One screen, applied to every project in the same order. It asks a fixed set of questions, records what each answer rests on, and produces an eligibility decision you can check against the rules published on this page.
Every portfolio, measured
There are two separate questions about any carbon credit. Confusing them is how the market got into trouble, so we answer them apart.
The first question is whether the tonne is real, and whether it can be transferred, retired and used for the claim you intend. Either the mitigation happened, was measured properly, can be transferred with clean title and will be counted once — or it cannot carry the claim. The assessment involves evidence and professional judgement. The procurement decision it produces is binary.
The second is what wider outcomes the project delivers. Health, livelihoods, gender equality, access to energy, biodiversity: these vary between projects and can be assessed on a scale. That is the 360° Impact assessment, and it is a different exercise.
Core integrity weaknesses are never averaged or traded against wider benefits. A strong impact assessment cannot compensate for a failure of additionality, double counting, independent verification or safeguards. There is no arrangement of strong credits that carries an ineligible one through.
Scoring
A numeric rating invites a buyer to accept the 82 because it is cheaper than the 95. That trade is exactly what a screen exists to prevent. We publish the eligibility decision, the criterion-level verdicts and the evidence behind them, rather than converting integrity into a saleable scale.
Pricing
A credit that does not clear the threshold for your claim is not offered at a lower price. It is not offered at all. Permitted evidence gaps on non-core criteria are recorded, constrained by the pass rule and disclosed to you — they never become a score, and points from the impact assessment can never be traded against them.
Every credit that carries carbon for you clears the same bar. Credits differ in what claim they support, how long the carbon stays put and what they cost — never in whether the tonne is real.
The four parts do different jobs. The twenty criteria draw on three frameworks; where two of them address the same question and differ, the stricter test binds.
Defects that end the assessment outright. There is no purpose in examining the baseline methodology of a project whose ownership is disputed.
The questions every surviving project is assessed against — and how much of the set the available evidence was able to answer.
Curable shortfalls are converted into additional volume rather than into verdicts. It happens before the verdicts are read.
Which combinations of verdicts we are willing to sell. It adds nothing new; it reads the sheet the first three parts produced.
The gates end an assessment. The criteria establish what the evidence shows. The pass rule states which patterns of evidence are sellable. Nothing is scored, weighted or averaged at any point in that sequence.
A project meeting any one of these is excluded and is not re-screened. No amount of strength elsewhere compensates, and no volume adjustment is available: these are defects in whether the tonne exists at all, not in how well it has been measured.
Excluded activity class
Sold-product avoided emissions — a category we do not screen at all.
Additionality failure
The activity would have occurred without carbon finance.
Regulatory-surplus failure
The activity was legally mandated in the host jurisdiction.
Double counting or title defect
The same tonne issued, sold or claimed twice; a competing claim to the same attribute; or defective legal title.
Safeguards or human-rights failure
Evidenced harm, consent absent where it was required, or an unresolved community dispute.
Ex-ante issuance
The credit does not represent an outcome that has already occurred.
No independent verification
The mitigation outcome has not been verified by an independent party.
Crediting-programme failure
The issuing programme itself fails the requirements we set for programmes.
Unmanaged reversal risk
Reversal risk is present with no buffer, insurance or make-good mechanism behind it.
Legacy activity creating a perverse incentive
A project type where carbon revenue itself created the incentive: industrial gas destruction, large hydro with limited demonstrable dependence on carbon finance, or any activity where credit income rewards increasing the underlying emitting activity.
Gates and criteria
With two exceptions they are the failure outcomes of the criterion set itself, checked first because they can often be established quickly. Where a defect is not obvious at the outset, the same conclusion is reached the long way round, through the criterion it belongs to. The two exceptions are activity-class rules and have no matching criterion.
Exclusions
Naming them would turn a documented assessment into a public dispute with developers we also buy from, and would make the next developer slower to share documentation — which costs the next client. We record the vintage, technology, country and ground of exclusion in our own register instead.
Each of the twenty criteria returns one of these. They are the codes that appear in your evidence pack, printed here in the same form, so you can read one against the other.
meets
The criterion is satisfied on evidence that can be cited.
gap
Partially satisfied, or satisfied on thin evidence. Recorded with the deficiency stated.
fail
Not satisfied.
na
Genuinely not applicable to this activity or programme type. The reason is recorded.
