Reporting checked October 31, 2026. Numerical examples are hypothetical sensitivity calculations, not measured results for named projects. This is technical and public-policy analysis, not tax, investment or compliance advice.
1. A current industry needs more exact language
The International Energy Agency published its Global Hydrogen Review 2026 on June 18, 2026, alongside updated project and infrastructure information. Its description distinguishes announced projects from operating production and examines deployment, costs, policy and infrastructure. Those distinctions matter because the environmental performance of a proposed route and the verified performance of an operating supply chain are different kinds of evidence. IEA Global Hydrogen Review 2026.
Public discussion often begins somewhere simpler: a colour. Green, blue and other labels can signal broad ideas about how hydrogen is produced. But they do not specify the complete emissions boundary, the source of electricity at each hour, the upstream methane loss, the amount of carbon dioxide captured or the energy used to deliver the product. The label cannot answer questions it was never designed to encode.
A hydrogen molecule carries no visible certificate of its production history. Two supplies with the same physical composition can have very different associated emissions. Conversely, two projects using different technologies can fall closer together in measured intensity than their colour labels suggest. The comparison needs a defined quantity, boundary and accounting method.
The useful starting question is therefore what exactly is being claimed. Is the supplier reporting emissions at the production site, emissions from feedstock extraction through production, or emissions through delivery and use? Is the figure a modelled design estimate or a result calculated from operating data? Does it describe an annual average or a particular batch and period?
This article follows those questions through electrolysis, fossil-based production with capture, electricity matching and downstream conversion. It also explains why a standard's threshold is not a universal physical definition and why a low emissions figure at one boundary cannot automatically support a broader claim about an end use.
The aim is not to rank every hydrogen pathway with a single number. It is to show what information would make a comparison credible. Hydrogen can serve useful industrial and energy roles. Public support and commercial claims become more defensible when the emissions account follows the actual system rather than stopping where a favourable label begins.
2. Choose the functional unit before comparing projects
An emissions intensity is a ratio. The numerator is a quantity of greenhouse-gas emissions expressed under a specified method. The denominator is a quantity of product or service. A common hydrogen production measure is kilograms of carbon-dioxide equivalent per kilogram of hydrogen, but other frameworks use energy-based units. A comparison must align those units before interpreting a difference.
The US Department of Energy's June 2023 Clean Hydrogen Production Standard guidance uses a well-to-gate target of four kilograms of carbon-dioxide equivalent per kilogram of hydrogen. The UK Low Carbon Hydrogen Standard page identifies a version-four framework from January 2026 and an energy-based threshold using lower heating value. These are specific policy frameworks with defined methods, not interchangeable colour definitions. DOE guidance, UK standard.
A hypothetical supplier might report emissions per kilogram at the factory gate while a buyer wants emissions per unit of useful heat delivered at a facility. Converting between them requires more than changing units. It requires information about transport, storage and conversion efficiency. The same kilogram can provide different useful services in different equipment.
Purity, pressure and physical form can also affect comparability. Hydrogen delivered at one pressure may require additional compression before use. A product requiring further purification is not equivalent to a product already meeting the customer's specification. If the extra processing falls outside one supplier's boundary but inside another's, the headline intensities can be misleading.
The denominator should represent actual usable output over the reporting period. A plant's nameplate production capacity is not the amount it produced. Dividing annual emissions by an optimistic design output can understate operating intensity when actual production is lower. The calculation should use compatible time periods and measured or clearly estimated quantities.
Co-products introduce another choice. A process may produce useful oxygen, heat or other outputs. Allocating part of the emissions to those products can change the hydrogen result. The allocation rule should be disclosed and consistent with the chosen framework. A credit based on a hypothetical buyer for a co-product should not be treated as an observed displacement without evidence.
These choices do not make lifecycle analysis arbitrary. They make its definition essential. A comparison becomes meaningful when the functional unit, product specification and allocation method are explicit. Without them, a lower number can reflect a narrower question rather than a better-performing supply chain.
