The voluntary carbon market has a credibility problem that goes deeper than the headlines suggest. When investigative reporting in 2023 and 2024 exposed that large portions of certain REDD+ project portfolios were generating credits against deforestation that would never have occurred anyway, the industry's response was to call for more audits, more third-party verification, tighter methodology standards. All of those responses are correct. But they miss the structural issue underneath: the market has no shared provenance infrastructure.
A retirement certificate tells you that a specific quantity of credits from a specific project was retired on a specific date. What it doesn't tell you — what the current market has no reliable mechanism for telling you — is the complete data trail from which that credit was derived. Where exactly was the forest canopy measured? What sensor? What baseline was applied? Which validation body reviewed the methodology submission? Which registry issuance batch does this credit belong to, and what was its chain of transfers between issuance and retirement? Those questions have answers, but they're scattered across project developer PDFs, third-party auditor reports, registry databases, and broker transaction logs that were never designed to connect to each other.
The Double-Counting Problem Is a Data Problem
Double-counting in voluntary carbon markets typically takes one of three forms: the same emissions reduction is claimed by both the host country under its Nationally Determined Contribution (NDC) and by the project's corporate buyer; the same credit is retired in multiple registries due to cross-listing without proper correspondent adjustment; or the credit's underlying measurement data covers an area that overlaps with an adjacent project's claim boundary. All three forms are fundamentally data integrity failures, not methodology failures.
The ICVCM's Core Carbon Principles, published in 2023, addressed this directly by requiring that credits meeting CCP eligibility demonstrate that the underlying emissions reductions or removals are "not double-counted." The guidance is clear; the enforcement mechanism is not. There is no cross-registry deduplication layer that can confirm, at credit-level granularity, that two serial numbers from two different registries don't reference the same underlying tonne of CO₂.
This isn't the registries' fault. Verra's VCS registry, Gold Standard's Impact Registry, the American Carbon Registry, the Climate Action Reserve — each built its issuance and tracking systems to manage projects within their own methodological framework. They weren't designed to function as nodes in a shared provenance graph. The market grew faster than the infrastructure, and now the infrastructure gap is destroying trust in the underlying asset.
What "More Audits" Actually Fixes — and What It Doesn't
Third-party validation and verification (VVBs in VCS terminology) do critical work. A qualified validator reviews a project's baseline scenario documentation, its additionality argument, and its monitoring plan before the first credit is issued. A verifier reviews monitoring reports against the approved methodology and confirms the quantification. This is not box-ticking; it's substantive technical review that catches real errors and discourages inflated claims.
We're not saying audits are inadequate — they're doing the job they were designed for. What they were not designed to do is maintain a continuous, queryable chain of custody from measurement event to retirement. An audit produces a report that attests to the quality of a project's claimed reductions at a point in time. It doesn't produce a data artifact that can be cryptographically linked to subsequent transfers and compared against other audits to prevent overlap. That's a different instrument entirely.
Consider a mid-size forest carbon project in the Brazilian Cerrado, operating under VCS VM0007 (the REDD+ methodology). A VVB conducts a five-year verification and signs off on 180,000 tonnes of emissions reductions. The registry issues 180,000 VCUs with sequential serial numbers. A carbon broker purchases a block of 40,000 VCUs and sells tranches to four different corporate buyers. Two of those buyers are based in jurisdictions where the host country's NDC accounting treatment creates a correspondent adjustment ambiguity. None of this information — the VVB report, the serial number block, the broker's fractional sale records, the buyers' jurisdictional exposure — is currently queryable in any unified way. An auditor reviewing one buyer's retirement certificate has no reliable mechanism to surface the full picture.
Why Provenance Infrastructure Is Different from Ratings
The market's initial response to the credibility crisis was to fund carbon credit rating agencies — firms that assign quality scores to projects or credit vintages based on methodology review and risk assessment. This is valuable work, and rating agencies like Sylvera and BeZero Carbon have built sophisticated analytical frameworks for it. But ratings are opinions about expected quality at the time of assessment. They are not records of what actually happened.
Provenance infrastructure records what actually happened. It answers: on what date was canopy height measured in this specific polygon? With what instrument? What was the above-ground biomass density estimate, and what uncertainty range was applied? When was this monitoring data submitted to the registry? What was the serial number assigned to the credit this data underpins? Where has that serial number been since issuance?
The pharmaceutical analogy is instructive here. A drug rating system could tell you that a manufacturing facility generally produces good output. But what regulators actually require is batch-level traceability: lot number, manufacturing date, raw material sourcing, quality control test results, cold chain temperature logs, distribution records. The rated quality of the facility is useful context. The batch-level chain of custody is the actual compliance document.
Carbon markets are moving toward exactly this level of scrutiny — not because regulators have mandated it yet, but because institutional buyers are demanding it and audit firms are beginning to ask for it. The question is whether the market builds the infrastructure proactively or waits for a regulatory mandate that arrives after more credibility damage is done.
What a Chain-of-Custody Layer Requires Technically
A functional chain-of-custody layer for voluntary carbon credits needs to do several things that current registry systems don't do natively. First, it needs to ingest and normalize measurement data at the source — SAR-derived canopy measurements, LiDAR point clouds, ground-truth plot surveys — and link that data to project geographic boundaries with enough granularity to detect overlapping claims. Second, it needs to record every custody transfer between issuance and retirement with enough metadata to reconstruct the full ownership chain. Third, it needs to be queryable by any legitimate stakeholder, not just the project developer or the registry, at serial-number level.
The last point is where the current system most clearly fails. Today, a corporate buyer who wants to verify the provenance of the credits they've retired has to manually request documentation from their broker, who may have purchased from another broker, who purchased from an aggregator, who purchased from the project developer. Each link in that chain involves a separate document request, variable response times, and documents in incompatible formats. None of it is programmatically queryable.
The architecture for fixing this exists. Cryptographic linking of custody records — the same underlying approach used in financial audit trail systems — makes tampering detectable and provenance queryable. Standardized data schemas for measurement inputs, based on existing frameworks like the IPCC's guidance on forest carbon accounting or the GHG Protocol's land sector guidance, would allow cross-registry comparison. Registry API integration would let custody records be appended to existing serial number structures without requiring registries to rebuild their core systems.
The Path Forward Isn't Comfortable
Building this infrastructure requires some uncomfortable conversations. It requires registries to open APIs to third-party data integrations rather than treating their serial number databases as proprietary assets. It requires project developers to share measurement data at a level of granularity that some may view as competitively sensitive. It requires corporate buyers to acknowledge that their current retirement certificates are not, by themselves, verifiable proof of impact.
None of those conversations are easy. But the alternative — continuing to run a multi-billion dollar market on the implicit trust that documents in disconnected systems are accurate and non-overlapping — is no longer viable. The investigative reporting of recent years didn't create the market's credibility problem; it revealed one that was already structural. The fix has to be structural too.
Chain of custody isn't a premium feature for the voluntary carbon market's future. It's the baseline documentation standard that every asset market with serious institutional participation already requires. The question isn't whether the VCM needs it. It's how fast the market can build it.
