Life cycle assessment (LCA) studies and disclosures under the EU’s Corporate Sustainability Reporting Directive (CSRD) runs through thousands of individual claims: emission factors, product category rule (PCR) clauses, double materiality assessments covering both a company’s environmental impact and the financial risks it faces. Each claim requires a qualified human sign off. However, there simply aren’t enough qualified reviewers to do that by hand at the volume the frameworks now require. Sustainability reporting has a volume problem before it has an accuracy problem, and that is the opening AI walks into.
\nAI hasn’t earned its way into the review role. It was pulled in because the alternative, manual review at current volume, doesn’t scale. That distinction matters, as trust in AI as a reviewer isn’t established, it’s assumed by necessity, and the two are not the same thing. An institution can adopt a tool because it has no better option and still owe itself a hard look at whether the tool deserves the role it’s been handed.
\nThe volume hasn’t eased either. The EU’s Omnibus I Directive narrowed how many companies fall under CSRD, but the companies still in scope face the same double materiality reporting requirements as before. Fewer filers, same depth per filer. The pressure that pulled AI into review in the first place hasn’t relaxed, it’s just concentrated on a smaller set of reports that are each just as dense as they were. A narrower scope was never going to be the thing that settled the question of whether AI belongs in the review chain.
\nWhat AI actually changes is what gets checked at all. A human reviewer working through a report has to sample: pull a handful of emission factors, spot check a few PCR clauses, and hope the pattern holds across the rest. Fatigue sets in a few hundred datapoints into a report that runs to thousands, and the inconsistencies that slip through tend to be the quiet ones, a boundary assumption that shifted halfway through, a clause addressed in spirit but not in the specific wording the standard requires. AI can cross-check every emission factor against its source, flag every PCR clause that’s only partially addressed, and catch exactly those quiet inconsistencies. That’s not a marginal improvement on the old process. It’s a different process altogether, checking a different fraction of the report, and a wider fraction checked isn’t automatically a more reliable check. Coverage and trust are two different questions. One asks how much of the report got checked and the other asks how much confidence we have on the checker itself. AI answers the first convincingly however, the second still remains unresolved.
\nWhich raises the actual question: whether or not AI should review sustainability data, since the volume math already answered that, but whether AI review can avoid becoming the failure point the review step exists to catch.
\nThe standards governing this work were not written with a non-human reviewer in mind. The International Organization for Standardization’s (ISO) 14044 critical review, a PCR verifier’s sign off, Greenhouse Gas (GHG) Protocol assurance: all three are built around a named, accountable person. The artifact of trust in each case isn’t just the number that got approved but the professional judgment of the person who approved it, and the fact that they can be asked, later and by someone else, to walk through why.
\nAI doesn’t fit cleanly into that role. It can flag an anomaly, but it can’t be cross examined about its reasoning the way a verifier can. A verifier who accepted an unusual boundary assumption can explain, on the record, why that assumption was reasonable for this product and this context. A model that flagged or cleared the same assumption has no equivalent account to give. Whether that gap matters comes down to governance: governance is what decides if AI stays a support function underneath the accountable person, or quietly erodes the premise that a human is the one actually answering for the result.
\nGood governance follows a specific sequence. AI flags an issue, a qualified human confirms or overrides that flag, and provides reasoning. That decision gets logged against the exact model and prompt version that produced the flag, because a model updated between one reporting cycle and the next can flag differently on the same underlying data, and a decision log that doesn’t say which version was in use can’t tell anyone whether that’s what happened.
\nBreak any one link and the chain stops doing its job. An AI flag nobody acts on is noise, indistinguishable from a false positive. There is no way to tell whether or not it was actually catching something real, because acting on it is what proves its value. The same logic applies to a human override, with no stated reasoning, nobody can reconstruct why the call was made, neither by an auditor checking the file nor by the reviewer’s own future self trying to remember the call. A model update that isn’t tracked against the decisions it produced makes last year’s results and this year’s results incomparable, even when nothing else about the underlying report has changed.
\nGovernance is the difference between “AI touched this” and something a person can actually trace, question, and defend. Without it, the first phrase is all anyone can honestly say, and it isn’t an answer to the question an auditor is actually asking.
\nCSRD assurance itself is still being defined. Omnibus I dropped what had been a planned move toward reasonable assurance, a higher evidentiary bar than the limited assurance most disclosures still face. The European Commission is due to publish harmonized limited assurance standards by mid 2027. Whatever bar eventually applies, the assurance statement is still issued by a named, accredited provider who is accountable for it, regardless of whether or not AI touched the file along the way.
\nThe clearest version of the pattern so far is showing up earlier in the pipeline, on the preparation side of environmental product declarations (EPD). AI tools are already interpreting PCR requirements, flagging data gaps, and suggesting system boundaries. What hasn’t moved is the step after: an independent, accredited verifier still has to review the submission and sign off before an EPD can be published. This pattern isn’t a settled template across sustainability reporting yet, but more a shape both frameworks keep landing on: whatever AI touches earlier in the process, a named person still has to be the one who signs at the end.
\nThis division changes what the human reviewer’s job actually is, in both directions at once. It narrows in scope: judging the anomalies AI surfaces rather than trying to catch everything cold across a report, rather than reading too closely. Such procedures deepens accountability: the reasoning behind a confirmation or an override has to be explicit enough to survive an audit, not just sit in the reviewer’s head as professional intuition nobody asked them to write down.
\nTrust in AI assisted review will end up resting on the strength and traceability of such governance trails, not on how fast or sophisticated the underlying tool is. A faster model that skips the logging step is not a more trustworthy reviewer. It’s an unaccountable one that happens to work quickly, and speed was never the thing the review step was built to protect. The review step exists to catch failure and a faster failure point is still a failure point nonetheless.
", "content_text": "Life cycle assessment (LCA) studies and disclosures under the EU's Corporate Sustainability Reporting Directive (CSRD) runs through thousands of individual claims: emission factors, product category rule (PCR) clauses, double materiality assessments covering both a company's environmental impact and the financial risks it faces. Each claim requires a qualified human sign off. However, there simply aren't enough qualified reviewers to do that by hand at the volume the frameworks now require. Sustainability reporting has a volume problem before it has an accuracy problem, and that is the opening AI walks into.\n\n## Necessity before trust\n\nAI hasn't earned its way into the review role. It was pulled in because the alternative, manual review at current volume, doesn't scale. That distinction matters, as trust in AI as a reviewer isn't established, it's assumed by necessity, and the two are not the same thing. An institution can adopt a tool because it has no better option and still owe itself a hard look at whether the tool deserves the role it's been handed.\n\nThe volume hasn't eased either. The EU's Omnibus I Directive narrowed how many companies fall under CSRD, but the companies still in scope face the same double materiality reporting requirements as before. Fewer filers, same depth per filer. The pressure that pulled AI into review in the first place hasn't relaxed, it's just concentrated on a smaller set of reports that are each just as dense as they were. A narrower scope was never going to be the thing that settled the question of whether AI belongs in the review chain.\n\nWhat AI actually changes is what gets checked at all. A human reviewer working through a report has to sample: pull a handful of emission factors, spot check a few PCR clauses, and hope the pattern holds across the rest. Fatigue sets in a few hundred datapoints into a report that runs to thousands, and the inconsistencies that slip through tend to be the quiet ones, a boundary assumption that shifted halfway through, a clause addressed in spirit but not in the specific wording the standard requires. AI can cross-check every emission factor against its source, flag every PCR clause that's only partially addressed, and catch exactly those quiet inconsistencies. That's not a marginal improvement on the old process. It's a different process altogether, checking a different fraction of the report, and a wider fraction checked isn't automatically a more reliable check. Coverage and trust are two different questions. One asks how much of the report got checked and the other asks how much confidence we have on the checker itself. AI answers the first convincingly however, the second still remains unresolved.\n\nWhich raises the actual question: whether or not AI should review sustainability data, since the volume math already answered that, but whether AI review can avoid becoming the failure point the review step exists to catch.\n\n## A trust built for a person\n\nThe standards governing this work were not written with a non-human reviewer in mind. The International Organization for Standardization's (ISO) 14044 critical review, a PCR verifier's sign off, Greenhouse Gas (GHG) Protocol assurance: all three are built around a named, accountable person. The artifact of trust in each case isn't just the number that got approved but the professional judgment of the person who approved it, and the fact that they can be asked, later and by someone else, to walk through why.\n\nAI doesn't fit cleanly into that role. It can flag an anomaly, but it can't be cross examined about its reasoning the way a verifier can. A verifier who accepted an unusual boundary assumption can explain, on the record, why that assumption was reasonable for this product and this context. A model that flagged or cleared the same assumption has no equivalent account to give. Whether that gap matters comes down to governance: governance is what decides if AI stays a support function underneath the accountable person, or quietly erodes the premise that a human is the one actually answering for the result.