{"@context":"https://schema.org","@type":"Article","headline":"Carbon Project Checker methodology","description":"The eight dimensions, their weights, the band thresholds and the data sources behind every assessment.","author":{"@id":"https://www.evertreen.com/#organization"},"publisher":{"@id":"https://www.evertreen.com/#organization"},"mainEntityOfPage":"https://www.evertreen.com/carbon-projects/methodology"}

How we assess carbon projects

A rating nobody can check is a rating nobody should trust. This page is the whole method: what we start from, what we look for, how the eight dimensions are weighted, and what each band actually means. 2 projects have been assessed this way.

Where the facts come from

Every profile begins with the registry's own record: the project id, the name, the proponent, the country, the project type, the methodology, the current status, and the credits issued and retired to date. None of that is written by us or by a model. It is imported from registry data and re-synced monthly, and every field change is logged.

The bulk of our registry data comes from the Voluntary Registry Offsets Database published by the Berkeley Carbon Trading Project at the Goldman School of Public Policy, University of California, Berkeley, which we use under its Creative Commons licence. Where a registry publishes its own machine-readable data we use that too.

What the assessment adds

On top of the registry record, we research what independent parties have published about the project: validation and verification reports, peer-reviewed research and preprints, court filings and regulatory action, named investigative journalism, and established watchdog analysis.

Two rules govern what reaches the page. Every claim must have a source URL that resolves and a publication date, and both are shown next to the claim. And every negative finding is attributed to whoever made it, and labelled as a finding, an allegation, a dispute, a regulatory action or a registry action, because those are different things and a buyer needs to be able to tell them apart.

Assessments are drafted with a large language model that searches the public record, then filtered. Any claim whose source will not resolve, or whose source is not on our list of accepted publishers, is dropped before publication. Any assessment landing in the two highest concern bands, and any assessment naming an individual, is reviewed by a person before it appears.

The eight dimensions

Each is scored from 0 to 100 for concern. A higher score means more concern, not more quality. Where we found nothing bearing on a dimension, it is left unscored and shown that way on the page.

DimensionWeight
Additionality20
Baseline and over-crediting risk18
Permanence and reversal risk14
Leakage10
Community and safeguards12
Verification quality and recency8
Registry status10
Independent scrutiny8
  • Additionality. Would this have happened anyway, without the credit revenue?
  • Baseline and over-crediting risk. Is the counterfactual the project is measured against credible?
  • Permanence and reversal risk. Fire, logging, political risk, and whether the buffer pool is adequate.
  • Leakage. Is deforestation or emission displaced somewhere else rather than prevented?
  • Community and safeguards. Free, prior and informed consent, land tenure, and documented rights concerns.
  • Verification quality and recency. Who validated it, how recently, and whether the auditor has had issues.
  • Registry status. Active, on hold, suspended or withdrawn.
  • Independent scrutiny. What journalism, academia, litigation and NGOs have found.

How the score becomes a band

The dimension scores are combined using the weights above, into a single concern score from 0 to 100. That score maps to a band as follows.

Serious concerns identified70 and above
Significant concerns identified50 to 69
Some concerns identified28 to 49
Limited concerns identifiedBelow 28
Insufficient public informationFewer than 4 dimensions have any evidence

Two rules that override the arithmetic

Thin evidence never produces a clean band. If fewer than 4 of the eight dimensions have any public evidence behind them, the result is insufficient public information, whatever the average of the scored dimensions happens to be. Absence of findings is not clearance. It is the single most common way a tool like this misleads people, and we would rather say we do not know.

Registry status can force the band up. A project the registry has suspended, withdrawn or placed on hold is shown as at least significant concerns identified, whatever the other seven dimensions say. A registry pulling a project is the strongest signal in this dataset and it should not be averaged away.

Why we do not say good or bad

We do not describe a project as good, bad, a scam or worthless. Those are verdicts, and a verdict on a named business is a different kind of statement from a summary of the public record. What we publish is what independent parties have found, who found it, and when. The bands summarise that, nothing more.

Removals and avoidance are different products

A removal credit represents CO2 taken out of the atmosphere, for example by planting trees, by biochar, or by direct air capture. An avoidance credit represents emissions a project says would otherwise have occurred, for example deforestation that was prevented. They are priced in the same units and sold in the same marketplaces, and they are not the same thing. Every profile labels which one a project issues, prominently, because a great many buyers discover the distinction only after they have bought.

How often assessments are refreshed

Every assessment is regenerated at least every 180 days. A registry status change triggers immediate regeneration rather than waiting for the cycle. Assessments are versioned and never overwritten, so we can always show exactly what was published on a given date and from which sources.

Corrections

If you represent a project and something on its page is inaccurate, request a correction. Tell us the statement you dispute, what the correct position is, and the supporting evidence. We respond within 10 working days, and every request is logged.

Our commercial interest

Evertreen sells tree planting and carbon credits. We show our own products underneath each assessment, clearly labelled and visually separated from the assessment itself. Where we sell something associated with a project we assess, the project page carries an explicit banner saying so. We would rather tell you than have you find out.

What this is not

This is not investment, legal or procurement advice, and it is not a substitute for your own due diligence. It is a starting point that puts the public record about a project in one place, with its sources attached, so that the questions worth asking are obvious before you spend money.

products View cart

Plant & offset from £ 1

Let’s find the best tree planting or Verra CO₂ offset solution for your needs.

I agree to be contacted by email and SMS from Evertreen based on my provided details