3. Draw the boundary all the way around the claim
A production-site account focuses on activities inside the facility. A well-to-gate account includes relevant upstream supply and ends at the production gate. A broader delivered-product account extends through transport and storage. A full end-use comparison also considers how the hydrogen supplies the desired service and what it replaces. Each boundary can be legitimate for a particular purpose.
The problem arises when the language expands but the calculation does not. A plant may accurately describe low direct carbon-dioxide emissions while advertising the product as having negligible lifecycle impact. A buyer may use a production certificate to imply a delivered-fuel result without adding downstream energy requirements. The wider statement needs wider evidence.
Consider a hypothetical project with production emissions of two kilograms of carbon-dioxide equivalent per kilogram of hydrogen at the gate. If delivery adds one kilogram on the same basis, the delivered figure becomes three. The production number remains correct. It simply does not answer the delivery question. A transparent report would preserve both rather than allowing one to stand in for the other.
A boundary diagram should identify inputs and outputs, including electricity, fuel, feedstock, water treatment, process heat, captured carbon dioxide and waste streams where relevant. It should state which infrastructure-related emissions are included and which are excluded. The aim is not to include every imaginable effect in every calculation, but to make the exclusions visible and consistent with the claim.
The time boundary matters too. A design study may use expected performance over a project lifetime. An operating report may cover one year with unusual outages or commissioning activity. These results can differ for valid reasons. The comparison should not mix a best-case steady-state model with a complete operating-year inventory without explaining the difference.
The IEA's 2024 emissions chapter distinguishes direct production emissions, upstream and midstream effects, electricity inputs and transport-related conversion. It provides a useful technical map, but its figures belong to that dated assessment and its assumptions. This article uses the distinctions rather than presenting its historical global estimates as current measurements. IEA emissions chapter.
A good public claim therefore names its endpoint. Low emissions at the production gate can be a meaningful achievement. It should be described that way until the evidence supports the delivered or end-use result. Precision about the boundary protects the achievement from being stretched into something the project has not established.
4. Electrolysis moves the emissions question to its energy supply
Electrolysis uses electricity to split water into hydrogen and oxygen. DOE's technical overview describes several electrolyser types and explains that high-temperature systems can use heat to reduce electrical demand. The reaction pathway therefore does not, by itself, determine the emissions associated with the energy used to run the system. DOE electrolysis overview.
For a simplified hypothetical calculation, assume a system uses 50 kilowatt-hours of electricity per kilogram of hydrogen and ignore other inputs temporarily. At an electricity emissions factor of 20 grams of carbon-dioxide equivalent per kilowatt-hour, that contribution is one kilogram per kilogram of hydrogen. At 400 grams per kilowatt-hour, it becomes 20 kilograms. The equipment's electricity requirement is unchanged; the assumed supply factor drives the difference.
This is a sensitivity example, not a claim about a particular electrolyser or grid. It shows why a statement that electrolysis produces no carbon dioxide in the water-splitting reaction cannot establish a low lifecycle result. The electricity and any heat supply need their own account, with units and boundaries that match the production calculation.
System electricity use should also be distinguished from stack electricity use. Pumps, controls, cooling, water treatment and other supporting equipment can consume energy. A manufacturer may legitimately report a stack performance metric, but a plant-level emissions assessment needs the relevant balance-of-plant demand as well. Comparing those two boundaries without adjustment can create an artificial efficiency advantage.
Operating conditions matter. A system may have different efficiency at different loads and may consume energy during startup, shutdown or standby. A single rated-point value does not automatically describe annual operation. The assessment should use measured operating data where available or a clearly specified performance model when the plant has not yet operated.
For high-temperature electrolysis, a lower electricity requirement can be accompanied by an external heat requirement. The heat cannot be treated as free of emissions merely because it is not electricity. Its source, availability and allocation need to be included under the chosen method. A comparison that counts one energy input and omits another can reward a bookkeeping choice rather than a real improvement.