\n\n## The chain that has to hold\n\nGood governance follows a specific sequence. AI flags an issue, a qualified human confirms or overrides that flag, and provides reasoning. That decision gets logged against the exact model and prompt version that produced the flag, because a model updated between one reporting cycle and the next can flag differently on the same underlying data, and a decision log that doesn't say which version was in use can't tell anyone whether that's what happened.\n\nBreak any one link and the chain stops doing its job. An AI flag nobody acts on is noise, indistinguishable from a false positive. There is no way to tell whether or not it was actually catching something real, because acting on it is what proves its value. The same logic applies to a human override, with no stated reasoning, nobody can reconstruct why the call was made, neither by an auditor checking the file nor by the reviewer's own future self trying to remember the call. A model update that isn't tracked against the decisions it produced makes last year's results and this year's results incomparable, even when nothing else about the underlying report has changed.\n\nGovernance is the difference between \"AI touched this\" and something a person can actually trace, question, and defend. Without it, the first phrase is all anyone can honestly say, and it isn't an answer to the question an auditor is actually asking.\n\n## The template already forming\n\nCSRD assurance itself is still being defined. Omnibus I dropped what had been a planned move toward reasonable assurance, a higher evidentiary bar than the limited assurance most disclosures still face. The European Commission is due to publish harmonized limited assurance standards by mid 2027. Whatever bar eventually applies, the assurance statement is still issued by a named, accredited provider who is accountable for it, regardless of whether or not AI touched the file along the way.\n\nThe clearest version of the pattern so far is showing up earlier in the pipeline, on the preparation side of environmental product declarations (EPD). AI tools are already interpreting PCR requirements, flagging data gaps, and suggesting system boundaries. What hasn't moved is the step after: an independent, accredited verifier still has to review the submission and sign off before an EPD can be published. This pattern isn't a settled template across sustainability reporting yet, but more a shape both frameworks keep landing on: whatever AI touches earlier in the process, a named person still has to be the one who signs at the end.\n\n## Judgment, not scale\n\nThis division changes what the human reviewer's job actually is, in both directions at once. It narrows in scope: judging the anomalies AI surfaces rather than trying to catch everything cold across a report, rather than reading too closely. Such procedures deepens accountability: the reasoning behind a confirmation or an override has to be explicit enough to survive an audit, not just sit in the reviewer's head as professional intuition nobody asked them to write down.\n\nTrust in AI assisted review will end up resting on the strength and traceability of such governance trails, not on how fast or sophisticated the underlying tool is. A faster model that skips the logging step is not a more trustworthy reviewer. It's an unaccountable one that happens to work quickly, and speed was never the thing the review step was built to protect. The review step exists to catch failure and a faster failure point is still a failure point nonetheless.", "image": "https://verdatir.com/_astro/ai-reviews-sustainability-hero.CYpaHuMV.jpg", "date_published": "2026-08-17T00:00:00.000Z", "date_modified": "2026-08-17T00:00:00.000Z", "authors": [ { "name": "Verdatir", "url": "https://verdatir.com/vision" } ], "tags": [ "Research", "AI", "CSRD", "Assurance", "Governance", "EPD", "PCR" ], "language": "en", "_verdatir": { "markdown": "https://verdatir.com/insights/ai-is-reviewing-sustainability-data-who-reviews-ai.md" } }, { "id": "https://verdatir.com/insights/verification-infrastructure-is-the-product", "url": "https://verdatir.com/insights/verification-infrastructure-is-the-product", "title": "Verification Infrastructure Is the Product", "summary": "Sustainability standards rely on accredited verifiers, but the real product of any programme is the verification infrastructure behind it. Consistency, independence, scalable capacity, and continuous feedback loops are what make a standard trustworthy in the market. Verdatir argues that verification should be engineered as a core design element, not treated as background plumbing.", "content_html": "Sustainability standards programmes work by delegation. The body that writes the rules, whether that is EPD International’s Product Category Rules (PCRs) or the EU’s Corporate Sustainability Reporting Directive (CSRD), hands the actual checking to a separate accredited verifier. The standard gets set in one place and confirmed in another, and the programme operator’s job stops well short of the moment a specific project or company is judged compliant.
\nThat split leaves a mismatch nobody designs away on its own. Trust is handed off to the verifier, but the reputational risk stays with the programme operator’s name. When a certified project turns out to be wrong, the market does not go looking for the accreditation body that signed off on it. It blames the standard, because the standard is the name it recognizes.
\nThe instinct is to treat verification as plumbing: necessary, but not part of the programme’s core design, something that runs in the background. That instinct is wrong, because a methodology is never applied mechanically, it is interpreted. Two accredited verifiers working from the same rulebook can read a boundary condition, an allocation choice, or a data gap differently, and both can be technically compliant while landing on different answers.
\nWhat holds a programme together is not each verifier’s individual compliance. It is the consistency across all of them. A single verifier can pass every audit of its own work and still be part of the problem, if the answer it reaches would not survive being handed to a different verifier working the same file. Without cross-checks and shared guidance, the market stops trusting the standard itself and starts trusting specific verifiers instead, which defeats the point of having a common standard in the first place.
\nEPD International shows what closing that gap looks like in practice. Its Product Category Rules set the technical rules for a given product category, but the rules alone do not stop the same rule being read differently across countries, industries, or individual verifiers. General Programme Instructions and PCR review committees exist specifically to catch this drift and correct it before it reaches the market, functioning as the mechanism that keeps a PCR meaning consistent with wherever it gets applied.
\nThe other structural weakness sits in who pays. Under the issuer-pays model, the project developer or the reporting company selects and pays its own verifier. That is a built-in incentive conflict, not a hypothetical one. The verifier’s commercial relationship runs directly to the party whose compliance it is judging, and no amount of professional integrity on the verifier’s side removes that structure by itself. Independence has to be engineered into the system. It cannot be assumed just because the verifier holds accreditation.
\nThe real safeguards are structural, not procedural. Rotating verifiers so no single relationship sets into familiarity. Findings kept visible to parties beyond the client who is paying for them, so a lenient call has an audience that notices. Overseeing bodies with actual sanction authority, not an advisory role that stops at a strongly worded letter.
\nCSRD assurance runs on the same structure. Companies choose and pay their own assurance provider, so the identical conflict sits underneath Europe’s flagship reporting regime and under product declaration schemes. The lesson remains the same for both: independence should be checked, not assumed, and a programme that skips the check is trusting the incentive structure to correct itself.
\nNone of this works if there are not enough qualified verifiers to go around. A safeguard like rotation only functions if there is a genuine pool of independent verifiers to rotate between, rather than a handful of firms trading the same clients back and forth. When supply lags demand, a programme operator has two options, and both are costly in different ways. It can delay issuance, which is visible, unpopular, and gets corrected by market pressure fairly quickly because everyone can see the backlog. Or it can lower accreditation standards to clear it, which is invisible right up until the damage it caused surfaces later.
\nSweden’s SWEDAC, accredited within the European cooperation for Accreditation network, shows the bottleneck concretely. As CSRD and EPD demand both grow, they draw on the same limited pool of nationally accredited capacity, and that capacity becomes the binding constraint on the whole system rather than the methodology or the willingness of companies to comply. A standard can be well-written and still fail in the market if the pipeline of people qualified to check it cannot keep pace with the number of projects and disclosures that need checking.
\nVerification methods built for periodic, sample-based audits assume a world that mostly no longer exists. That model made sense when the underlying data was static and expensive to collect. Satellite imagery, Internet of Things (IoT) sensors, and remote sensing now generate continuous data about the thing being verified, not a snapshot pulled once a year and treated as representative of the whole period.
\nThe shift that verification infrastructure needs to make is from a one-time, retrospective judgment to continuous reconciliation: live data checked against the rules as it arrives, with people handling the edge cases and interpretation calls that data alone cannot resolve. That is a different job than auditing, closer to monitoring with judgment built in, and most verification infrastructures are not built to do it.
\nCSRD’s digital tagging requirement pushes reporting toward machine-readable data, which is a step in that direction. But the EU’s detailed assurance standard covering how that machine-readable data actually gets checked is not due until 1 October 2026. Until then, operators are left to build the bridge themselves, deciding case by case how much of the old sample-based approach still applies to data that no longer arrives in samples.
\nVerifiers are the ones who see where a methodology actually breaks down in the field, before anyone else does. They are the first to hit the ambiguous case, the boundary the rule did not anticipate, the allocation question the drafters never considered. That makes them a source of information the standard itself needs, not just an enforcement layer sitting downstream of it. Operators that build a deliberate feedback loop with verification findings, end up with a standard that improves with use, which is roughly what has happened with European Sustainability Reporting Standards (ESRS) updates and with EPD’s General Programme Instructions revisions.
\nWhat a market actually trusts is never the methodology document by itself. It is proof that the document was applied consistently, independently, and at scale, across every verifier working under it and every year the programme runs. That proof is the verification infrastructure, and it is the programme’s real product, whether or not the programme was ever designed to treat it that way.
\nThis perspective is part of Verdatir’s ongoing research on verification infrastructure, programme governance, and the future of sustainability assurance.
