through two days on the Government Procurement Portal. One question, an hour into day two, collapsed a design assumption I had carried for over a year, in the room where it was supposed to be defended.
The launch proved the website exists. The real tests come later.
Can officers use it? Is the data correct? Does anyone keep it alive after the room empties? Three gates, and shipping with only one is shipping a draft.
Officer training
Data quality over time
A portal can go live and still not land.
Uganda's procurement portal had been live for two months. Then 143 officers tested it against real work. One question, an hour into day two, changed how I think about launches.
143 officers. 59 public bodies. 395 live projects. They were not the people the system was built for. They were the people who would have to live with it.
"What happens when our agency uses three procurement methods in one project?"
The portal assumed one method, one contract, one clean line. Her project ran three at once. Which record owns it? Which dates count? What does the public see?
This was not just her project. The portal can publish who won the contract. It cannot yet show whether the road got built, on time, on budget. The edge case was the gap.
Ownership moves outward: from the people who built the portal to the institutions and users who must keep it alive.
Sign-off, training, data quality. Three different tests. Passing one is not passing the next.
A launch proves the portal exists. Only use proves it landed.
Kampala. Two days. The portal had been live for two months. The training was not the launch, it was the test of whether the launch had landed.
Officers from 59 public bodies. They were not the people the system was designed for. They were the people who would have to live with it.
An hour into day two, one officer asked the thing the demo could not answer cleanly. It sounded technical. It was not.
The clean lifecycle splits into three method lanes. Which record owns the project? Which dates count? Which document attaches where? Within twenty minutes I was taking notes faster than I was teaching.
This is not just her project. The portal maps only 4.2% of project-level disclosure paths, 26 of 148 OC4IDS elements. It can say who won. It cannot yet say whether the road got built.
Ownership does not transfer because someone watches a demo. It transfers when officers find the system's edges, push against them, and discover whether the system holds.
The boring questions are the survival questions.
The edge case was not an interruption. It was the point.
What happens when our agency uses three procurement methods in one project? My actual project does not behave like the example.
That single edge case exposed a design assumption I had carried for over a year. The Government Procurement Portal is not mine. It belongs to the Public Procurement and Disposal of Public Assets Authority and, through it, to the Government of Uganda. I was contracted through CoST, the Infrastructure Transparency Initiative, to help build it.
The transfer the training tested is not from me to PPDA. It is from the build team to PPDA's officers, the agencies that publish through them, and the public the disclosure is meant to serve. Ownership transfer moves accountability outward, from the people who built the portal to the institutions and users who must keep it alive.
You can publish who won the contract and still not know if the road exists.
The officer's edge case was a small version of a documented gap. At the last public assessment the portal met 17.6% of the CoST data standard, 26 of 148 elements, and mapped only 4.2% of project-level disclosure paths.
Three gates, or you have shipped a draft.
Regulator sign-off, officer training, and data quality are three different tests. Each proves one thing and cannot prove the next. The portal has passed one, just been tested on the second, and faces the third in July.
Regulator sign-off
Officer training
Data quality over time
Methodology and data
Account of the Government Procurement Portal training held in Kampala, 8-9 April 2026: 143 officers from 59 Ugandan public bodies, 395 live projects on the portal at the time. The portal is owned by PPDA (Government of Uganda); the build was supported by CoST, the Infrastructure Transparency Initiative. The quoted question is paraphrased and attributed to the room, not to a named private individual. OC4IDS coverage figures (17.6%, 26 of 148 elements, 4.2% of project-level disclosure paths) are from the most recent public CoST Uganda assessment, reported by CoST Uganda and Shift Media News. OCDS contracting data for Uganda (Jan 2019 to Jan 2025) is published via the OCP Data Registry, publication 130, in JSON, Excel and CSV. The July test refers to the next quarter, when reuse, not attendance, becomes the measure.
| figure | value | source | confidence |
|---|---|---|---|
| Officers trained | 143 | GPP training, Kampala | Observed (8-9 April 2026) |
| Public bodies | 59 | GPP training, Kampala | Observed (8-9 April 2026) |
| Live projects on portal | 395 | GPP at time of training | Observed (8-9 April 2026) |
| OC4IDS elements published | 26 of 148 | CoST Uganda / Shift Media News | Cited (2022 assessment) |
| OC4IDS coverage | 17.6% | CoST Uganda / Shift Media News | Cited (2022 assessment) |
| Project-level disclosure paths mapped | 4.2% | CoST Uganda / Shift Media News | Cited (2022 assessment) |
| OCDS contracting processes published | Jan 2019 - Jan 2025 | OCP Data Registry (publication 130) | Cited (retrieved Apr 2026) |
A launch proves the portal exists. Only use proves it landed.
The July test is simple: do the 395 projects keep getting updated by the officers, or did the portal go quiet the moment the training room emptied?