The proper question is therefore the emissions of the complete energy supply per unit of usable hydrogen. Technology labels can guide the investigation, but operating data and supply assumptions determine the result. The public should be able to see which parts of that result were measured and which remain projections.
5. Annual electricity matching leaves timing unresolved
An electrolyser can consume electricity at times that differ from the output of the renewable generation associated with its contracts. Matching annual quantities establishes one relationship between consumption and procurement. It does not show that clean generation was available during every hour of production. The distinction is physical and temporal before it becomes a policy debate.
Imagine a hypothetical plant consuming 100 megawatt-hours every day. A contracted renewable resource produces the same annual total but concentrates output in different hours and seasons. The annual balance can be exact while the plant still relies on the grid during periods when the contracted resource produces little. The emissions interpretation depends on the accounting framework and the wider system response.
Hourly matching asks a more granular question: whether qualifying clean production aligns with consumption within each hour under the relevant rules. Geographic matching asks whether the associated generation is in a location meaningfully connected to the load. Additionality asks whether the procurement supports new clean supply rather than simply reallocating existing output. These are related dimensions, not synonyms.
A January 2026 research preprint by Michael Giovanniello and Dharik Mallapragada examines annual and hourly matching using a Texas capacity-expansion case study and multiple weather scenarios. Its abstract highlights the limits of relying on a single weather year. The study is model-based evidence, not a measurement of every hydrogen project's real-world consequences. Time-matching research preprint.
The broader analytical point is that matching rules need to be evaluated over realistic variability. A portfolio that works well in one weather year may require different backup or curtailment in another. A plant designed around an annual average can still face difficult hours when renewable output and hydrogen demand diverge.
Certificates can document attributes under a system of rules, but they do not trace individual electrons through a meshed grid. A credible claim explains what the certificate establishes, how double counting is prevented and what temporal and geographic relationship is required. The physical and contractual accounts should be connected without pretending they are identical.
The public should also avoid a false binary in which only direct connection has any value or every certificate proves zero emissions. Different arrangements can support useful investment and different levels of confidence. The appropriate conclusion follows the actual matching method, project configuration and evidence, not the presence of a renewable logo on a procurement document.
6. Average and marginal emissions answer different questions
An average electricity emissions factor allocates the emissions of a defined electricity supply across its output. A marginal analysis asks how emissions change when demand changes. Both can be useful, but they answer different questions. An emissions certificate under one accounting framework should not automatically be described as a causal estimate of the project's effect on the power system.
Suppose a hypothetical region has a low annual average emissions intensity because much of its electricity comes from low-emissions sources. An additional industrial load at a particular hour may still cause a higher-emitting unit to increase output. Conversely, a flexible load operating during otherwise curtailed clean generation may have a different incremental effect from the annual average. These examples illustrate the distinction rather than establish a specific grid result.
A consequential analysis needs a counterfactual power system. What would generation, investment and consumption have been without the hydrogen project? That is more demanding than multiplying measured electricity by an average factor. It requires assumptions about dispatch, transmission, future capacity and market response, and those assumptions can materially affect the answer.
An attributional inventory can remain valuable even when it does not answer that causal question. It provides a consistent accounting framework for comparing products under specified rules. The problem is overclaiming: using an accounting result to state that the project caused an equivalent reduction in global emissions without examining what changed elsewhere.
A public report should identify which type of analysis it uses and why it fits the decision. A certification program may need a consistent attributional method. A government deciding whether to support a large new load may also want a consequential sensitivity analysis. The two can be presented together without forcing their numbers to match.
Uncertainty should be handled openly. A marginal-emissions model can produce precise hourly values while depending on uncertain future conditions. The analyst should test alternative demand, generation and transmission assumptions. A range of outcomes can be more informative than a single exact figure attached to a long-lived project.