", "content_text": "Sustainability standards programmes work by delegation. The body that writes the rules, whether that is EPD International's Product Category Rules (PCRs) or the EU's Corporate Sustainability Reporting Directive (CSRD), hands the actual checking to a separate accredited verifier. The standard gets set in one place and confirmed in another, and the programme operator's job stops well short of the moment a specific project or company is judged compliant.\n\nThat split leaves a mismatch nobody designs away on its own. Trust is handed off to the verifier, but the reputational risk stays with the programme operator's name. When a certified project turns out to be wrong, the market does not go looking for the accreditation body that signed off on it. It blames the standard, because the standard is the name it recognizes.\n\n## Consistency Matters More Than Individual Compliance\n\nThe instinct is to treat verification as plumbing: necessary, but not part of the programme's core design, something that runs in the background. That instinct is wrong, because a methodology is never applied mechanically, it is interpreted. Two accredited verifiers working from the same rulebook can read a boundary condition, an allocation choice, or a data gap differently, and both can be technically compliant while landing on different answers.\n\nWhat holds a programme together is not each verifier's individual compliance. It is the consistency across all of them. A single verifier can pass every audit of its own work and still be part of the problem, if the answer it reaches would not survive being handed to a different verifier working the same file. Without cross-checks and shared guidance, the market stops trusting the standard itself and starts trusting specific verifiers instead, which defeats the point of having a common standard in the first place.\n\nEPD International shows what closing that gap looks like in practice. Its Product Category Rules set the technical rules for a given product category, but the rules alone do not stop the same rule being read differently across countries, industries, or individual verifiers. General Programme Instructions and PCR review committees exist specifically to catch this drift and correct it before it reaches the market, functioning as the mechanism that keeps a PCR meaning consistent with wherever it gets applied.\n\n## Independence Has to Be Engineered\n\nThe other structural weakness sits in who pays. Under the issuer-pays model, the project developer or the reporting company selects and pays its own verifier. That is a built-in incentive conflict, not a hypothetical one. The verifier's commercial relationship runs directly to the party whose compliance it is judging, and no amount of professional integrity on the verifier's side removes that structure by itself. Independence has to be engineered into the system. It cannot be assumed just because the verifier holds accreditation.\n\nThe real safeguards are structural, not procedural. Rotating verifiers so no single relationship sets into familiarity. Findings kept visible to parties beyond the client who is paying for them, so a lenient call has an audience that notices. Overseeing bodies with actual sanction authority, not an advisory role that stops at a strongly worded letter.\n\nCSRD assurance runs on the same structure. Companies choose and pay their own assurance provider, so the identical conflict sits underneath Europe's flagship reporting regime and under product declaration schemes. The lesson remains the same for both: independence should be checked, not assumed, and a programme that skips the check is trusting the incentive structure to correct itself.\n\n## Verifier Supply Has to Scale Ahead of Demand\n\nNone of this works if there are not enough qualified verifiers to go around. A safeguard like rotation only functions if there is a genuine pool of independent verifiers to rotate between, rather than a handful of firms trading the same clients back and forth. When supply lags demand, a programme operator has two options, and both are costly in different ways. It can delay issuance, which is visible, unpopular, and gets corrected by market pressure fairly quickly because everyone can see the backlog. Or it can lower accreditation standards to clear it, which is invisible right up until the damage it caused surfaces later.\n\nSweden's SWEDAC, accredited within the European cooperation for Accreditation network, shows the bottleneck concretely. As CSRD and EPD demand both grow, they draw on the same limited pool of nationally accredited capacity, and that capacity becomes the binding constraint on the whole system rather than the methodology or the willingness of companies to comply. A standard can be well-written and still fail in the market if the pipeline of people qualified to check it cannot keep pace with the number of projects and disclosures that need checking.\n\n## From Periodic Audits to Continuous Reconciliation\n\nVerification methods built for periodic, sample-based audits assume a world that mostly no longer exists. That model made sense when the underlying data was static and expensive to collect. Satellite imagery, Internet of Things (IoT) sensors, and remote sensing now generate continuous data about the thing being verified, not a snapshot pulled once a year and treated as representative of the whole period.\n\nThe shift that verification infrastructure needs to make is from a one-time, retrospective judgment to continuous reconciliation: live data checked against the rules as it arrives, with people handling the edge cases and interpretation calls that data alone cannot resolve. That is a different job than auditing, closer to monitoring with judgment built in, and most verification infrastructures are not built to do it.\n\nCSRD's digital tagging requirement pushes reporting toward machine-readable data, which is a step in that direction. But the EU's detailed assurance standard covering how that machine-readable data actually gets checked is not due until 1 October 2026. Until then, operators are left to build the bridge themselves, deciding case by case how much of the old sample-based approach still applies to data that no longer arrives in samples.\n\n## Verification Findings Should Feed Back Into the Standard\n\nVerifiers are the ones who see where a methodology actually breaks down in the field, before anyone else does. They are the first to hit the ambiguous case, the boundary the rule did not anticipate, the allocation question the drafters never considered. That makes them a source of information the standard itself needs, not just an enforcement layer sitting downstream of it. Operators that build a deliberate feedback loop with verification findings, end up with a standard that improves with use, which is roughly what has happened with European Sustainability Reporting Standards (ESRS) updates and with EPD's General Programme Instructions revisions.\n\nWhat a market actually trusts is never the methodology document by itself. It is proof that the document was applied consistently, independently, and at scale, across every verifier working under it and every year the programme runs. That proof is the verification infrastructure, and it is the programme's real product, whether or not the programme was ever designed to treat it that way.\n\n---\n\n*This perspective is part of Verdatir's ongoing research on verification infrastructure, programme governance, and the future of sustainability assurance.*", "image": "https://verdatir.com/_astro/verification-infrastructure-hero.B73qaCbc.jpg", "date_published": "2026-08-12T00:00:00.000Z", "date_modified": "2026-08-17T00:00:00.000Z", "authors": [ { "name": "Verdatir", "url": "https://verdatir.com/vision" } ], "tags": [ "Research", "Verification", "EPD", "CSRD", "Assurance", "Governance", "Programme Operators" ], "language": "en", "_verdatir": { "markdown": "https://verdatir.com/insights/verification-infrastructure-is-the-product.md" } }, { "id": "https://verdatir.com/insights/why-verified-lcas-still-arent-comparable", "url": "https://verdatir.com/insights/why-verified-lcas-still-arent-comparable", "title": "Why Verified LCAs Still Aren't Comparable", "summary": "Third-party verification confirms a life cycle assessment followed its own rules. It says nothing about whether two studies followed the same rules. Comparability needs shared Product Category Rules and a structured format to carry them.", "content_html": "A life cycle assessment (LCA) tallies a product’s environmental footprint from raw material extraction through disposal. Procurement teams, investors, and regulators expect to know by name, often before a supplier gets shortlisted or a claim gets reported. Third-party verification against ISO, the International Organization for Standardization, is supposed to settle whether that number can be trusted.
\nTwo ISO-compliant, third-party verified LCAs can still be incomparable. Verification checks whether a study followed its own stated rules consistently. It says nothing about whether two studies used the same rules as each other. A verified LCA is internally sound but whether it lines up with the LCA sitting next to it is a separate question entirely, and verification was never built to answer it.
\nOrganizations working through life cycle impact tend to pass through the same three stages. First, LCA is not on their radar, as the need hasn’t surfaced, which means no studies exist for the product. Then comes a data-collection bottleneck: gathering the material, energy, and process inputs which an LCA needs is slow, and getting suppliers to hand over usable data is even slower. Clearing this bottleneck used to be treated as the finish line. It isn’t. Once the LCAs exist, a third stage shows up, and it’s the one that actually stalls decisions: the studies are sitting there, produced in good faith, and they still can’t be compared to each other. A supplier scorecard, a portfolio-wide reduction target, a claim that one option beats another: all of them assume the underlying numbers were built to matching assumptions. That assumption is usually untested.
\nTwo studies of what looks like the same product can diverge from various variables: functional unit, system boundary, impact assessment method, dataset and its version, allocation rules, and geography or end-of-life assumptions. Move any one of those and the result shifts, independent of anything about the product itself. Neither study has to be wrong for the two numbers to be unusable side by side.
\nOne example: a systematic review of electric truck life cycle studies found cradle-to-gate carbon footprints ranging from 23 to 313 tonnes of CO2 equivalent, a variation driven entirely by which system boundary was applied, not by any difference between the trucks themselves. Same category of vehicle, honest work in every underlying study, and a range that spans more than tenfold.
\nThis reframes the useful question: while faced with two LCAs that disagree, the instinct is to ask which number is correct. However, the better question is whether both assessments were built under the same rules. If they weren’t, correctness isn’t even on the table yet. The two numbers are answers to different questions that happen to share a unit, and no amount of scrutinizing either one in isolation will make them line up. The scrutiny has to happen upstream, at the level of what each study assumed, not downstream at the level of the figure it produced.
\nProduct Category Rules, or PCRs, exist to close this gap. A PCR standardizes the assessment rules within a product category: it fixes the functional unit, the system boundary, and the allocation approach so two LCAs in that category finally answer one question instead of two adjacent ones. Where a PCR exists and both studies actually followed it, the comparability problem mostly disappears.
\nHowever, that is still just half the fix. The other half gets overlooked: format. A PCR governs how a study gets built. It says nothing about how the result gets transmitted to whoever has to use it next, and that handoff is usually where the context disappears. A raw number carries no methodology with it. A PDF buries the functional unit and boundary assumptions in an appendix nobody opens before the number gets copied into a spreadsheet. Both formats strip out exactly the information a reader would need to confirm the rules being actually followed, even when a perfectly good PCR was sitting behind the study the whole time.
\nA rulebook without a structured format to carry its outputs is unverifiable at the point of use. An Environmental Product Declaration, or EPD, is the standard document for reporting an LCA result, and a handful of structured, machine-readable formats now carry one, including ILCD+EPD, openEPD, and PACT/Pathfinder. Each carries the functional unit, boundary, and methodology alongside the figure itself, in a form which a system can read rather than a human having to dig for it. A procurement analyst, a supplier database, or another company’s own assessment tool can check that two numbers were built to matching rules before treating them as comparable. A bare number or a PDF can’t support that check, no matter how rigorous the underlying study was. The context needed to verify the rules is gone by the time the figure arrives at its destination, which means every recipient downstream is trusting the number on faith rather than confirming it.
