Carbon offsets have a trust problem, and most of the people selling them know it. The market has been hit by a steady run of stories about phantom forests, double-counted credits, and projects that priced in benefits which never showed up. When someone tells me they have solved this with a blockchain, I usually ask them to slow down. A public ledger fixes some of the fraud surface in carbon accounting. It does not fix the rest. The honest version of the pitch is narrower than the marketing version, and worth setting out plainly.

What a public ledger genuinely fixes

Strip the slogans away and a blockchain is a shared append-only record that anybody can read and nobody can quietly rewrite. For on-chain carbon, that single property addresses a specific, well-documented failure mode: the same tonne being sold twice. In legacy registries, retirement of a credit is a database row managed by a single operator. If the operator is sloppy, captured, or simply offline, the same serial number can end up claimed by two buyers. Auditors then spend months untangling it.

Move the retirement event onto a public chain and that ambiguity goes away at the bookkeeping layer. The transaction is timestamped. The wallet that retired the credit is visible. Anyone can verify that a given serial has been burned and cannot be resold. That is a real improvement, and it is the reason we built IMPT.io's offset flow the way we did. Every booking on our platform offsets one tonne of CO₂, and that retirement is recorded on-chain so a buyer, an auditor, or a curious journalist can check it without asking us for permission.

So the ledger gives you three things, and only three things, for free:

• Uniqueness. A retired credit cannot be retired again.
• Public observability. The retirement event is visible without a login.
• Tamper-evidence. Nobody can rewrite the retirement record after the fact, including us.

That is the entire honest list. Everything else people claim for on-chain carbon sits on top of work that the chain itself does not do.

What the chain does not fix

The hard part of carbon credit fraud has never been the bookkeeping. It has been the underlying claim. Did the trees actually get planted? Were they going to be planted anyway? Did they burn down two years later? Was the methodology honest about leakage and additionality? A blockchain has nothing to say about any of that. It records what someone wrote down. If what they wrote down was nonsense, the chain will faithfully preserve the nonsense forever.

I want to be blunt about this because the industry has not been. On-chain does not mean verified. A token minted against a bad project is still a bad credit. Tokenising a junk methodology gives you a tradeable junk methodology. The chain is downstream of the science, the field measurement, the satellite imagery, the third-party audit, and the standards body. If any of those upstream steps are weak, the ledger inherits the weakness without complaint.

This is the trap that the first wave of crypto-carbon products fell into. Projects rushed to bridge old, low-quality credits onto chains, generated huge supply, and then acted surprised when the standards bodies pushed back and the secondary market collapsed. The lesson was not that blockchains are useless for carbon. The lesson was that putting bad data on a fast settlement layer just lets you distribute bad data faster.

Where the real verification has to live

If the ledger is downstream, the question becomes: what do you put upstream of it? In practical terms, three things have to be in place before a tonne of CO₂ shows up on a chain in a way I would defend.

A methodology you would publish

The standards bodies — Verra, Gold Standard, the newer ICVCM Core Carbon Principles work — exist for a reason. They are not perfect, and they have been criticised, fairly, for being slow. But the discipline of writing down what counts as a tonne, how additionality is tested, and how leakage and permanence are handled is the whole job. A project that cannot point to a published methodology is asking you to take its word for it. The chain will not save you from that.

Measurement you can re-run

Modern remote sensing has changed what is possible here. Satellite-based forest monitoring, soil carbon sampling protocols, and independent MRV (monitoring, reporting, verification) providers can be cross-checked by a third party years after the credit was issued. The right question to ask any offset provider is not "is it on-chain" but "if I hire someone independent to re-measure this in 2029, what will they find?" If the answer is hand-waving, walk away.

A retirement event that matches a real claim

The retirement on the ledger needs to correspond to a specific, named project with a specific, dated vintage. "One tonne offset" is not a claim. "One tonne, vintage 2024, project X, methodology Y, retired on date Z by wallet W" is a claim. The first is marketing. The second is accounting. The chain helps with the second and does nothing for the first.

Why we still chose to build this way

Given all of the above, you might fairly ask why IMPT bothers with on-chain retirement at all. Why not just buy credits, retire them in a normal registry, and put a PDF on the website?

Two reasons. The first is scale and speed. We are stitching offsets onto travel bookings across 1.7 million hotels in 195 countries. The volume of individual retirement events that produces is not something a quarterly PDF handles gracefully. A programmable ledger does. Each booking can trigger a retirement that is visible within minutes, not months.

The second is that we wanted the accounting to be inspectable by people who do not trust us, and should not have to. I run the company. I am not a neutral party on whether IMPT.io is doing what it says it does. The chain lets a sceptic verify the retirement side of the claim without asking my permission, reading my marketing, or believing my staff. That is a healthier arrangement for everyone. It also forces us to be honest, because the ledger does not care about quarterly narratives.

What it does not let the sceptic do, and I want to keep saying this, is verify the project upstream. For that, we lean on the standards bodies and on the project documentation we link to. The chain is the receipt. It is not the audit.

How to read a "verifiable offsets" claim

If you are evaluating any platform, ours included, that uses words like "on-chain carbon" or "verifiable offsets," here is the short checklist I would use.

1. Find the project. Not the platform. The actual project where the carbon was avoided or sequestered. Name, country, methodology, vintage.
2. Find the standard. Verra, Gold Standard, Plan Vivo, Climate Action Reserve, or a peer of those. If the standard is one nobody outside the company has heard of, that is a flag.
3. Find the retirement. A serial number, a transaction hash, or both. You should be able to click through.
4. Check the claim. Does the retirement match the marketing? "One tonne offset per booking" should be one retirement per booking, not one retirement per thousand bookings rounded up.
5. Ignore the logos. Partner logos and badges tell you nothing about credit quality. A credit either has a defensible methodology and a real retirement, or it does not.

That checklist will get you further than any whitepaper.

The narrower, more useful claim

The version of "on-chain carbon" worth defending is small. It says: the bookkeeping of retirement is now public, unique, and tamper-evident, which removes one specific class of fraud — double-counting and quiet reversal — from the system. It does not say the underlying credits are good. It does not say tokenisation creates new climate value out of thin air. It says the receipts are honest. The project quality question lives where it has always lived: with standards bodies, with independent measurement, and with the willingness of buyers to ask uncomfortable questions before they pay.

I think that narrower claim is enough. It is genuinely useful, it is testable, and it stops the conversation from drifting into the magical thinking that has hurt this market for a decade.

What to do this week

If you buy offsets for a company, pull one of your recent retirements and try to walk the chain backwards yourself: receipt, serial number, project, methodology, vintage. If any link is missing, ask your provider for it in writing. If you book travel through IMPT.io, the on-chain retirement is there to be inspected, and we would rather you checked it than took our word for it. We are working on making that walk-back even shorter for a non-technical reader, because a system that is only verifiable in theory is not really verifiable. The ledger is the easy part. The hard part is making sure the tonne it points at was real in the first place, and that work never stops.