How to actually offset carbon
Rees Calder · 17 April 2026 · 7 min read
In January 2023 the Guardian ran a nine-month investigation into Verra, the world's largest certifier of voluntary carbon offsets. The finding was startling. About ninety-four percent of Verra's rainforest credits did not represent a real tonne of avoided emissions. The investigation drew on three separate scientific studies. Verra disputed the methodology. The market largely agreed with the investigation.
The next year a paper in Nature Communications looked at the twenty largest corporate buyers in the voluntary market between 2020 and 2023. Eighty-seven percent of what they had purchased sat in credit categories the authors judged high-risk for not producing real or additional reductions. The total traded value of the market fell from a 2021 peak to $723 million in 2023, then $535 million in 2024.
If you have ever paid Delta fifteen dollars to offset a flight, there is a very good chance your offset did nothing.
Avoidance credits are usually theatre
Offsets come in two broad families. The first, and historically the largest, is avoidance. You pay a landowner or a project developer not to cut down a forest, or not to burn a stand of trees, or not to release some methane. The counterfactual is the problem. To claim the credit, the project has to show what would have happened without it. In practice, these baselines are set by the project itself, audited loosely, and almost always flatter the seller.
The Nature Communications study found that REDD+ avoided-deforestation credits, the category at the heart of the Verra investigation, delivered around six percent of the emissions reductions they claimed on average. The other ninety-four percent were counted but not real. Old cookstove projects in East Africa had similar problems. So did many industrial gas destruction schemes.
Avoidance is not always worthless. It is almost never reliably worth what it costs.
Removal credits are newer and much better
The second family is removal. You pay someone to pull a tonne of carbon out of the atmosphere and put it somewhere it cannot return for a long time. The obvious version is direct air capture, where machines pull CO2 from ambient air and store it underground. Climeworks in Iceland is the best-known operator. Heirloom and 1PointFive are scaling in the US. The cost is between five hundred and a thousand dollars per tonne today, falling by about ten percent a year.
Biochar sequestration converts agricultural waste into a stable carbon form that stays in soil for centuries. Enhanced rock weathering spreads crushed basalt on farmland, where it reacts with CO2 and mineralises it. Both are cheaper than direct air capture at roughly one to three hundred dollars per tonne, with more durability uncertainty than machines but still far more credible than a typical avoidance credit.
These are real. They are audited by people who actually measure the outcome. They are also two orders of magnitude more expensive per tonne than the cheapest avoidance credit you can buy. This is not coincidence. The cheap credits were cheap because they were not really removing anything.
In 2024, even while the overall voluntary market shrank by about a quarter, removals grew by three hundred and eighty-one percent off a tiny base. This is the direction the serious money has moved.
If you want to actually offset
There is no free version of this that works. You have three honest options.
The first is to buy removal credits directly from an operator with a public, peer-reviewed monitoring protocol. Climeworks, Charm Industrial (bio-oil injection), Heirloom. Expect to pay between $150 and $800 per tonne. Stripe Climate, Frontier (a coalition of Stripe, Alphabet, Shopify, McKinsey and Meta that committed $1B to durable removal), and the Milkywire Climate Transformation Fund aggregate these for smaller donors.
The second is to give through a charity that funds political and systemic work rather than buying individual credits. Founders Pledge's Climate Fund, Clean Air Task Force, Good Food Institute, and the Gold Standard's policy work all target changes that prevent emissions at the source. Cost-effectiveness estimates here are highly uncertain but the direction is right.
The third is to skip the offset and reduce the emission instead. Every serious climate researcher who has written about this in the last decade has landed in roughly the same place. Flight emissions are best addressed by flying less or taking the train. Home emissions are best addressed by insulation and a heat pump. Offsetting a flight you took anyway is what an economist would call "paying to feel better."
The one thing to do this week
If you have an active offset subscription with any airline or travel aggregator, go and look at what they are buying. It will usually be Verra REDD+ credits or cookstove projects. Cancel it.
Redirect the same money to Milkywire's Climate Transformation Fund, Founders Pledge's Climate Fund, or direct purchase at Stripe Climate. You will offset less carbon in tonnes on paper. You will offset more in reality.
Sources used
- Guardian, "Revealed: more than 90% of rainforest carbon offsets by biggest certifier are worthless," 18 January 2023 (based on West et al., Coffield et al., and Guizar-Coutiño et al. peer-reviewed analyses)
- Trencher et al., "Demand for low-quality offsets by major companies undermines climate integrity of the voluntary carbon market," Nature Communications, 2024, DOI 10.1038/s41467-024-51151-w
- S&P Global Commodity Insights, "Verra makes major changes to methodology of forest-based carbon offsets," April 2023
- Ecosystem Marketplace, State of the Voluntary Carbon Markets reports 2023 and 2024
- Frontier Climate purchase framework, https://frontierclimate.com
- LSE International Development blog, "The Verra scandal explained," January 2023