\nCarbon numbers don’t carry the intuitive trust that price does, and the gap isn’t about rigor. Price earned that trust over centuries, through shared currencies, standardized receipts, and accounting rules built specifically so a number could move between parties and still mean the same thing on arrival. Environmental data has neither piece of that infrastructure yet: no consistent methodological alignment across categories, and no standardized way to disclose a result once it’s calculated. Both gaps are solvable. Neither has been solved yet, which is why swings of this size keep showing up under a different product name.
\nThe fix isn’t more audits stacked on top of the ones already in place. Think about a beer’s alcohol percentage, no third party re-verifies every bottle at the point of sale. It is trusted because of a measurement and disclosure system behind the label doing that work permanently, at the source. Carbon data needs an equivalent move: trust engineered into the format itself, not bolted on afterward through repeated certification of the same underlying number.
\nFixing only one half doesn’t solve the problem. It moves the failure downstream. A category with a solid PCR and no structured disclosure format still ships numbers that look incomparable, because nobody receiving the data can check that the rules behind it were followed. A structured format with no PCR behind it just moves the disagreement into a common container. The numbers become easy to compare and are still built on different assumptions, except now the mismatch is invisible instead of obvious.
\nClosing this gap from both sides means a few concrete shifts:
\nComparability fails without shared rules. Trust fails without a shared format to carry them. Fix one and the other failure is just waiting downstream.
", "content_text": "A life cycle assessment (LCA) tallies a product's environmental footprint from raw material extraction through disposal. Procurement teams, investors, and regulators expect to know by name, often before a supplier gets shortlisted or a claim gets reported. Third-party verification against ISO, the International Organization for Standardization, is supposed to settle whether that number can be trusted.\n\nTwo ISO-compliant, third-party verified LCAs can still be incomparable. Verification checks whether a study followed its own stated rules consistently. It says nothing about whether two studies used the same rules as each other. A verified LCA is internally sound but whether it lines up with the LCA sitting next to it is a separate question entirely, and verification was never built to answer it.\n\n## From Data Gap to Comparability Gap\n\nOrganizations working through life cycle impact tend to pass through the same three stages. First, LCA is not on their radar, as the need hasn't surfaced, which means no studies exist for the product. Then comes a data-collection bottleneck: gathering the material, energy, and process inputs which an LCA needs is slow, and getting suppliers to hand over usable data is even slower. Clearing this bottleneck used to be treated as the finish line. It isn't. Once the LCAs exist, a third stage shows up, and it's the one that actually stalls decisions: the studies are sitting there, produced in good faith, and they still can't be compared to each other. A supplier scorecard, a portfolio-wide reduction target, a claim that one option beats another: all of them assume the underlying numbers were built to matching assumptions. That assumption is usually untested.\n\n## Several Variables, One \"Same\" Product\n\nTwo studies of what looks like the same product can diverge from various variables: functional unit, system boundary, impact assessment method, dataset and its version, allocation rules, and geography or end-of-life assumptions. Move any one of those and the result shifts, independent of anything about the product itself. Neither study has to be wrong for the two numbers to be unusable side by side.\n\nOne example: a systematic review of electric truck life cycle studies found cradle-to-gate carbon footprints ranging from 23 to 313 tonnes of CO2 equivalent, a variation driven entirely by which system boundary was applied, not by any difference between the trucks themselves. Same category of vehicle, honest work in every underlying study, and a range that spans more than tenfold.\n\nThis reframes the useful question: while faced with two LCAs that disagree, the instinct is to ask which number is correct. However, the better question is whether both assessments were built under the same rules. If they weren't, correctness isn't even on the table yet. The two numbers are answers to different questions that happen to share a unit, and no amount of scrutinizing either one in isolation will make them line up. The scrutiny has to happen upstream, at the level of what each study assumed, not downstream at the level of the figure it produced.\n\n## Rules Solve Half the Problem\n\nProduct Category Rules, or PCRs, exist to close this gap. A PCR standardizes the assessment rules within a product category: it fixes the functional unit, the system boundary, and the allocation approach so two LCAs in that category finally answer one question instead of two adjacent ones. Where a PCR exists and both studies actually followed it, the comparability problem mostly disappears.\n\nHowever, that is still just half the fix. The other half gets overlooked: format. A PCR governs how a study gets built. It says nothing about how the result gets transmitted to whoever has to use it next, and that handoff is usually where the context disappears. A raw number carries no methodology with it. A PDF buries the functional unit and boundary assumptions in an appendix nobody opens before the number gets copied into a spreadsheet. Both formats strip out exactly the information a reader would need to confirm the rules being actually followed, even when a perfectly good PCR was sitting behind the study the whole time.\n\n## The Missing Format Layer\n\nA rulebook without a structured format to carry its outputs is unverifiable at the point of use. An Environmental Product Declaration, or EPD, is the standard document for reporting an LCA result, and a handful of structured, machine-readable formats now carry one, including ILCD+EPD, openEPD, and PACT/Pathfinder. Each carries the functional unit, boundary, and methodology alongside the figure itself, in a form which a system can read rather than a human having to dig for it. A procurement analyst, a supplier database, or another company's own assessment tool can check that two numbers were built to matching rules before treating them as comparable. A bare number or a PDF can't support that check, no matter how rigorous the underlying study was. The context needed to verify the rules is gone by the time the figure arrives at its destination, which means every recipient downstream is trusting the number on faith rather than confirming it.\n\n## What Carbon Data Is Still Missing\n\nCarbon numbers don't carry the intuitive trust that price does, and the gap isn't about rigor. Price earned that trust over centuries, through shared currencies, standardized receipts, and accounting rules built specifically so a number could move between parties and still mean the same thing on arrival. Environmental data has neither piece of that infrastructure yet: no consistent methodological alignment across categories, and no standardized way to disclose a result once it's calculated. Both gaps are solvable. Neither has been solved yet, which is why swings of this size keep showing up under a different product name.\n\nThe fix isn't more audits stacked on top of the ones already in place. Think about a beer's alcohol percentage, no third party re-verifies every bottle at the point of sale. It is trusted because of a measurement and disclosure system behind the label doing that work permanently, at the source. Carbon data needs an equivalent move: trust engineered into the format itself, not bolted on afterward through repeated certification of the same underlying number.\n\n## Fixing Both, Not One\n\nFixing only one half doesn't solve the problem. It moves the failure downstream. A category with a solid PCR and no structured disclosure format still ships numbers that look incomparable, because nobody receiving the data can check that the rules behind it were followed. A structured format with no PCR behind it just moves the disagreement into a common container. The numbers become easy to compare and are still built on different assumptions, except now the mismatch is invisible instead of obvious.\n\nClosing this gap from both sides means a few concrete shifts:\n\n- Treating LCA figures as scenario-based results rather than absolute measurements.\n- Developing PCRs for the product categories that carry the most weight in a portfolio.\n- Demanding supplier data in structured exchange formats instead of PDFs or bare numbers.\n- Building the internal capability to preserve methodological context as data changes hands, rather than letting it evaporate at each handoff.\n- Training the people who are making purchasing and reporting decisions to ask about, functional unit and system boundary, before they ask about the number itself.\n\nComparability fails without shared rules. Trust fails without a shared format to carry them. Fix one and the other failure is just waiting downstream.", "image": "https://verdatir.com/_astro/lca-comparability-hero.e_-jIp15.jpg", "date_published": "2026-08-04T00:00:00.000Z", "date_modified": "2026-08-17T00:00:00.000Z", "authors": [ { "name": "Verdatir", "url": "https://verdatir.com/vision" } ], "tags": [ "Insights", "LCA", "PCR", "EPD", "Comparability", "Data Standards" ], "language": "en", "_verdatir": { "markdown": "https://verdatir.com/insights/why-verified-lcas-still-arent-comparable.md" } }, { "id": "https://verdatir.com/insights/cbam-2026-price-and-playbook", "url": "https://verdatir.com/insights/cbam-2026-price-and-playbook", "title": "A Price and a Playbook: What EU's Latest CBAM Related Moves Mean", "summary": "CBAM's definitive phase is live, the certificate price is set, and two open negotiations in Brussels will decide what importers actually pay. A practical read on prices, deadlines, exemptions, and scope.", "content_html": "The Carbon Border Adjustment Mechanism, or CBAM, charges importers of carbon-intensive goods the same carbon price EU producers already pay under the EU Emissions Trading System (EU ETS). It covers six sectors: cement, iron and steel, aluminium, fertilizers, electricity, and hydrogen. Since January 1, 2026, when CBAM’s definitive phase began, that price has stopped being theoretical.
\nOn July 6, 2026, the European Commission set the CBAM certificate price for the second quarter of 2026 at EUR 75.28 per tonne of embedded CO2, the volume-weighted average of EU ETS auction prices for April through June. That’s down eight cents from the EUR 75.36 set for the first quarter. Four such quarterly prices apply across 2026. From January 1, 2027, the Commission switches to publishing a price every Monday instead, tracking the EU ETS market in something close to real time.
\nThat date isn’t the same as when certificates go on sale though. Purchases open a month later, on February 1, 2027, and cover all of 2026’s imports retroactively. Nobody pays for 2026 emissions until then, which gives importers most of a year to model their exposure before a euro changes hands.
