Illustrative forest, landfill gas-capture equipment and cookstove shown in three adjacent panels
Three synthetic scenes illustrate different project activities, not the quality or eligibility of a named project.
Diagram text description

A measured tree trunk in a forest, gas-handling pipes beside a landfill, and a cooking stove appear in separate panels. The synthetic collage does not show verified units, project approval or measured mitigation.

Carbon offset projects should be compared by the change they claim and the evidence supporting it. A forest photograph, a gas-capture installation or a programme logo can describe an activity without establishing that the offered carbon units exist, represent the stated outcome or support your intended use.

Begin with the mechanism and baseline, then examine the monitored result, independent review and offered-unit status. This order prevents a familiar project label from substituting for a project-specific assessment. A project can avoid a release, reduce ongoing emissions or remove atmospheric carbon into storage. These mechanisms need different evidence even when credits are expressed in the same unit of carbon dioxide equivalent.

The comparison below is for a business reader evaluating an offer. It contrasts nature, methane and industrial-gas, energy and cookstove activities, then provides a common screening matrix. It does not rank whole project families as good or bad. The useful outcome is a traceable decision about a named project and unit batch, including a clear reason to pause when the evidence does not answer the claim.

Start with the claim a project makes

Ask the seller to complete a concrete sentence: what changes because of this activity, compared with what would otherwise happen? The second part is the baseline or counterfactual. It is an estimate of the alternative, not an observed parallel world. A persuasive project narrative can coexist with an overstated baseline, so the reasoning behind that estimate matters as much as descriptions of the activity itself.

Avoidance concerns a release that is prevented relative to that reference. A forest-protection claim, for example, depends on what loss would credibly occur without the intervention. Reduction concerns a lower emissions flow, such as a process releasing less greenhouse gas than its baseline. Removal involves atmospheric carbon entering a store, which introduces questions about the amount added and how long it remains there. Programme terminology and accounting rules must still be read for the actual activity.

These categories are not interchangeable labels for success. Protecting an existing forest is not automatically a new removal, and destroying methane before it reaches the atmosphere is not atmospheric carbon storage. A project may contain several activities, but their quantities and methods need to remain distinguishable. Equal tonnes on an offer sheet do not mean equal uncertainty, storage duration or exposure to a failed assumption.

Conceptual mechanisms differ: preventing a release, lowering an emissions flow, and storing atmospheric carbon are not interchangeable claims.
Conceptual mechanisms differ: preventing a release, lowering an emissions flow, and storing atmospheric carbon are not interchangeable claims.
Diagram text description

Three equal-width paths contrast a baseline release with prevention, continuing emissions with lower emissions, and atmospheric carbon with storage. Equal box sizes show no measured tonnes, time scale or quality ranking.

For comparison, hold the question constant while changing the requested evidence. “What proves the claimed change?” might lead to a land-use baseline for one project, operating records for another, and measured carbon-stock changes for a third. Comparing the strengths of those answers is more informative than choosing the most appealing technology name.

Nature projects: the carbon is not all measured the same way

Nature projects require a description of the carbon pools and processes being credited. Trees, soil, peat and coastal vegetation are not interchangeable measuring devices. A hectare of land does not establish a universal credit yield, and a restoration activity does not prove that every reported tonne is additional removal.

Use the specific setting to direct the evidence request:

  • Forests: identify whether the claim concerns avoided loss, restoration or another activity; ask which biomass and other pools are included, how changes are monitored, and how fire, harvest or other losses are addressed.
  • Agricultural soils: examine the practice change, sampling or modelling design, soil depth and represented area. Ask how a change in management or later loss of stored carbon affects the credited outcome.
  • Peatlands: establish which drainage or restoration processes the method addresses and how its baseline represents continued degradation. Do not assume that a rewetting claim is entirely a new carbon-removal claim.
  • Coastal systems: establish the vegetation and sediment pools, project boundary and monitoring access. Land and resource rights, local use and disturbances can affect both the activity and its durability.

The questions are different, but the comparison discipline is shared: specify the credited change, identify the evidence that can detect it, and explain what would invalidate the interpretation. A monitoring report should make it possible to understand the area and period represented rather than offering a single unexplained total.

