Why Transparency Portals Die

The most successful infrastructure disclosure system I helped build stopped updating eighteen months after launch.

The technology worked. The dashboards looked good. The data standard was implemented correctly. Ministers and donors attended the launch. Two years later, the portal still loads. Nobody maintains it.

I've spent over a decade building these systems. Uganda's Government Procurement Portal publishes contracts from over 200 government entities. Kaduna State's platform in Nigeria tracks over 1,000 infrastructure projects. I've worked on implementations in Malawi, Mozambique, Zambia, and Thailand. Across all of it, I've helped publish data on tens of thousands of contracts and projects.

Here's what I've learned: solving the data structure problem doesn't mean you've built a system that survives.

1. The standard is not the system

Let me be precise. A data standard gives you a shared schema, comparability across contexts, and portability of tools. A disclosure system requires incentives, maintenance budgets, political cover, and users who depend on its survival.

OC4IDS is a standard. It emerged from CoST, the Infrastructure Transparency Initiative, and the Open Contracting Partnership. It defines 40 core disclosure items across the project lifecycle, from identification through procurement to completion and maintenance. It connects project-level information to contract-level procurement data.

The evidence for open contracting approaches is solid. Ukraine's ProZorro system, built on the Open Contracting Data Standard, has saved over $6 billion since 2016 and grown the domestic supplier pool from 14,000 to 140,000 (EBRD, 2024). Thailand's Comptroller General's Department found 8-9% lower bid prices on disclosed projects, roughly $70 million in savings in 2024 (CoST, 2024).

But ProZorro survived a revolution, a pandemic, and a war. That context is not replicable. Most disclosure systems face quieter deaths.

2. Three problems standards cannot solve

Every portal I've built faces the same three challenges. They're not technical.

The durability problem. Project funding ends. Staff rotate. Ministers change. The developer who understood the codebase takes a private sector job. Two years after launch, the portal runs but nobody maintains it. Data quality degrades. New agencies never get onboarded.

The test: Has the system survived at least one political transition? At year ten, has it been upgraded, or is it running on infrastructure nobody maintains? The question isn't whether the portal loads. It's whether anyone's job depends on keeping the data current.

I've watched portals launch with 200 projects, then show 200 projects three years later. Not because nothing was built. Because nobody was accountable for updating.

The political survival problem. Disclosure systems make powerful people uncomfortable. They reveal which contractors consistently win bids. They show which projects run over budget. They expose delays someone would prefer to keep quiet.

In Honduras, the World Bank reported that the now-defunct road fund (Fondo Vial) awarded contracts to firms run by a drug cartel. CoST's scrutiny of those contracts contributed to reviews and reforms. The disclosure approach worked. But working means creating enemies.

The question isn't whether your portal can publish embarrassing data. It's whether it will still be publishing next year.

The constituency problem. Most disclosure portals are built for imaginary users. We assume journalists will download the data. We assume civil society will run analyses. We assume citizens will demand accountability.

Ukraine built Dozorro, a monitoring platform with over 100,000 users per month. Of thousands of suspicious activity reports filed, roughly a quarter resulted in tenders being re-awarded (Dozorro, 2023). That's impact.

That's not typical. The more common pattern: usage spikes during a scandal, journalists write stories, then traffic flatlines. Civil society gets trained once. Nobody budgets for refreshers. The portal sits there. Data publishes. Nobody uses it. When nobody uses it, nobody defends it. When nobody defends it, the next minister quietly defunds it.

3. What survival requires

The next generation of disclosure systems won't succeed by adding data points. They'll succeed by building systems that create their own reasons to survive. Here's how.

Make disclosure a byproduct of existing work

The killer feature isn't transparency. It's automation.

If the portal auto-generates the compliance report the Minister demands every Friday, the procurement officer will use it. If it pre-populates the quarterly infrastructure update the Finance Ministry requires, the planning department will keep the data current. Transparency becomes the byproduct of making someone's job easier.

Thailand's implementation works because disclosure is integrated into workflows the Comptroller General's Department already runs. The data isn't extra work. It's how the system functions.

Compare that to portals where a procurement officer logs into a separate system, re-enters data entered elsewhere, and sees no personal benefit. In that design, disclosure is optional. Optional dies.

The practical test: Can you name the report your portal generates that someone would miss if the system went offline tomorrow? If you can't, you've built a ghost.

Implementation steps:

1. Map the reports your target agencies produce monthly and quarterly
2. Identify which data points overlap with your disclosure requirements
3. Design your system to generate those reports automatically from disclosed data
4. Make the disclosure interface the easiest path to compliance

Build constituencies that profit from the data

Who loses money if the portal dies? Find those people.

Ukraine's coalition emerged from political upheaval. You cannot manufacture that. But Colombia offers a replicable model.

When Colombia standardized tender documents for public transport works in 2019, the push came partly from the Society of Engineers and the Chamber of Infrastructure. These industry groups were concerned about single-bid awards and contracts tailored to favor specific firms. They weren't transparency advocates. They were businesses tired of losing to rigged competitions. Open data served their commercial interests.

That's a defender. Not because they care about accountability in the abstract, but because the data helps them forecast demand, identify opportunities, and challenge competitors who win through connections rather than competence.

Civil society matters. But the funding landscape for accountability work is shifting. This month, major donors announced plans to fund governments directly rather than through CSOs. Civil society organizations depend on grant cycles and donor priorities. That dependency is becoming more precarious.

Businesses operate on different incentives. A construction firm doesn't need a three-year grant to care about pipeline visibility. An industry association doesn't lose interest when a donor exits the sector.

Implementation steps:

1. List every potential commercial user of your data: materials suppliers, subcontractors, equipment vendors, industry associations, financial institutions assessing project risk
2. Interview five of them. Ask: "If you could see every infrastructure project in the pipeline six months ahead, how would you use that?"
3. Design data outputs for their use cases, not just oversight use cases
4. Track which commercial users access the data. When the system is threatened, call them

What if no commercial constituency exists? In markets dominated by state-connected firms who benefit from opacity, the commercial case may be weak. In that context, durability requires embedding disclosure in donor conditionality or regulatory frameworks that outlast political transitions. The system survives because removing it has external costs, not because users defend it.

Prove value in terms agencies recognize

Thailand can point to $70 million in estimated savings. That number ends arguments about whether disclosure is worth the effort.

Most disclosure systems measure outputs, not outcomes. Contracts published. Projects disclosed. Nobody in government cares about outputs. Without outcome evidence, disclosure remains expendable when budgets tighten.

What would change this: measure time saved on audit preparation. Count procurement complaints avoided because data was already public. Track whether disclosed projects have fewer cost overruns than undisclosed ones.

Implementation steps:

1. Before launch, establish baseline metrics: audit preparation time, procurement complaint volume, average project cost overrun
2. Build measurement into the system. Track these automatically.
3. Report savings in currency and time every quarter
4. Share savings reports with budget decision-makers, not just transparency advocates

4. The question for your next RFP

If you're writing a request for proposal for a disclosure system today, don't ask for a dashboard. Ask for a dependency map.

Ask: Who loses if this system goes offline?

Ask: What report stops getting generated?

Ask: Which workflow breaks?

Ask: What business loses its competitive intelligence?

If the vendor can't answer those questions, you're funding a monument, not infrastructure.

The standard was the easy part. Building systems that survive is harder. It's also the only work that matters.