\nBuying a certificate and staying compliant run through two different pieces of infrastructure. The CBAM Registry is where declarants submit their annual declarations and where certificates are issued, held, surrendered, and cancelled. Payment is processed separately, through a Common Central Platform the Commission built to handle the financial side. Declarants can buy certificates in any quantity, as often as they like, throughout the year: there’s no cap and no single annual purchase window to wait for.
\nThat flexibility has a limit. From 2027, declarants must hold certificates covering at least 50% of their year-to-date embedded emissions at the end of each quarter, a threshold the Commission cut from 80% to ease cash flow. Imports are tracked on the calendar year, January 1 through December 31, and every holding requirement resets on that clock.
\nThree dates govern a certificate’s life, and none of them fall within the year the certificates cover: both the surrender and repurchase happens in the subsequent year, whilst the expiry of certificates happens 2 years after.
\nSeptember 30, annually: the deadline to submit the annual declaration and surrender enough certificates to cover the full prior year’s emissions. Miss it, and the penalty is EUR 100 per tonne of the shortfall.
\nOctober 31, annually: the deadline to request repurchase of any surplus. The request can only be filed once a year, cannot be edited or withdrawn after submission, and is repaid at exactly the price the certificate was bought for.
\nNovember 1, annually: whatever is left uncancelled and unsurrendered from certificates bought two calendar years earlier disappears, with no compensation.
\nNone of this can be worked around by trading. CBAM certificates are non-tradable and non-transferable, so the one-shot repurchase window above is the only way to unwind a surplus.
\nImporters whose combined imports of cement, iron and steel, aluminium, and fertilizers stay at or below 50 tonnes of net mass in a calendar year are exempt from CBAM altogether: no registration, no declaration, no certificate surrender. Electricity and hydrogen carry no such exemption.
\nCBAM’s actual bite depends on two negotiations still running in Brussels, and they’re not the same fight.
\nThe first is timing. CBAM is meant to replace the free EU ETS allowances domestic producers currently receive, phasing them out between 2026 and 2034 as CBAM’s own cost factor climbs from 2.5% to 100%. Germany and Austria are both pushing to stretch that phase-out past 2034, which would leave domestic producers with free allowances for longer and push back the point at which CBAM’s full cost lands on importers.
\nThe second is scope. CBAM currently taxes raw materials, not the finished goods made from them, and the Commission’s own reading is that this leaves a downstream leak: a fridge or an engine built from steel carries no carbon charge at all today. The Commission’s December 2025 proposal would add 180 downstream products to close that gap. The Council’s position, adopted in June 2026, pushes that closer to 380. Parliament’s environment committee voted in July for more than 400. That fight runs through the ordinary legislative process, with Parliament and Council negotiating toward a final number. It has nothing to do with the separate delegated act that governs how certificates are bought and sold, whose public consultation closed as of August 6 and now awaits adoption. The two are easy to conflate; only one of them changes what gets taxed.
\nBoth fights are still open as of August 2026. Either outcome will matter more to an importer’s actual bill than any single quarterly price print.
", "content_text": "The Carbon Border Adjustment Mechanism, or CBAM, charges importers of carbon-intensive goods the same carbon price EU producers already pay under the EU Emissions Trading System (EU ETS). It covers six sectors: cement, iron and steel, aluminium, fertilizers, electricity, and hydrogen. Since January 1, 2026, when CBAM's definitive phase began, that price has stopped being theoretical.\n\nOn July 6, 2026, the European Commission set the CBAM certificate price for the second quarter of 2026 at EUR 75.28 per tonne of embedded CO2, the volume-weighted average of EU ETS auction prices for April through June. That's down eight cents from the EUR 75.36 set for the first quarter. Four such quarterly prices apply across 2026. From January 1, 2027, the Commission switches to publishing a price every Monday instead, tracking the EU ETS market in something close to real time.\n\nThat date isn't the same as when certificates go on sale though. Purchases open a month later, on February 1, 2027, and cover all of 2026's imports retroactively. Nobody pays for 2026 emissions until then, which gives importers most of a year to model their exposure before a euro changes hands.\n\n## Two systems, one compliance trail\n\nBuying a certificate and staying compliant run through two different pieces of infrastructure. The CBAM Registry is where declarants submit their annual declarations and where certificates are issued, held, surrendered, and cancelled. Payment is processed separately, through a Common Central Platform the Commission built to handle the financial side. Declarants can buy certificates in any quantity, as often as they like, throughout the year: there's no cap and no single annual purchase window to wait for.\n\nThat flexibility has a limit. From 2027, declarants must hold certificates covering at least 50% of their year-to-date embedded emissions at the end of each quarter, a threshold the Commission cut from 80% to ease cash flow. Imports are tracked on the calendar year, January 1 through December 31, and every holding requirement resets on that clock.\n\n## The deadlines that actually bind\n\nThree dates govern a certificate's life, and none of them fall within the year the certificates cover: both the surrender and repurchase happens in the subsequent year, whilst the expiry of certificates happens 2 years after.\n\n**September 30, annually:** the deadline to submit the annual declaration and surrender enough certificates to cover the full prior year's emissions. Miss it, and the penalty is EUR 100 per tonne of the shortfall.\n\n**October 31, annually:** the deadline to request repurchase of any surplus. The request can only be filed once a year, cannot be edited or withdrawn after submission, and is repaid at exactly the price the certificate was bought for.\n\n**November 1, annually:** whatever is left uncancelled and unsurrendered from certificates bought two calendar years earlier disappears, with no compensation.\n\nNone of this can be worked around by trading. CBAM certificates are non-tradable and non-transferable, so the one-shot repurchase window above is the only way to unwind a surplus.\n\n## Who's exempt\n\nImporters whose combined imports of cement, iron and steel, aluminium, and fertilizers stay at or below 50 tonnes of net mass in a calendar year are exempt from CBAM altogether: no registration, no declaration, no certificate surrender. Electricity and hydrogen carry no such exemption.\n\n## The two fights that will set the real price\n\nCBAM's actual bite depends on two negotiations still running in Brussels, and they're not the same fight.\n\nThe first is timing. CBAM is meant to replace the free EU ETS allowances domestic producers currently receive, phasing them out between 2026 and 2034 as CBAM's own cost factor climbs from 2.5% to 100%. Germany and Austria are both pushing to stretch that phase-out past 2034, which would leave domestic producers with free allowances for longer and push back the point at which CBAM's full cost lands on importers.\n\nThe second is scope. CBAM currently taxes raw materials, not the finished goods made from them, and the Commission's own reading is that this leaves a downstream leak: a fridge or an engine built from steel carries no carbon charge at all today. The Commission's December 2025 proposal would add 180 downstream products to close that gap. The Council's position, adopted in June 2026, pushes that closer to 380. Parliament's environment committee voted in July for more than 400. That fight runs through the ordinary legislative process, with Parliament and Council negotiating toward a final number. It has nothing to do with the separate delegated act that governs how certificates are bought and sold, whose public consultation closed as of August 6 and now awaits adoption. The two are easy to conflate; only one of them changes what gets taxed.\n\nBoth fights are still open as of August 2026. Either outcome will matter more to an importer's actual bill than any single quarterly price print.", "image": "https://verdatir.com/_astro/cbam-2026-hero.DalvEkjO.jpg", "date_published": "2026-07-24T00:00:00.000Z", "date_modified": "2026-08-17T00:00:00.000Z", "authors": [ { "name": "Verdatir Research", "url": "https://verdatir.com/vision" } ], "tags": [ "Research", "CBAM", "EU ETS", "Carbon Pricing", "Regulation", "Compliance" ], "language": "en", "_verdatir": { "markdown": "https://verdatir.com/insights/cbam-2026-price-and-playbook.md" } }, { "id": "https://verdatir.com/insights/iso-14025-2026-epd-ecosystem-maturity", "url": "https://verdatir.com/insights/iso-14025-2026-epd-ecosystem-maturity", "title": "ISO 14025:2026 reflects the growing maturity of the global EPD ecosystem", "summary": "On 24 June 2026 ISO published the biggest update to the EPD standard since 2006. The six changes look unrelated but form one sequence: tightened terminology, a harder boundary against unverified claims, mandatory harmonisation, verified digital tools, formalised prospective EPDs, and a governance regime built for a regulated market.", "content_html": "When ISO 14025 first came out in 2006, Environmental Product Declarations (EPDs) were a fairly niche way to report a product’s environmental footprint. Since then, they’ve become central to different use cases such as green building certification, sustainable procurement, carbon reporting, and product transparency. As the ecosystem grew, so did the pressure for a clearer structure behind it.
\nOn 24 June 2026, ISO published the biggest update to the EPD standard since the original 2006 edition. Listed out, the six changes look unrelated. They aren’t: each one sets up the building block for the next.
\nFor twenty years the ISO 14020 family split environmental communication into Type I (ecolabels), Type II (self-declared claims), and Type III (LCA-based declarations). Almost nobody in practice asked for “a Type III declaration for this insulation panel.” They asked for an EPD. ISO 14025:2026 catches up to that usage: an EPD is now defined directly as an independently verified, LCA-based environmental statement, and the Type I/II/III language is dropped from the standard altogether. “Type III environmental declaration,” “environmental label,” and “environmental declaration” are formally deprecated in favor of this one unified term, closing a gap where practitioners already said “EPD” colloquially while the formal documents still said “Type III”, a mismatch that caused real confusion in tenders and green building credit submissions. It also shuts down a specific abuse pattern the old terminology quietly enabled: manufacturers marketing self-declared claims as fictitious “Type II EPDs,” borrowing the credibility of the EPD name for a declaration that never went through independent verification.
