How to Spot a Weak Carbon Credit: A Buyer's Due Diligence Guide

Every credit on every registry has passed validation and verification. That is the floor, not a quality rating. Two credits with identical registry status can rest on completely different measurement, and the difference only becomes visible if you ask the right questions before buying.

A registry records that a unit exists, who owns it, and whether it has been retired. That prevents double counting, which is genuinely important, but it says nothing about how accurately the underlying tonnes were measured.

This is the single most common misunderstanding in carbon procurement. "It's Verra-registered" answers a legal question, not a scientific one.

1. How was the sampling designed? Ask for the stratification logic and how plot counts were derived. A good answer references measured variance and a target precision. A weak answer references a plot count with no derivation, or declines to share the design at all.

2. What uncertainty was reported, and what deduction followed? Every credible estimate has a confidence interval. If none is quoted, either it was not calculated or it is unflattering. Both are informative.

1. Can raw field data be inspected? Not just a summary, but the actual measurements. Projects with defensible data chains can produce them. Projects that cannot usually have gaps between field sheet and headline number.

4. What is the baseline, and what evidence supports it? Baselines are counterfactual claims. Look for contemporaneous documentation rather than reconstruction, and for conservatism rather than optimism.

5. What is the reversal exposure and detection latency? How is the project monitored between verifications, and how long would a fire go unnoticed?

A proprietary accuracy figure with no published method behind it. Accuracy is a property of a method applied to a population; a single percentage without the method is a marketing number.

Wall-to-wall satellite measurement with no ground sample. Sensors measure proxies, and proxies need calibration.

Precision quoted without an interval. "Accurate to within 5%" means little without knowing at what confidence, over what population, and validated against what.

And methodology silence, which refers to a project that will not say which version of which methodology it used, particularly across a transition period.

Compliance demand behaves differently from voluntary demand. When a credit offsets a statutory liability, a later finding of ineligibility returns that liability to the buyer. The diligence bar rises accordingly, and buyers increasingly employ technical reviewers rather than relying on the registry.

That shift rewards projects that can answer the five questions above and penalises those that cannot, regardless of how the credits are marketed.

Check legal status first, because it is binary and fast: registry listing, methodology, and for international use, authorization and corresponding adjustment.

Then request the measurement file: sampling design, uncertainty calculation, validation approach. Projects that treat this as a normal request are usually the ones with nothing to hide.

Then assess durability: buffer risk rating, leakage treatment, monitoring cadence.

Only then negotiate price. A cheap credit that fails review later is the most expensive kind.

1. ICVCM Core Carbon Principles — https://icvcm.org/core-carbon-principles/ (accessed 16 Sep 2026)

2. Verra VCS programme — https://verra.org/ (accessed 16 Sep 2026)

3. IPCC 2006 Guidelines Vol. 4 (AFOLU) — https://www.ipcc-nggip.iges.or.jp/public/2006gl/vol4.html (accessed 16 Sep 2026)