Missing evidence
gap, not meetsA criterion the documentation does not address has not been satisfied, and is not treated as though it had been. Twenty questions go in and twenty verdicts come out, each recorded against the document it rests on. There is no fifth verdict, no partial credit and no aggregation.
Close calls
This is designed in, and it is not discretionary. Where documentation is silent or ambiguous, a criterion is recorded as gap rather than meets, and a goal takes the lower step. Variability resolves toward the more conservative answer rather than in whichever direction suits the sale. On the pass rule, that means a borderline project fails.
na is controlledNot applicable is a legitimate verdict. Storage durability is not a meaningful question for an avoidance credit, and recording it as a failure would be wrong.
But it would otherwise be the easy route through the screen. A project with almost no published documentation could return na across most of the set and pass on a handful of remaining verdicts. So we measure how much of the set the evidence was actually able to answer, and record that figure on every screen and in your evidence pack.
Below a defined level the screen is not valid and further documentation is required before any decision. Below a lower level the project is not screenable at all, and is excluded rather than assessed. Those two outcomes are different, and the difference matters: a project that is not yet assessable has not been rejected, and may become assessable when more is published. A project cannot pass by being poorly documented.
Grouped into six areas for organisation only. There are no weights and no group scores, because there is no score. Each criterion is assessed on its own, against the document its verdict rests on.
Each block is one criterion, drawn at the same width, because none counts for more than another. A bar beneath a block marks a Tier 1 criterion.
credible-baseline Tier 1
Is the baseline a realistic representation of what would have happened, and has it been kept realistic across the crediting period?
ex-post-verified Tier 1
Does the credit represent an outcome that has already occurred and been independently verified — and, for end-state removals, verified before issuance by an appropriately accredited body?
accurate-accounting
Is the quantification arithmetically sound and conservatively treated, with all material upstream and downstream process emissions deducted from gross removal?
leakage
Are emissions displaced outside the project boundary identified and deducted?
emissions-matching Tier 1
Does one credit correspond to one net tonne actually mitigated and, for an end-state claim, to a removal delivered in the same reporting period as the residual emissions it counterbalances?
additionality Tier 1
Would the activity have occurred without carbon finance?
regulatory-surplus
Is the activity beyond what law or regulation requires in the host jurisdiction?
causality
Are the reported outcomes a direct result of the project rather than background trend or coincident policy?
On regulatory surplus we apply the test absolutely. The SBTi standard permits mitigation used for regulatory or compliance purposes to be counted at its higher recognition levels; we do not take that option. A legally mandated activity fails the criterion and gates, whatever the recognition level.
no-double-counting Tier 1
Is the mitigation counted once and once only, and not allocated to another distinct residual-emissions source?
registry-tracking
Is the credit uniquely identified, publicly traceable and retired through a functioning registry?
unique-attribution
Is the environmental attribute capable of attribution to one holder only, with an auditable chain of custody where a product-based removal is used?
These two are assessed here at project stage only. Whether the specific credits you buy were actually transferred and cancelled, and whether title reached you, is verified at procurement and recorded separately in your evidence pack.
crediting-programme
Does the programme issuing the credit satisfy the requirements set for crediting programmes, and is an end-state removal issued under a science-based methodology developed or independently validated through a credible process?
governance
Is the programme independently governed, with a functioning grievance and appeals process?
transparency
Is sufficient information publicly available to assess the credit, including registry status, monitoring and any applicable benefit-sharing arrangements?
On transparency we also record which documents were public and which were supplied privately. A screen resting on privately supplied documents is reproducible only by someone with the same access, and you need to know that before your assurance provider discovers it.
permanence
Is the mitigation durable, with the storage period, monitoring, remediation and responsible party evidenced?
reversal-risk
Are reversal risks identified, quantified and managed through funded monitoring, remediation, replacement, buffer or insurance arrangements?
durable-removals
For a removal credit, does the evidenced storage duration satisfy the durability role it has been assigned?
avoid-lock-in
Does the activity entrench high-carbon infrastructure or delay transition?