Storage also creates an obligation beyond the initial measurement. Ask about the monitoring duration, reversal provisions, responsibilities and remedies under the applicable programme. A buffer or insurance-like arrangement is evidence of a risk-management mechanism, not proof that loss is impossible. Community rights and ecological safeguards require their own records; a carbon estimate cannot stand in for them.

Methane and industrial-gas projects: check operational evidence

Methane and industrial-gas projects often depend on a different evidence chain: the gas source, measurement system, treatment equipment and operating record. An installed flare or destruction device shows that equipment exists. It does not establish the volume or composition of gas treated during the credited period, the equipment's performance, or the emissions that would have occurred without the project.

Request the baseline explanation alongside the operating evidence. What was the previous practice, what regulation or operating requirement applies, and what activity does the methodology treat as additional? An activity already required or expected under the relevant rules needs the programme's actual additionality assessment; the buyer cannot resolve that question from a photograph of new equipment.

For measurement, look for a coherent connection between gas-flow records, composition, instrument calibration, downtime, treatment performance and the monitoring report. A period with incomplete data should have an explained treatment under the method. Design capacity and a short demonstration run cannot simply be extended across the full crediting period.

The risk profile also differs from stored carbon. A failure to operate equipment next year can reduce future mitigation, while losing a biological carbon store can reverse a previously credited stock gain. Those are different events to examine, not a basis for calling every operational project permanent or every nature project unreliable. Compare the particular method, period and failure mechanism.

Energy and cookstove projects: use is the missing question

Energy and cookstove offers often begin with a visible installation or distribution count. The evidential question is what happens after delivery. A renewable plant's nameplate capacity is not its eligible monitored output. A stove handed to a household is not proof that the household uses it consistently or stops using the previous cooking arrangement.

For cookstoves, ask how use is established, how sampling represents participating households and how fuel displacement is estimated. Continued use of an old stove alongside a new one can matter to the claimed change. The method must explain the baseline, relevant fuel assumptions, project emissions and any other adjustments used to reach a net monitored result. Do not replace that chain with a device count multiplied by an unsupported per-stove saving.

Distribution is the start of an evidence chain, not the same as a net monitored result.
Distribution is the start of an evidence chain, not the same as a net monitored result.
Diagram text description

Cookstove distribution is followed by evidence of use and sampling, then fuel displacement and a net monitored result under the applicable method. The arrows show evidence dependencies rather than automatic credit issuance.

For energy projects, examine the applicable eligibility and additionality assessment, metered generation and the baseline supply being displaced. Grid conditions, project circumstances and methodology versions can change the comparison. The fact that a technology generates renewable electricity does not establish that every installation can issue credits under every programme.

This is where two superficially similar offers may separate. One may provide period-specific monitoring, a documented sample and a method-supported result; another may provide only delivery numbers and a forecast. That is a difference in the evidence available for the offered claim. It is not a reason to infer the same quality for all cookstove or renewable-energy projects.

A project type is not a methodology

A project type describes an activity family. A methodology specifies how an eligible activity establishes its baseline, tests additionality, defines its boundary, monitors outcomes and calculates credited results. The programme supplies the broader rules and registry arrangements. These layers are related, but the name of one does not establish compliance with the others.

Obtain the exact methodology identifier and version, then connect them to the project's documents and monitoring period. Ask why that method applies to the activity, geography and circumstances described. If the offer cites a newer version while the monitoring report uses an earlier one, request the explanation of which rules govern the offered vintage. Do not silently replace the historical method with the newest document you can find.

A programme or methodology assessment is useful context, including any relevant quality label, but it is not a substitute for tracing the offered units. Confirm the applicable assessment and eligibility rather than inferring that all categories or all vintages under a recognised programme have the same status. The remaining project-specific tests still need answers.

Pause when the project description, method and monitoring period cannot be reconciled. This is a request for missing evidence, not an invitation to invent a more favourable method. A comparison is useful only if the competing offers are evaluated against the rules that actually govern their claimed outcomes.