\nThis kind of cleanup isn’t unique to this standard. ISO 14001:2026, published earlier this year, made a similar move elsewhere in its text, retiring older inherited jargon once a standard has matured past the point where it still adds clarity.
\nThe same push for precision runs through the rest of the standard’s vocabulary. ISO 14025:2026 introduces or tightens definitions for general programme instructions (GPI), EPD tool, a formally defined declared unit, and environmental performance, and it replaces the vague 2006 term “predetermined parameters” with the more specific “predetermined impact/inventory indicators.” A full chain-of-custody vocabulary, covering mass balance, credit method, and book-and-claim, is imported from ISO 22095, signaling that programme operators are now expected to explicitly address how mixed or attributed data is handled rather than leaving it implicit.
\nNaming these terms clearly only helps if the standard’s own place in the ISO family is just as clear. ISO 14025 is no longer standalone: its core principles have moved up into the revised ISO 14020, making 14025 a specialization within a broader family that also pulls in ISO/TS 14027, which governs how PCRs are developed, and ISO/TS 14029, which governs how operators recognise each other’s work, as normative references rather than optional guidance. Requirements are now split cleanly into general requirements inherited from ISO 14020 versus EPD-programme-specific requirements, which makes the standard easier to keep synchronised as the broader 14020 umbrella evolves. That structural realignment is what makes the next change possible. Once EPD sits inside a defined family with a defined scope, the standard can draw a much harder line around what falls outside that scope entirely.
\nISO 14025:2026 states directly that self-declared claims under ISO 14021 and ecolabels under ISO 14024 don’t count as EPDs, or as a type of EPD. Once “EPD” carries more weight as a trusted term, the standard has to be equally clear about what sits outside it.
\nThat line matters because the market it’s protecting is genuinely messy. The European Commission’s own analysis found that over half of the green claims it examined were vague, misleading, or unfounded, and identified around 230 different sustainability labels circulating across the EU with wildly inconsistent levels of transparency. ISO 14025:2026 is drawing a harder line against that confusion from the standards side.
\nThe EU is arriving at a similar point from the regulatory side, on its own separate timeline. The proposed Green Claims Directive is a messier story on its own. In June 2025 the European Commission announced that it intended to withdraw the proposal, after pushback over how far it would reach. Yet it still hasn’t taken the formal step, so the proposal sits stuck in between: not dead, but not moving forward either.
\nA different law, already in force, is closing much of the same gap on its own. The Directive on Empowering Consumers for the Green Transition (EmpCo), in force since March 2024 and applying to businesses from 27 September 2026, only allows a sustainability label if it’s established by a public authority or backed by an independent, third-party certification scheme. Companies inventing their own trust marks won’t be able to use them after that date.
\nTwo different processes, one technical standard and one binding consumer law, are converging on the same idea at roughly the same time: unverified environmental claims are losing legitimacy, and who verifies matters more than the label itself. For companies that already hold an EPD, that’s an advantage rather than a new compliance cost. The verification was already built in.
\nA clean definition doesn’t solve a separate problem: roughly fifteen to twenty EPD programme operators are active within Europe alone (IBU in Germany, EPD International in Sweden, EPD Norge, EPD Danmark, GlobalEPD in Spain, and others), each running its own General Programme Instructions and PCR library. Most build on the same core standard, EN 15804, so PCRs from different operators end up containing almost identical methodology anyway. In practice, a manufacturer selling across three or four European markets can still end up navigating three or four separate registries, verification processes, and slightly different interpretations of the same rules.
\nLeft alone, this fragmentation tends to widen rather than shrink. France’s operator, INIES, discontinued its membership in the cross-programme harmonisation body ECO Platform at the end of 2024 and stopped acting as an ECO Platform operator from 1 January 2025, choosing to run independently instead.
\nThe European Commission already tried solving a version of this problem from a different angle: a single centralised methodology, the Product Environmental Footprint (PEF), with Category Rules approved directly by the Commission. The idea was that one methodology, defined centrally, would remove fragmentation by design. The results have been mixed. A peer-reviewed comparison of PEF and EN 15804 EPDs for construction products found the two methods have different enough requirements that their results can’t really be compared or used interchangeably. Centralising the rulebook didn’t remove fragmentation. It just moved it from many operators down to two competing systems.
\nISO 14025:2026 is making a different bet. Rather than centralising authority in one body, it keeps programme operators independent but makes ISO/TS 14027 and ISO/TS 14029 normative references instead of optional guidance, backed by a new informative annex addressing PCR and programme harmonisation for comparability, tied directly to both technical specifications. Mutual recognition arrangements (MRAs) are woven into operator requirements throughout, rather than left as a voluntary add-on, which moves comparability from an aspirational goal into an actual process with defined mechanisms. IBU and EPD Norge have had a signed, working bilateral mutual recognition agreement on this basis since 2014, a reasonable proof point that convergence doesn’t need a single owner, just a shared and enforced process for getting there.
\nThe revision also tackles a subtler version of the comparability problem: two EPDs for the same product category can carry the “EPD” name while resting on very different amounts and quality of underlying data. Programmes must now classify each EPD by criteria such as the number of products or facilities covered, the percentage of GWP from supply-chain-specific data versus generic background data, the life cycle stages included, and the months of primary data used, and this classification has to appear on the declaration itself. It’s a shift toward comparability with honesty rather than comparability by assumption. It distinguishes a single-facility EPD built on a full year of primary data from an industry-average EPD leaning heavily on generic background data, without banning either one. This convergence has to reach the tools generating the data too, not just the paperwork.
\nHarmonising PCRs on paper only goes so far if the software producing the underlying LCA data isn’t held to the same bar. ISO 14025:2026 introduces new terminology and requirements for EPD tools, including a requirement that the tools themselves go through independent verification, not just the EPD output they generate. For the first time, “EPD tool” is a formally defined term, and programme operators are required to set explicit verification requirements for tools rather than only reviewing what a tool produces. A verified tool can unlock streamlined verification for the individual EPDs it generates, rewarding a single rigorous engine check that then applies to every output the tool produces afterward, a meaningful efficiency gain for manufacturers running many similar products through the same software. That efficiency is deliberately paired with an independence rule: automation is endorsed, but only if the tool’s verifier isn’t the vendor itself, a caveat that connects directly to the broader independence requirements described later in this revision.
\nThat shift lands right as a much bigger digital requirement arrives from EU regulation. Under the Ecodesign for Sustainable Products Regulation and the revised Construction Products Regulation, the EU’s Digital Product Passport won’t replace EPDs, but is expected to reference and embed EPD data directly in structured, machine-readable form. The Battery Passport is first, mandatory from February 2027, with construction and other product groups following on a phased schedule after that.
\nThe sequence makes sense together: harmonised PCRs mean the same product category gets assessed the same way everywhere, and verified tools mean the resulting data can be trusted enough to flow automatically into a legally mandated passport, instead of sitting in a PDF that only a person ever opens.
\nThese changes ensure the consistency within the processes and tool providers; however, there is no specific emphasis on LLMs and non-deterministic generation of the datapoints. As a result the standardisation in processing might not solve the complete inconsistency and data quality issue, but is a step in the right direction.
\nOnce the definition is precise, the boundary is clear, the methodology is converging, and the tools are verified, ISO 14025:2026 can do something it couldn’t have done casually back in 2006: let EPDs be published before a full year of production data exists. This revision formally introduces requirements for prospective EPDs, built from data on a comparable existing EPD, inventory data for similar technologies already on the market, or forecast and design data from a manufacturing plant. Programme operators have already been offering informal versions of this for years. What changes is that it’s now written into the international standard with defined requirements, including explicit guardrails: limited validity periods, mandatory annual re-evaluation against actual data as it becomes available, clear labeling so a prospective EPD is never mistaken for a fully verified one, and formal quantification of the uncertainty involved. That matters for companies that need an EPD to unlock financing or bid on a tender before they’ve completed a full reference year of data.
\nThere’s a useful comparison in carbon markets, which ran into the same trade-off earlier. Project-based carbon crediting distinguishes ex-ante crediting, issued against forecast future reductions before a project is fully operational, from ex-post crediting, issued once measured performance data exists. Ex-ante credits unlock financing for projects that couldn’t get built otherwise; ex-post credits are the ones markets treat as fully robust. Prospective EPDs sit in the same space: deliberately less rigorous, clearly labelled as such, and useful precisely because they let innovation get financed before the historical data catches up.
\nThe open risk is the same one carbon markets have already dealt with: “prospective” quietly becoming the default instead of the exception. Whether programme operators keep that distinction sharp in their own General Programme Instructions is the real test, and the annual re-evaluation requirement is the standard’s main lever for keeping that discipline in place rather than leaving it to goodwill.
\nEverything above, the tighter definition, the harder boundary, the harmonisation push, the verified tools, the disciplined use of prospective data, depends on one thing: that verification itself can be trusted. ISO 14025:2026 addresses that directly, with stronger emphasis on independence, impartiality, and conflict-of-interest management throughout EPD programme processes. This is the biggest structural tightening in the revision. “Third party” is now explicitly defined as independent of the manufacturer, the programme operator, the LCA practitioner, and, notably, the EPD tool developer. Independence is also folded into the definition of “verifier” itself, upgraded from someone who merely carries out verification to a party who is explicitly competent and independent. This directly targets arrangements where verifiers were trained, approved, or supplied by the same entity whose tool or scheme they were checking, closing a loophole the old definition left open.