Lock-in matters most for older efficiency, fuel-switch and industrial cohorts, where an activity can reduce emissions today and extend the life of the system producing them. An activity that does not meet its assigned durability role is ineligible for that claim; it is not averaged through by other removals.
safeguards Tier 1
Does the project identify, evaluate and reduce risks to people and the environment; respect human rights and the rights of Indigenous Peoples and local communities; apply free, prior and informed consent where relevant; and provide remediation and fair benefit-sharing arrangements?
sustainable-development-benefit
Does the activity evidence at least one positive sustainable-development benefit that occurs through, or is an unavoidable physical consequence of, the implemented activity?
Positive benefit is an eligibility floor wherever the claim route requires it, and a portfolio average cannot cure an activity-level failure. What a project delivers above that floor is measured by the 360° Impact assessment, and is never assessed twice.
Tier 1 criteria are not weighted more heavily. Nothing is weighted, because nothing is scored. They carry a stricter passing verdict: meets is the only result that lets a project through.
additionality
Would it have happened anyway?
no-double-counting
Has anyone else claimed this tonne?
ex-post-verified
Has the outcome actually been verified by someone independent?
credible-baseline
Is the comparison it is measured against defensible?
emissions-matching
Does one credit correspond to one tonne mitigated?
safeguards
Have people and their environment been protected?
On the other fourteen a gap is tolerable: it is recorded, it is disclosed to you, and the project can still pass. On these six it is not, because thin evidence here puts the tonne itself in question rather than its paperwork.
A credit is issued against assumptions fixed when the project was registered. Some of those assumptions decay, and the methodology does not always follow them down. Where the shortfall is a matter of degree, we correct it with volume rather than recording a failure.
Correctable
A project delivering less than its methodology assumes, because a parameter it was correctly credited against has since moved, is not fictitious. It is delivering a smaller quantity than the printed number. Where the evidence shows that, we retire additional credits so that the tonne you claim is the tonne that was delivered. You pay for one tonne and you claim one tonne; the additional volume is ours to buy.
Not correctable
Where the defect concerns whether the tonne exists, is additional, is uniquely owned or is safe to use, buying more cannot cure it. The tonne was sold twice, the project was never additional, someone was harmed: no multiplier exists, retiring more of a non-existent tonne delivers nothing, and the criterion fails.
The inputs are published and independent. Emission factors and other key parameters come from recognised official or authoritative sources, identified and dated in the assessment. We do not select a figure because it produces a convenient result.
A parameter correctly sourced at the time and since superseded is not a defect in the project’s work. The question at each criterion is whether the project did what was asked of it under the methodology in force. Whether the credited tonne still matches physical reality today is a separate question, answered with arithmetic rather than with a verdict.
Adjustment has a limit. Beyond a defined point, correcting the quantification would require so much additional volume that the project is excluded from carrying a carbon claim rather than adjusted.
The adjustment happens before the pass rule is applied, which is why the rule below refers to verdicts after adjustment. The cost of that additional volume sits with us, not with you.
By this point every question has a verdict. The pass rule adds nothing new. It states which combinations we are willing to sell.
There is no partial pass and no conditional pass. There is no arithmetic in which three conditions outweigh a fourth.
No exclusion gate triggered
None of the ten.
No failed criterion
None of the twenty, after any conservatism adjustment has been applied.
No gap on a Tier 1 criterion
Meets is the only verdict that lets a project through on the six.
Enough evidence to complete a valid assessment
Enough of the criterion set graded, rather than recorded as not applicable, for the result to mean anything.
Both projects below clear all ten gates. Both return a single gap and no failures. Only one is offered, and the difference is which criterion the gap landed on.
Gap on leakage
No gate triggered. Seventeen meets. Two na, each with the reason recorded. One gap, on leakage, which is not a Tier 1 criterion. Enough of the set graded to make the assessment valid.
The gap appears in your evidence pack with the deficiency stated, so you know it is there and you know it was not sufficient to fail the project.
Gap on ex-post-verified Tier 1
No gate triggered. Eighteen meets. One na, with the reason recorded. One gap, on ex-post-verified, which is Tier 1. Evidence completeness higher than Project A’s.
It fails on that single verdict, and the other nineteen do not carry it. Nothing about this project is offered for a carbon-carrying role at any price.
Both examples are illustrations of how the rule reads, not records of assessed projects.
A more complete evidence file does not rescue a Tier 1 gap, and a thinner one does not condemn a project that has answered the six questions that matter.
Tell us the claim you need to support and the Impact Band you want the finished portfolio to reach. Every credit we source against that specification will have been through the screen on this page.