This distinction is particularly important when hydrogen is promoted as using surplus electricity. The claim needs evidence about when and where surplus exists, what would otherwise happen to it and whether the electrolyser's operating pattern matches those periods. Low electricity prices alone do not prove zero marginal emissions or permanent abundance of otherwise wasted energy.
7. Fossil-based hydrogen requires an upstream account
Natural-gas reforming uses methane as a feedstock and requires process energy. DOE's technical explanation describes steam reforming followed by a water-gas-shift reaction that produces additional hydrogen and carbon dioxide. That chemistry establishes why capturing carbon dioxide at the plant can matter, but it does not describe the entire upstream supply chain. DOE natural-gas reforming overview.
An emissions assessment should account for the relevant production, processing and transport of the gas under its defined boundary. Methane losses upstream can affect the lifecycle result even if the hydrogen plant performs well. The supplier's gas origin, measurement quality and chosen emissions factors therefore matter to a claim about low-emissions production.
A hypothetical calculation can illustrate sensitivity without pretending to represent a real supply chain. Suppose one assessed pathway has three kilograms of carbon-dioxide-equivalent emissions per kilogram of hydrogen from all components except a disputed upstream term. If that term is estimated at one kilogram, the total is four. If better evidence changes it to three, the total becomes six. A seemingly secondary input changes the classification under some thresholds.
The conversion of methane to carbon-dioxide equivalent also depends on the chosen climate metric and time horizon. A report should state that convention and use it consistently. Comparing two studies with different conventions can produce a difference that looks technological but is partly methodological.
Supplier-specific measurements can improve an inventory, but their representativeness needs examination. A short measurement campaign at selected facilities may not describe all sources or all operating conditions. An annual average can conceal episodic events. The analyst should explain how measured observations are extended to the full supply and what uncertainty remains.
The IEA's 2024 emissions assessment explicitly notes that carbon capture alone does not address all upstream and midstream emissions. That is the source-supported point here; this article does not assign a universal leakage rate or current emissions intensity to every gas-based project. IEA emissions assessment.
A colour label associated with capture therefore cannot settle the lifecycle question. The project needs a documented feedstock account, plant performance and downstream handling of captured carbon dioxide. Strong performance in one part of the chain should receive credit, while unresolved emissions elsewhere remain visible rather than disappearing into the label.
8. Capture percentages need a numerator and a denominator
A claim that a project captures a high percentage of carbon dioxide can refer to a particular process stream, the plant's total generated carbon dioxide or a wider lifecycle denominator. Those quantities differ. The reader needs to know which emissions are captured and which are outside the percentage before interpreting its environmental significance.
Consider a hypothetical plant generating ten units of carbon dioxide across two streams. Eight units arise in a stream equipped for capture, and two arise elsewhere. Capturing 95 percent of the first stream removes 7.6 units. Relative to all ten units, the capture fraction is 76 percent, not 95 percent. The smaller-stream percentage can be accurate while the broader interpretation is wrong.
The example does not imply that all plants have this configuration. It shows why the denominator must be explicit. A technical report should identify the streams included, the operating period and whether the figure is a design rating or measured performance. A best-hour result should not be presented as an annual average.
Capture systems can also require energy and supporting equipment. The lifecycle assessment should include relevant additional energy consumption under the chosen boundary. If the capture process changes fuel use or electricity demand, simply subtracting captured carbon dioxide from an unchanged baseline can misstate the net effect.
Transport and storage of captured carbon dioxide introduce further responsibilities. A production claim should explain whether the carbon is permanently stored, used in a way that releases it later or handled through another route. Capturing a stream at the plant gate is an intermediate operation; the climate account depends on its subsequent fate under the applicable method.
Operational downtime deserves attention. A plant may produce hydrogen during periods when capture is unavailable, depending on its design and rules. The emissions account should include those periods rather than assuming continuous rated capture. If production stops instead, the effect on output and economics should be reflected consistently.
