The voluntary carbon market spent 2023 in crisis. Three independent investigations found that a large share of avoided-deforestation credits — the dominant category at the time — represented far less abatement than they claimed. A handful of household-name corporates discovered, in the same news cycle, that their net-zero claims were resting on those credits. The market contracted; prices fell; the cynics declared the experiment over.
Two years later, the diagnosis is clearer and the response is well underway. The voluntary market has a working verification architecture — a multi-layer chain of custody from the carbon-removal site to the buyer's disclosure. Corporate buyers who are paying attention can now build credit portfolios that survive external scrutiny. Most are not paying attention. This article is for the buyer who would like to.
The four verification gaps the 2023 crises exposed
The criticisms levelled in 2023 reduced to four gaps in the system. Each has been substantively addressed, though not all at the same maturity.
Baseline integrity. Avoided-emissions credits depend on a counterfactual: "what would have happened without the project?" The 2023 investigations showed that baselines were systematically inflated. The response — moving to dynamic, ex-post baselines benchmarked against regional deforestation patterns — is now embedded in the major standards (Verra VCS v4.1, Gold Standard 2024 update). Baselines today are tighter and updated more frequently.
Permanence and reversal risk. A forest credit assumes the forest stays. When the forest burns or is logged, the claimed abatement reverses. The response: ringfenced buffer pools (10–25% of issued credits held back), insurance-style reversal protections, and explicit permanence terms (now typically 40+ years for nature-based credits).
Additionality. Would the project have happened anyway? The newer methodologies require explicit financial-viability testing — proof that the project is not viable without carbon revenue — and project-level disclosure of the financial gap closed by the credit.
Chain of custody. The crisis showed that the same tonne of CO₂ could, in some cases, be sold twice across different registries. The fix is registry interoperability and serialized retirement — every credit has a unique ID, traceable from issuance to retirement, and cannot be in two places at once.
What verification actually looks like now
Most corporate buyers shop for credits the way they shop for office supplies — by price and category. The verification trail is the product.
Integrity scores: what to look for
The market has converged on a few independent integrity-rating providers (BeZero, Sylvera, Calyx Global) who score credits against a consistent rubric. Their scores are not perfect, but they are correlated, and the dispersion within a category is now larger than the dispersion between categories.
The shift in the table is obvious. Engineered removals — direct air capture, biochar, enhanced weathering — score highest because the science is mechanistic and the chain of custody is short. Nature-based credits, particularly avoided-deforestation, score lower because the counterfactual problem is structurally harder. The market is repricing accordingly: high-integrity removals trade at 5–8x the price of low-integrity avoidance, and the spread is widening.
What a defensible portfolio looks like
A corporate buyer asked to defend its portfolio in front of a journalist, a regulator, or a sceptical board should be able to say four things. First: every credit has a unique ID and a published rating. Second: the portfolio mix is balanced toward higher-integrity categories (today, that means ≥60% in engineered removals or top-quartile nature-based). Third: the buyer can show the chain — what was measured, by whom, when validated, when verified, when retired. Fourth: the credit story is part of an emissions-reduction plan, not a substitute for one.
The last is the most important and the most often missed. Credits are not a license. They sit on top of an absolute reduction trajectory and address what cannot, at this stage, be abated directly. A net-zero claim that depends primarily on credit volumes will not survive serious scrutiny — not because the credits are necessarily bad, but because the story they support is structurally wrong.
- Buy by rating, not category. Within-category dispersion is larger than between-category dispersion. A top-quartile reforestation credit can be defensible; a bottom-quartile direct-air-capture credit may not be.
- Tilt the portfolio toward engineered removals. They are more expensive, but the integrity floor is structurally higher. Price will normalize as supply scales.
- Lead with absolute reduction. Credits address residual emissions. If they are doing more than that in your story, the story is the problem.
What this means for sustainability leaders
The voluntary carbon market is messier than it looked in 2021 and more defensible than it looked in 2023. The verification architecture is now sufficient for a careful buyer to assemble a portfolio that survives external scrutiny — and the data, ratings, and audit trail are available without proprietary research.
The leaders who get this right do not outsource the judgment. They build internal capability to assess projects against the rating data, they retire credits on a serialized basis, and they publish their portfolio composition annually. The leaders who don't are still buying on price-per-tonne and category label — and they are the ones who will appear in the next round of news investigations.
The bar has moved. The good news is the bar can be cleared.