A project listing is not an issued and retired unit

A project can be described publicly before it has issued the units a buyer expects to receive. A validated design concerns the plan and its conformity with applicable requirements. Monitoring records what happened over a period. Verification evaluates the reported result under the relevant rules. Issuance creates identified units through the programme's process. Ownership and retirement then concern those units, rather than a new measurement of the project's physical activity.

Keep these stages separate when reading an offer. A project listing does not establish an issued quantity, and a verification report does not by itself identify which serials a seller owns or can deliver. An issued unit may still be available for transfer; a retired unit has been taken out of circulation. Check the actual registry record and the retirement purpose or beneficiary relevant to the proposed use, rather than accepting a screenshot with no identifying context.

The illustrative figure shows one project associated with two batches in different states. It is a relationship map, not a claim that either batch exists or a promise of automatic progression. The same project can have several monitoring periods and unit histories. A claim about one batch cannot be extended to all of them merely because the project name matches.

Illustrative unit batches under one project can have different statuses; the project listing alone does not establish the offered serials.
Illustrative unit batches under one project can have different statuses; the project listing alone does not establish the offered serials.
Diagram text description

A hypothetical project record connects to batch A marked issued and batch B marked retired. These are separate associated records, not real units or a promise that an issued batch automatically reaches retirement.

When the seller offers future delivery, evaluate it as future delivery. A contract to fund an activity or acquire later units is different from evidence that a specified unit has already been issued and retired. Public wording should preserve that distinction instead of describing an expected outcome as a completed one.

Use a risk-specific project screening matrix

Use the matrix to compare the evidence offered, not to assign a star rating to whole project types. The examples are prompts: each project's applicable methodology determines its detailed requirements. Shared quality criteria do not mean that identical documents will answer them across all mechanisms.

Risk testStored-carbon example evidenceOperational or use-based example evidenceDecision question
AdditionalityLand-use baseline, project incentives and applicable additionality testRegulatory context, existing practice and applicable additionality testWhat supports a change beyond what would otherwise happen?
QuantificationDefined pools, sampling or models, represented area and uncertaintyGas flow or usage evidence, equipment performance and missing-data treatmentWhich assumptions or observations drive the credited total?
DurabilityStorage duration, continuing monitoring, reversal provisions and remediesVerified period outcome, operating reliability and future-performance exposureCan a past outcome reverse, or is the risk that a future benefit will not occur?
LeakageEvidence of displaced land-use activity beyond the projectEffects outside the activity boundary, including displaced practices where relevantWhere could the claimed gain shift emissions rather than prevent them?
RightsLand and carbon rights, affected rights-holders and documented consent where applicableAuthority over assets and the activity, with participant rights where relevantWho has the authority to undertake and credit this activity?
SafeguardsCommunity and biodiversity assessment, grievance process and relevant benefit arrangementsParticipant safety, local impacts and access to a grievance processWhat harms are assessed, and what happens when concerns are raised?
Double countingIdentified units and checks for overlapping projects or claimsIdentified units and checks for overlapping issuance, use or claimsWhat prevents the same outcome or unit being counted more than once?

Table 1, shown one row at a time. Every field is retained.

Row 1

Risk test
Additionality
Stored-carbon example evidence
Land-use baseline, project incentives and applicable additionality test
Operational or use-based example evidence
Regulatory context, existing practice and applicable additionality test
Decision question
What supports a change beyond what would otherwise happen?

Row 2

Risk test
Quantification
Stored-carbon example evidence
Defined pools, sampling or models, represented area and uncertainty
Operational or use-based example evidence
Gas flow or usage evidence, equipment performance and missing-data treatment
Decision question
Which assumptions or observations drive the credited total?

Row 3

Risk test
Durability
Stored-carbon example evidence
Storage duration, continuing monitoring, reversal provisions and remedies
Operational or use-based example evidence
Verified period outcome, operating reliability and future-performance exposure
Decision question
Can a past outcome reverse, or is the risk that a future benefit will not occur?

Row 4

Risk test
Leakage
Stored-carbon example evidence
Evidence of displaced land-use activity beyond the project
Operational or use-based example evidence
Effects outside the activity boundary, including displaced practices where relevant
Decision question
Where could the claimed gain shift emissions rather than prevent them?