A credible disclosure therefore reports absolute quantities as well as percentages: hydrogen produced, carbon dioxide generated within the defined scope, carbon dioxide captured and relevant uncaptured emissions. These quantities make it possible to check the ratio and connect the capture claim to the product's emissions intensity. The percentage then becomes useful evidence rather than a substitute for the rest of the system.
9. Delivery can change the product's emissions intensity
Hydrogen may require compression, storage, transport or conversion into another carrier before reaching a user. Those operations can consume energy and lose material. A low production-gate intensity does not automatically remain unchanged at the destination. The delivered product needs its own denominator and added process account.
The IEA's 2024 emissions chapter discusses energy losses associated with converting hydrogen into transport carriers and recovering it. Its specific pathway comparisons depend on the report's assumptions. The general lesson is that downstream conversion can amplify upstream emissions per unit of final delivered hydrogen when more initial production is required to supply that unit. IEA hydrogen-derivatives analysis.
Suppose a hypothetical route delivers only 80 percent of the initially produced hydrogen quantity on an equivalent usable-output basis. To deliver one unit, it needs 1.25 units upstream. If production emissions are two units per unit produced, their contribution becomes 2.5 per unit delivered before adding transport or conversion emissions. This is arithmetic about a defined yield, not a claim about any specific transport technology.
The analyst must be careful about what the yield means. Material loss, energy conversion efficiency and chemical conversion are related but different concepts. A percentage describing energy retained in a carrier should not automatically be treated as a percentage of hydrogen molecules lost. The calculation needs a consistent mass or energy basis throughout.
Delivery specifications can also change the comparison. A user needing compressed gaseous hydrogen has a different supply requirement from a user consuming a derivative directly. Converting a derivative back into hydrogen when the end use could consume it without reconversion may add a stage that another pathway avoids. The functional service should determine the comparison.
Storage time matters where losses or energy consumption continue during holding. A model assuming immediate delivery may understate the burden of a supply chain that keeps inventory for long periods. Conversely, a conservative long-storage assumption may overstate impacts for a short local route. The reporting should describe the actual or planned logistics rather than a generic distance alone.
The buyer should therefore ask for both production and delivered intensities when delivery is material to the claim. Keeping the two figures separate makes the supply chain easier to improve: it reveals whether the main opportunity lies in production, conversion, transport or end-use efficiency. A single colour label cannot identify that opportunity.
10. Hydrogen leakage deserves its own evidence
Hydrogen escaping into the atmosphere is a different issue from carbon dioxide emitted during production. The UK's Environment Agency monitoring review explains that atmospheric hydrogen can extend methane's lifetime and that smaller leaks need attention beyond traditional safety monitoring. The review supports a monitoring concern; it does not justify assigning an invented leakage rate to a project. Environment Agency monitoring review.
Safety detection and climate accounting can have different measurement objectives. A system designed to identify dangerous concentrations near equipment may not quantify small total emissions across a facility or supply chain. A public claim about fugitive emissions needs to state what the monitoring system can detect and how observations are converted into an annual quantity.
A hypothetical facility might report that no safety alarm occurred during a year. That is relevant to the alarm's purpose, but it does not establish zero leakage. The detection threshold, sensor placement, maintenance and operating conditions determine what the absence of an alarm means. An emissions inventory needs evidence appropriate to its own question.
Mass balance can help but may be difficult when the expected loss is small relative to measurement uncertainty. If production and delivery meters each have uncertainty comparable with the suspected loss, subtracting their readings can produce an unstable estimate. The report should examine instrument accuracy and uncertainty before describing the difference as measured leakage.
Direct measurements, component estimates and operational records may need to be combined. Maintenance events, venting and unusual conditions can contribute differently from routine operation. A representative account should state which categories are measured, which are estimated and which remain uncertain. It should avoid silently replacing missing data with zero.
The UK standard's January 2026 page includes a requirement for a fugitive-hydrogen risk-mitigation plan and annual reporting within its framework. That is a jurisdiction-specific compliance feature, not a universal rule for every producer. It demonstrates that hydrogen loss is a distinct reporting issue rather than something a production colour automatically resolves. UK Low Carbon Hydrogen Standard.
