A Data Space Is Not a Business — Why Commercial Viability Must Come First

The room agrees that the data should be shared. The manufacturer wants better visibility. Suppliers want less repetitive reporting. The technology team can describe a secure architecture. A funding programme can support the first phase. Everyone sees a reason to participate. Then someone asks: who will pay to keep it running?

This is a hypothetical scene, but a useful test of any proposal. Agreement that sharing is desirable does not establish a buyer, a recurring service or a sustainable operating model. A data space can provide valuable infrastructure without itself being a commercial business. Confusing those propositions makes an investment decision harder, not easier.

The starting question is not “How do we build a data space?” It is “Which decision or transaction becomes sufficiently better that someone will sustain the arrangement?”

What the OECD report finds — and what it does not prove

On 20 October 2026 the OECD published Data Spaces between Co-ordination Challenges, Financial Viability and Public Policy. It describes data spaces as trusted multi-party data sharing under federated governance and draws on analysis of more than 200 initiatives, case studies and an expert workshop.

Its executive summary identifies four conditions that, taken together, favour a data space over alternatives such as bilateral contracts, centralised platforms or privacy-enhancing technologies: sharing among three or more parties; competitive or jurisdictional relationships that preclude trusting a single intermediary; collective value that bilateral contracts alone cannot capture; and ongoing data flows that need continuous governance.

The report’s viability picture is sobering. Citing the International Data Spaces Association Radar, it reports 235 listed initiatives as of February 2026, of which only 25 had reached operational status and none the scaling stage it associates with financial viability. It adds that most depend on public funding.

It calls the central barrier a “bootstrapping problem”: shared infrastructure must precede the network effects that justify it. Where that was overcome, it points to quantifiable use-case value from the outset; it cites Agdatahub, which declared bankruptcy in 2024, as infrastructure built without sufficient demonstrated use-case value once grant funding ended.

These are the OECD’s figures and analysis, attributed rather than independently re-measured here. The registry counts listings, not audited outcomes, and the OECD itself notes that few initiatives collect the data needed to evaluate whether they deliver value alternatives could not. My interpretation is narrower: a data space merits consideration when the coordination problem genuinely requires it, and its additional costs must be justified by an outcome a simpler arrangement cannot deliver.

Separate the outcome from the institution

A buyer rarely starts by needing federated governance. A procurement director may need to approve a supplier without repeatedly requesting the same evidence. A logistics operator may need to resolve a delivery exception before it disrupts a customer commitment. A research consortium may need usable access to sensitive data under agreed conditions. These are illustrative buyer problems, not client cases. Each could support shared infrastructure, but each could also support a much simpler service.

Separating the outcome from the institution exposes three propositions: the service offered to the user, the exchange among participants and the organisation that operates the arrangement. Value in the first does not automatically fund the other two. A supplier might incur preparation costs while the procurement team captures the savings. That does not make the arrangement worthless. It makes the funding model a decision that must be explicit.

A practical framework: buyer, exchange, economics, commitment

Before approving a substantial build, write a one-page decision brief around four tests. This is my proposed management framework, not an OECD scoring method or a validated assessment.

1. Buyer — whose decision changes? Name the role, the recurring task and the consequence of today’s friction. “The industry needs better data” is a theme, not an answer.
2. Exchange — why would every essential participant contribute? Map who supplies, uses, verifies and can refuse access, with the cost and benefit for each role. If the smallest supplier bears repeated onboarding work for little benefit, the exchange may rest on pressure rather than durable incentives.
3. Economics — what pays for repeatable operation? Separate setup funding from ongoing support, data preparation, assurance, integration and dispute handling. Compare membership fees, service charges, an anchor buyer’s budget and explicit public funding as candidate models, not evidence of demand.
4. Commitment — what evidence permits the next step? Expressions of interest are weaker than tested use; tested use is not the same as payment. A grant proves funding for its agreed purpose, not post-grant demand.

The brief should end with a decision: proceed, narrow the use case, choose a simpler alternative or stop. A useful framework must permit all four.

Compare the alternatives before defending the architecture

A bilateral agreement and API may be sufficient when two organisations need a bounded exchange. An existing industry platform may be preferable when participants already trust its operator. A marketplace may address discovery and licensing more directly than a new consortium. A secure collaboration environment may support joint analysis without a broader institution. Each has trade-offs: central operators create dependency, bilateral integrations multiply, and federation distributes control while adding coordination work.

The relevant competitor is often the current spreadsheet, reporting portal or manual process. A proposed data space must beat that baseline on the buyer’s actual decision, not only on architectural elegance.

Test a transaction, not merely a demonstration

Consider a hypothetical supplier-evidence service. Instead of first assembling a comprehensive sector platform, test one evidence request with a small set of consenting organisations. Can the supplier prepare it within an acceptable burden? Can the buyer use it to make the intended decision? Who corrects an error? What happens when permission changes?

A technical demonstration proves that records can move. A commercial experiment must also investigate usefulness, acceptance, incentives and operating responsibility. This is consistent with the OECD’s observation that successful cases had a concrete initial use case, but my transaction-first approach is a working hypothesis, not a measured universal result. Its advantage is that it makes assumptions visible before they become expensive commitments.

Public value deserves an honest funding model

Some arrangements are intended to produce public goods, enable research or fulfil policy objectives. Requiring every one to become a standalone profit-making company would be the wrong test; the OECD itself sets out policy considerations, including government anchoring where no natural anchor exists. Operational sustainability still matters. Name the sponsoring institution, recurring budget, service obligations and conditions for continuation. A deliberately funded public service and a venture expected to earn its own revenue are both legitimate; an unexplained transition between them is the risk.

Clarity before commitment

Before the next platform commitment, ask: who is the buyer, why will participants keep contributing, and what will sustain operation after the initial funding period? If those answers are unclear, the free Workflow Fit Diagnostic is a starting point for framing the problem, and it may be enough on its own. If one bounded go/no-go or options decision remains, the Commercial Viability & Vendor Dependency advisory area on the Strategic Reviews page describes how an optional Decision Snapshot can address it. Neither is a data-space certification, market-demand validation, evidence audit or solution design.

The strongest case for shared infrastructure begins outside the infrastructure: a useful outcome, a credible exchange and an explicit commitment to sustain it. — Ronny Boesing