\nThat same logic extends to how reviews are organised. PCR review and EPD verification must now run as two entirely separate processes, the PCR review panel must have at least a chair plus two members, and programme operators must implement formal policies to identify, disclose, and manage conflicts of interest across operators, verifiers, and panel members. This closes loopholes where in-house reviewers were still labeled “third party” despite having ties to the scheme or tool being assessed.
\nThese verification rules sit inside a much broader governance overhaul for programme operators. The 2026 edition converts a short list of operator duties into an extensive governance checklist: disclosing ownership, governance, and revenue structures, publishing GPIs and public digital registries of PCRs and EPDs, running transparent PCR development with voting and public consultation, maintaining complaints procedures that cover already-published EPDs, adopting anti-trust policies, and actively participating in mutual recognition arrangements. Read together, this looks less like a simple labeling standard and more like a governance regime built for a regulated market, likely anticipating tightening EU and US green claims regulation rather than waiting for it. One piece of this tightening takes effect with no phase-in at all: for consumer-facing EPDs specifically, all verification must be third-party and the verifier must be named in the declaration. It’s the most immediate, hard deadline embedded in the revision, with far less room for gradual transition than the operator-level governance changes.
\nThis isn’t happening to EPDs alone. ISO 14001:2026 separately strengthened internal audit rigour and extended accountability across the value chain. ISO has also published a dedicated horizontal standard for this exact problem, the new ISO 14019 series, which sets out requirements for the competence and impartiality of bodies that validate or verify sustainability information, and distinguishes validation of forward-looking assumptions (relevant to the prospective EPDs above) from verification of historical, declared data.
\nEPDs aren’t being singled out here. They’re catching up to a governance bar ISO is raising across its sustainability standards at the same time, which is part of why the rest of this revision holds together. Because ISO 14025 remains a voluntary standard with no automatic enforcement mechanism, none of this lands on day one across the board. A staggered rollout is the realistic expectation: GPI and terminology updates likely come first, verifier approval and tool verification procedure changes follow, and the unwinding of conflict-of-interest arrangements the new “third party” definition no longer permits comes last, simply because those arrangements take longer to dismantle than they took to build.
\nNone of this changes how an LCA itself is done. ISO 14040 and ISO 14044 remain the methodological foundation, untouched by this revision. What ISO 14025:2026 changes is everything around it: what counts as an EPD, what doesn’t, who has to agree on the rules, whose tools are trusted to produce the data, how early that data can responsibly be shared, and who’s accountable for checking all of it.
\nPut the six pieces back together and they read as one sequence rather than six separate updates. Fixing the term first (1) is what makes it worth defending with a hard boundary (2). A defended term is what makes harmonised methodology across a dozen-plus operators (3) worth pursuing instead of ignoring. Harmonised methodology is what makes it safe to trust software that automates part of the process (4). Trusted tools and converging methodology are what make it defensible to publish an EPD before a full year of data exists (5). And every one of those five steps only holds up if the verification underneath all of it is genuinely independent (6), which is why governance is the load-bearing change rather than the last item on the list.
\nThe practical read for anyone holding or issuing EPDs: the rule on named, third-party verification for consumer-facing declarations is the one item that needs attention now, since it carries no phase-in window at all. Everything else, terminology updates, tool verification procedures, the unwinding of conflicted review arrangements, will phase in on the operators’ own schedules rather than all at once.
\nThe next phase of EPDs will be defined by clarity, consistency, and credibility. That’s where real progress begins.
", "content_text": "When ISO 14025 first came out in 2006, Environmental Product Declarations (EPDs) were a fairly niche way to report a product's environmental footprint. Since then, they've become central to different use cases such as green building certification, sustainable procurement, carbon reporting, and product transparency. As the ecosystem grew, so did the pressure for a clearer structure behind it.\n\nOn 24 June 2026, ISO published the biggest update to the EPD standard since the original 2006 edition. Listed out, the six changes look unrelated. They aren't: each one sets up the building block for the next.\n\n## 1. Terminology is comprehensively updated\n\nFor twenty years the ISO 14020 family split environmental communication into Type I (ecolabels), Type II (self-declared claims), and Type III (LCA-based declarations). Almost nobody in practice asked for \"a Type III declaration for this insulation panel.\" They asked for an EPD. ISO 14025:2026 catches up to that usage: an EPD is now defined directly as an independently verified, LCA-based environmental statement, and the Type I/II/III language is dropped from the standard altogether. \"Type III environmental declaration,\" \"environmental label,\" and \"environmental declaration\" are formally deprecated in favor of this one unified term, closing a gap where practitioners already said \"EPD\" colloquially while the formal documents still said \"Type III\", a mismatch that caused real confusion in tenders and green building credit submissions. It also shuts down a specific abuse pattern the old terminology quietly enabled: manufacturers marketing self-declared claims as fictitious \"Type II EPDs,\" borrowing the credibility of the EPD name for a declaration that never went through independent verification.\n\nThis kind of cleanup isn't unique to this standard. ISO 14001:2026, published earlier this year, made a similar move elsewhere in its text, retiring older inherited jargon once a standard has matured past the point where it still adds clarity.\n\nThe same push for precision runs through the rest of the standard's vocabulary. ISO 14025:2026 introduces or tightens definitions for general programme instructions (GPI), EPD tool, a formally defined declared unit, and environmental performance, and it replaces the vague 2006 term \"predetermined parameters\" with the more specific \"predetermined impact/inventory indicators.\" A full chain-of-custody vocabulary, covering mass balance, credit method, and book-and-claim, is imported from ISO 22095, signaling that programme operators are now expected to explicitly address how mixed or attributed data is handled rather than leaving it implicit.\n\nNaming these terms clearly only helps if the standard's own place in the ISO family is just as clear. ISO 14025 is no longer standalone: its core principles have moved up into the revised ISO 14020, making 14025 a specialization within a broader family that also pulls in ISO/TS 14027, which governs how PCRs are developed, and ISO/TS 14029, which governs how operators recognise each other's work, as normative references rather than optional guidance. Requirements are now split cleanly into general requirements inherited from ISO 14020 versus EPD-programme-specific requirements, which makes the standard easier to keep synchronised as the broader 14020 umbrella evolves. That structural realignment is what makes the next change possible. Once EPD sits inside a defined family with a defined scope, the standard can draw a much harder line around what falls outside that scope entirely.\n\n## 2. The boundary around the terminology got stricter\n\nISO 14025:2026 states directly that self-declared claims under ISO 14021 and ecolabels under ISO 14024 don't count as EPDs, or as a type of EPD. Once \"EPD\" carries more weight as a trusted term, the standard has to be equally clear about what sits outside it.\n\nThat line matters because the market it's protecting is genuinely messy. The European Commission's own analysis found that over half of the green claims it examined were vague, misleading, or unfounded, and identified around 230 different sustainability labels circulating across the EU with wildly inconsistent levels of transparency. ISO 14025:2026 is drawing a harder line against that confusion from the standards side.\n\nThe EU is arriving at a similar point from the regulatory side, on its own separate timeline. The proposed Green Claims Directive is a messier story on its own. In June 2025 the European Commission announced that it intended to withdraw the proposal, after pushback over how far it would reach. Yet it still hasn't taken the formal step, so the proposal sits stuck in between: not dead, but not moving forward either.\n\nA different law, already in force, is closing much of the same gap on its own. The Directive on Empowering Consumers for the Green Transition (EmpCo), in force since March 2024 and applying to businesses from 27 September 2026, only allows a sustainability label if it's established by a public authority or backed by an independent, third-party certification scheme. Companies inventing their own trust marks won't be able to use them after that date.\n\nTwo different processes, one technical standard and one binding consumer law, are converging on the same idea at roughly the same time: unverified environmental claims are losing legitimacy, and who verifies matters more than the label itself. For companies that already hold an EPD, that's an advantage rather than a new compliance cost. The verification was already built in.\n\n## 3. Harmonisation becomes non-negotiable\n\nA clean definition doesn't solve a separate problem: roughly fifteen to twenty EPD programme operators are active within Europe alone (IBU in Germany, EPD International in Sweden, EPD Norge, EPD Danmark, GlobalEPD in Spain, and others), each running its own General Programme Instructions and PCR library. Most build on the same core standard, EN 15804, so PCRs from different operators end up containing almost identical methodology anyway. In practice, a manufacturer selling across three or four European markets can still end up navigating three or four separate registries, verification processes, and slightly different interpretations of the same rules.\n\nLeft alone, this fragmentation tends to widen rather than shrink. France's operator, INIES, discontinued its membership in the cross-programme harmonisation body ECO Platform at the end of 2024 and stopped acting as an ECO Platform operator from 1 January 2025, choosing to run independently instead.\n\nThe European Commission already tried solving a version of this problem from a different angle: a single centralised methodology, the Product Environmental Footprint (PEF), with Category Rules approved directly by the Commission. The idea was that one methodology, defined centrally, would remove fragmentation by design. The results have been mixed. A peer-reviewed comparison of PEF and EN 15804 EPDs for construction products found the two methods have different enough requirements that their results can't really be compared or used interchangeably. Centralising the rulebook didn't remove fragmentation. It just moved it from many operators down to two competing systems.