The appropriate conclusion is proportionate. Leakage can affect the environmental account and deserves measurement. Uncertainty about it should motivate better evidence, not a confident assertion that all hydrogen systems are either harmless or ineffective. The supply chain should be evaluated with the same care given to its electricity and feedstock inputs.
11. Utilization links economics and emissions without making them identical
An electrolyser's utilization affects how much output is produced from a capital investment. Running more hours can spread some fixed costs over more hydrogen. But the additional hours may have different electricity prices and emissions. A strategy that minimizes production cost under one set of assumptions may not minimize emissions intensity.
The JRC describes research on electrolyser performance, durability and testing under different operating conditions, while noting remaining knowledge gaps about costs and effective applications. This supports the need for operating-profile evidence rather than assuming that a rated efficiency establishes annual performance. JRC electrolyser research.
Consider a hypothetical plant choosing between operating only during low-emissions hours and operating more continuously. The flexible strategy may need more installed equipment or storage to supply the same annual hydrogen demand. The continuous strategy may use more carbon-intensive electricity in some hours. A fair comparison includes both the changed operating profile and the changed system configuration.
The buyer's demand pattern matters too. A customer requiring steady delivery may need storage or another supply arrangement when production is intermittent. The emissions and cost of that arrangement belong in the delivered-service analysis. It is not enough to show that the electrolyser can operate cleanly during selected hours if the commercial promise requires supply at other times as well.
Equipment degradation can further affect the trajectory. Performance at commissioning may differ from performance after years of operation. Replacement and maintenance assumptions should be explicit in a long-term model. A projected cost or emissions figure should identify whether it assumes constant efficiency, scheduled replacement or another treatment of aging.
The analyst should resist combining the most favorable assumptions from incompatible scenarios. A model cannot simultaneously assume very cheap intermittent electricity, continuous high utilization and no cost for balancing the mismatch unless the configuration genuinely supports those conditions. Each scenario needs an internally consistent energy and production balance.
This is where a transparent sensitivity analysis becomes valuable. It can show how the result changes with electricity intensity, utilization, equipment efficiency and delivery requirements. The purpose is not to produce an endless range that avoids conclusions. It is to identify which assumptions dominate and what evidence would most improve confidence in the project's claimed performance.
12. Standards and certificates need version control
A standard defines an accounting method and eligibility conditions within a particular framework. It may specify boundaries, factors, evidence requirements and thresholds. A certificate can establish conformity with that framework if issued and verified appropriately. It does not automatically establish compliance with another jurisdiction's rules or prove every broader environmental claim.
The UK government's January 2026 update identifies version four of its standard and revised data annexes and calculators. The changes include updated factors and methodologies. That is a practical reminder that an emissions result is connected to a method version, not merely to the name of a scheme. UK standard and update record.
A hypothetical project assessed under an earlier version may obtain a different result when a factor or boundary changes, even if its physical operation remains the same. The report should distinguish a methodological revision from an actual performance improvement. Otherwise a lower published intensity can be mistaken for an engineering achievement when it reflects changed accounting.
The evidence package should identify the standard version, reporting period, facility boundary, data sources and verifier's scope. It should state whether the assessment concerns a prospective design or operating production. It should also explain how electricity attributes and other claims are tracked to avoid more than one party using the same environmental attribute inconsistently.
A buyer comparing certificates should check that the products and boundaries align. One certificate may concern production only, another delivered hydrogen, and another a derivative with different conversion assumptions. The presence of a seal does not make these results directly comparable. The underlying method remains part of the product information.
Commercial confidentiality can limit publication of detailed operating data, but a useful public summary can still explain the method and the assurance scope. It can identify which inputs were measured, estimated or supplied by third parties. A claim of independent verification should say what was verified rather than implying an unlimited guarantee.