Row 5

Risk test
Rights
Stored-carbon example evidence
Land and carbon rights, affected rights-holders and documented consent where applicable
Operational or use-based example evidence
Authority over assets and the activity, with participant rights where relevant
Decision question
Who has the authority to undertake and credit this activity?

Row 6

Risk test
Safeguards
Stored-carbon example evidence
Community and biodiversity assessment, grievance process and relevant benefit arrangements
Operational or use-based example evidence
Participant safety, local impacts and access to a grievance process
Decision question
What harms are assessed, and what happens when concerns are raised?

Row 7

Risk test
Double counting
Stored-carbon example evidence
Identified units and checks for overlapping projects or claims
Operational or use-based example evidence
Identified units and checks for overlapping issuance, use or claims
Decision question
What prevents the same outcome or unit being counted more than once?

Record an evidence reference, an unresolved question and a decision for each material risk. “Document received” is not the same as “question answered”. If a forest document discusses a buffer but not the store or period it protects, the durability row remains incomplete. If an operational project supplies meter totals without explaining downtime or its baseline, the quantification and additionality rows may still need work.

Avoid compensating for a missing critical answer with strengths elsewhere. A good community narrative cannot prove the carbon quantity, and a strong monitoring system cannot establish ownership rights by itself. The matrix makes such dependencies visible. It is a practical buyer screen, not a substitute for the programme's assessment or an independent project audit.

A portfolio still needs project-by-project proof

A portfolio can spread exposure across activities, but the evidence still belongs to individual projects and units. A mix of forests, methane and cookstoves does not cure an unsupported baseline in one constituent project. Nor does a shared seller or programme ensure that all offered vintages have the same method, status or unresolved questions.

For each component, retain the project identifier, methodology and version, monitoring period or vintage, offered quantity, serial information when available, and the relevant registry status. Connect its screening decisions to those records. If an offer permits substitution, ask which attributes and evidence must remain equivalent and how the actual delivered units will be identified. A portfolio description can change while its marketing name stays the same.

Look for common assumptions as well as different labels. Projects in different activity families may still depend on the same disputed data source, a shared interpretation or incomplete records. The useful comparison identifies those dependencies and preserves exceptions. Treat any unresolved component as unresolved; do not hide it inside an average quality statement about the whole collection.

Keep company and project evidence distinct

Keep project evidence separate from the company's own emissions measurement. An internal inventory or product calculation describes the company's defined activities and boundaries. A project packet describes mitigation elsewhere and the status of associated units. Buying or retiring a unit does not silently revise the company's gross activity data or demonstrate an internal physical reduction.

A record system can connect the two for a clearly stated purpose while preserving their different meanings. Keep the inventory version, project documents, unit identity and intended wording distinguishable. Organising those records does not itself develop a project, verify its outcome, sell a credit or authorise a public environmental claim. Those roles require their own evidence and responsibilities.

Choose one project and ask for a traceable packet

Choose one named project from an actual offer and request a packet that can answer the sequence below. Starting with one complete case exposes missing links before a portfolio creates more records to chase.

  1. Identify the activity, claimed change and baseline, then obtain the programme and exact methodology version governing it.
  2. Connect the monitoring period to its data, reported outcome and relevant independent validation or verification records. Note qualifications rather than copying only a positive conclusion.
  3. Identify what is being offered: future delivery, issued units or a completed retirement. For units, reconcile project, vintage, quantity and serial information with the current registry trail.
  4. Complete the risk matrix and name the person responsible for resolving each material open question. Decide which evidence is sufficient for the intended use before approving any outward wording.

Stop if the project identifier cannot be matched to the documents, the method does not fit the described activity, the outcome cannot be traced to monitoring, or the offered unit status remains unclear. A seller's willingness to explain a gap is useful, but an explanation still needs support. A proposed future activity should remain described as proposed until the relevant evidence changes its status.

The final choice is about a supported claim, not the most attractive project category. A buyer should be able to say what changed, how the result was established, which units it concerns and what remains uncertain. If that chain is incomplete, preserve the question and narrow the wording rather than using a broad offsetting claim to bridge the missing evidence.