\n\nISO 14025:2026 is making a different bet. Rather than centralising authority in one body, it keeps programme operators independent but makes ISO/TS 14027 and ISO/TS 14029 normative references instead of optional guidance, backed by a new informative annex addressing PCR and programme harmonisation for comparability, tied directly to both technical specifications. Mutual recognition arrangements (MRAs) are woven into operator requirements throughout, rather than left as a voluntary add-on, which moves comparability from an aspirational goal into an actual process with defined mechanisms. IBU and EPD Norge have had a signed, working bilateral mutual recognition agreement on this basis since 2014, a reasonable proof point that convergence doesn't need a single owner, just a shared and enforced process for getting there.\n\nThe revision also tackles a subtler version of the comparability problem: two EPDs for the same product category can carry the \"EPD\" name while resting on very different amounts and quality of underlying data. Programmes must now classify each EPD by criteria such as the number of products or facilities covered, the percentage of GWP from supply-chain-specific data versus generic background data, the life cycle stages included, and the months of primary data used, and this classification has to appear on the declaration itself. It's a shift toward comparability with honesty rather than comparability by assumption. It distinguishes a single-facility EPD built on a full year of primary data from an industry-average EPD leaning heavily on generic background data, without banning either one. This convergence has to reach the tools generating the data too, not just the paperwork.\n\n## 4. Digital EPD tools got officially recognised\n\nHarmonising PCRs on paper only goes so far if the software producing the underlying LCA data isn't held to the same bar. ISO 14025:2026 introduces new terminology and requirements for EPD tools, including a requirement that the tools themselves go through independent verification, not just the EPD output they generate. For the first time, \"EPD tool\" is a formally defined term, and programme operators are required to set explicit verification requirements for tools rather than only reviewing what a tool produces. A verified tool can unlock streamlined verification for the individual EPDs it generates, rewarding a single rigorous engine check that then applies to every output the tool produces afterward, a meaningful efficiency gain for manufacturers running many similar products through the same software. That efficiency is deliberately paired with an independence rule: automation is endorsed, but only if the tool's verifier isn't the vendor itself, a caveat that connects directly to the broader independence requirements described later in this revision.\n\nThat shift lands right as a much bigger digital requirement arrives from EU regulation. Under the Ecodesign for Sustainable Products Regulation and the revised Construction Products Regulation, the EU's Digital Product Passport won't replace EPDs, but is expected to reference and embed EPD data directly in structured, machine-readable form. The Battery Passport is first, mandatory from February 2027, with construction and other product groups following on a phased schedule after that.\n\nThe sequence makes sense together: harmonised PCRs mean the same product category gets assessed the same way everywhere, and verified tools mean the resulting data can be trusted enough to flow automatically into a legally mandated passport, instead of sitting in a PDF that only a person ever opens.\n\nThese changes ensure the consistency within the processes and tool providers; however, there is no specific emphasis on LLMs and non-deterministic generation of the datapoints. As a result the standardisation in processing might not solve the complete inconsistency and data quality issue, but is a step in the right direction.\n\n## 5. \"Prospective EPDs\" defined as a part of the standard\n\nOnce the definition is precise, the boundary is clear, the methodology is converging, and the tools are verified, ISO 14025:2026 can do something it couldn't have done casually back in 2006: let EPDs be published before a full year of production data exists. This revision formally introduces requirements for prospective EPDs, built from data on a comparable existing EPD, inventory data for similar technologies already on the market, or forecast and design data from a manufacturing plant. Programme operators have already been offering informal versions of this for years. What changes is that it's now written into the international standard with defined requirements, including explicit guardrails: limited validity periods, mandatory annual re-evaluation against actual data as it becomes available, clear labeling so a prospective EPD is never mistaken for a fully verified one, and formal quantification of the uncertainty involved. That matters for companies that need an EPD to unlock financing or bid on a tender before they've completed a full reference year of data.\n\nThere's a useful comparison in carbon markets, which ran into the same trade-off earlier. Project-based carbon crediting distinguishes ex-ante crediting, issued against forecast future reductions before a project is fully operational, from ex-post crediting, issued once measured performance data exists. Ex-ante credits unlock financing for projects that couldn't get built otherwise; ex-post credits are the ones markets treat as fully robust. Prospective EPDs sit in the same space: deliberately less rigorous, clearly labelled as such, and useful precisely because they let innovation get financed before the historical data catches up.\n\nThe open risk is the same one carbon markets have already dealt with: \"prospective\" quietly becoming the default instead of the exception. Whether programme operators keep that distinction sharp in their own General Programme Instructions is the real test, and the annual re-evaluation requirement is the standard's main lever for keeping that discipline in place rather than leaving it to goodwill.\n\n## 6. Governance and verification requirements get spelled out in more detail\n\nEverything above, the tighter definition, the harder boundary, the harmonisation push, the verified tools, the disciplined use of prospective data, depends on one thing: that verification itself can be trusted. ISO 14025:2026 addresses that directly, with stronger emphasis on independence, impartiality, and conflict-of-interest management throughout EPD programme processes. This is the biggest structural tightening in the revision. \"Third party\" is now explicitly defined as independent of the manufacturer, the programme operator, the LCA practitioner, and, notably, the EPD tool developer. Independence is also folded into the definition of \"verifier\" itself, upgraded from someone who merely carries out verification to a party who is explicitly competent and independent. This directly targets arrangements where verifiers were trained, approved, or supplied by the same entity whose tool or scheme they were checking, closing a loophole the old definition left open.\n\nThat same logic extends to how reviews are organised. PCR review and EPD verification must now run as two entirely separate processes, the PCR review panel must have at least a chair plus two members, and programme operators must implement formal policies to identify, disclose, and manage conflicts of interest across operators, verifiers, and panel members. This closes loopholes where in-house reviewers were still labeled \"third party\" despite having ties to the scheme or tool being assessed.\n\nThese verification rules sit inside a much broader governance overhaul for programme operators. The 2026 edition converts a short list of operator duties into an extensive governance checklist: disclosing ownership, governance, and revenue structures, publishing GPIs and public digital registries of PCRs and EPDs, running transparent PCR development with voting and public consultation, maintaining complaints procedures that cover already-published EPDs, adopting anti-trust policies, and actively participating in mutual recognition arrangements. Read together, this looks less like a simple labeling standard and more like a governance regime built for a regulated market, likely anticipating tightening EU and US green claims regulation rather than waiting for it. One piece of this tightening takes effect with no phase-in at all: for consumer-facing EPDs specifically, all verification must be third-party and the verifier must be named in the declaration. It's the most immediate, hard deadline embedded in the revision, with far less room for gradual transition than the operator-level governance changes.\n\nThis isn't happening to EPDs alone. ISO 14001:2026 separately strengthened internal audit rigour and extended accountability across the value chain. ISO has also published a dedicated horizontal standard for this exact problem, the new ISO 14019 series, which sets out requirements for the competence and impartiality of bodies that validate or verify sustainability information, and distinguishes validation of forward-looking assumptions (relevant to the prospective EPDs above) from verification of historical, declared data.\n\nEPDs aren't being singled out here. They're catching up to a governance bar ISO is raising across its sustainability standards at the same time, which is part of why the rest of this revision holds together. Because ISO 14025 remains a voluntary standard with no automatic enforcement mechanism, none of this lands on day one across the board. A staggered rollout is the realistic expectation: GPI and terminology updates likely come first, verifier approval and tool verification procedure changes follow, and the unwinding of conflict-of-interest arrangements the new \"third party\" definition no longer permits comes last, simply because those arrangements take longer to dismantle than they took to build.\n\n## What this doesn't change\n\nNone of this changes how an LCA itself is done. ISO 14040 and ISO 14044 remain the methodological foundation, untouched by this revision. What ISO 14025:2026 changes is everything around it: what counts as an EPD, what doesn't, who has to agree on the rules, whose tools are trusted to produce the data, how early that data can responsibly be shared, and who's accountable for checking all of it.\n\n## Where the six changes land\n\nPut the six pieces back together and they read as one sequence rather than six separate updates. Fixing the term first (1) is what makes it worth defending with a hard boundary (2). A defended term is what makes harmonised methodology across a dozen-plus operators (3) worth pursuing instead of ignoring. Harmonised methodology is what makes it safe to trust software that automates part of the process (4). Trusted tools and converging methodology are what make it defensible to publish an EPD before a full year of data exists (5). And every one of those five steps only holds up if the verification underneath all of it is genuinely independent (6), which is why governance is the load-bearing change rather than the last item on the list.\n\nThe practical read for anyone holding or issuing EPDs: the rule on named, third-party verification for consumer-facing declarations is the one item that needs attention now, since it carries no phase-in window at all. Everything else, terminology updates, tool verification procedures, the unwinding of conflicted review arrangements, will phase in on the operators' own schedules rather than all at once.\n\nThe next phase of EPDs will be defined by clarity, consistency, and credibility. That's where real progress begins.", "image": "https://verdatir.com/_astro/iso-14025-2026-hero.W7dP3Vve.jpg", "date_published": "2026-07-10T00:00:00.000Z", "date_modified": "2026-07-10T00:00:00.000Z", "authors": [ { "name": "Verdatir Research", "url": "https://verdatir.com/vision" } ], "tags": [ "Research", "EPD", "ISO 14025", "Sustainability", "Regulation", "Verification" ], "language": "en", "_verdatir": { "markdown": "https://verdatir.com/insights/iso-14025-2026-epd-ecosystem-maturity.md" } } ] }