For public discussion, the safest habit is to cite the applicable version and avoid giving an informal colour label the force of a legal conclusion. This article does not determine eligibility for any subsidy or tax credit. It explains why such determinations require the actual current rules and project evidence, not a general description of the production technology.
13. Compare the end use with a real alternative
A low-emissions hydrogen supply can still be an inefficient way to provide a particular service if another feasible option uses less energy or creates fewer emissions. Conversely, some industrial uses require hydrogen as a chemical input rather than merely as an energy carrier. The appropriate alternative depends on the job being performed.
A hypothetical industrial buyer should define the service first: a chemical reaction, a unit of heat at a required temperature, a transport task or stored energy returned later. The comparison then follows each feasible pathway through its equipment and supply chain. Comparing kilograms of hydrogen with kilowatt-hours of electricity without accounting for the service is not enough.
The alternative must be realistic. A model should not compare a proposed hydrogen system with an obsolete, unusually inefficient incumbent if a more relevant replacement is available. Nor should it assume an alternative that cannot meet the required process conditions. The analyst should explain the technical feasibility and boundaries of both sides.
Avoided emissions depend on that counterfactual. A production certificate may show that the hydrogen has a low attributed intensity. The claim that buying it reduces emissions by a particular amount also depends on what production or fuel use is displaced and whether the change actually occurs. These are additional causal questions, not automatic consequences of certification.
Infrastructure timing can affect the decision. A project may have low operating emissions in a future scenario but require years of construction and new delivery facilities. A public investment analysis should state the schedule and interim assumptions rather than treating the future system as available today. Announced capacity and operating supply remain separate categories.
The IEA's 2026 review explicitly examines acceptable costs across applications and regions, reflecting the fact that hydrogen's role cannot be evaluated independently of its use. This article does not borrow an unverified universal cost threshold from that report; the relevant point is the need for application-specific comparison. IEA 2026 review.
A well-framed decision can therefore conclude that hydrogen is useful in one setting and unattractive in another without inconsistency. The evidence follows the service, the available alternatives and the complete supply chain. A colour label cannot resolve that comparison because it says too little about both the hydrogen system and what it would replace.
14. The claim should travel with an emissions account
A credible hydrogen claim can be summarized in a compact set of facts: the product specification, reporting period, system boundary, method version, measured production and emissions intensity with its main uncertainties. Supporting records should identify electricity, feedstocks, capture performance and relevant downstream operations. The exact details vary by route, but the logic remains consistent.
The account should separate design expectations from operating results. A proposed plant can publish a well-supported forecast, including sensitivity to uncertain inputs. Once operating, it should compare that forecast with measured performance and explain material differences. A favourable design study should not remain the sole evidence after years of actual operation become available.
Absolute quantities should accompany intensities where possible. A low intensity applied to a small amount of production is a different contribution from the same intensity at sustained commercial output. Likewise, a growing plant can reduce intensity while increasing total emissions. Both dimensions matter, depending on the public claim being made.
Uncertainty should be concentrated where it is most consequential. If electricity supply dominates the result, hourly consumption and matching evidence deserve attention. If upstream methane is decisive, the supply-chain data need scrutiny. If delivery conversions create large losses, the analysis should follow the product beyond the gate. A long checklist is less useful than identifying the inputs that can change the conclusion.
Corrections should remain visible. Revised emissions factors, improved measurements or changed operating conditions can alter the reported result. A versioned record allows readers to distinguish those causes. It also permits a supplier to demonstrate real improvement rather than relying on a series of incomparable numbers.
Hydrogen's environmental value does not require every project to look identical. Different routes can contribute under different conditions. What they share is the need to connect the physical process with an honest account of its inputs, outputs and consequences. That account should be specific enough for a buyer, regulator or independent reviewer to challenge it.
The colour can remain a conversational starting point. It should not be the conclusion. The question that matters is how much climate impact is associated with delivering the required service, under a stated method and supported by current evidence. Following that question through the whole system is the difference between a label that reassures and a claim that can be trusted.