What it Really Takes to Prove a Tonne

It's the first week of the month, and a small team of us is doing something that doesn't look like climate work.

We're reconciling propane invoices. Cross-checking originating manifest weights against receiving scale tickets. Sweeping email for a missing argon receipt. Reading a pH value off a clipboard photo. Confirming the lab analyzed the right tote.

This is what producing a tonne of carbon removal actually requires — and we think it's worth talking about.

The thing nobody asks about

When a buyer purchases a tonne of CDR from us, they're not buying a number on a spreadsheet. They're buying a chain of evidence — a verifiable trail of how a pile of woody biomass from wildfire fuel reduction projects in Colorado becomes a stable cargo of bio-oil sequestered thousands of feet underground in Louisiana.

That trail is what MRV — Measurement, Reporting, and Verification — is for. It's the unsexy backbone of every credit we deliver.

So how thick is the trail?

We counted.

175 measurement types. One Charm tonne.

Every tonne we deliver inherits the same schema: 175 distinct measurements, organized across seven buckets that span the lifecycle from biomass intake to verified removal. Our measurements are broken in two categories: 

1. Activity Data Measurements - Live measurements that we collect every month for our reporting.

2. Scaling Factor Measurements - Scaling Factors that are verified by a third party, these measurements are tied into LCA models and our digital MRV software modeling each month for a repeatable closing process. 


Our exciting buckets:

• Wildfire Mitigation Biomass Supply Chain — every load we accept gets weighed, sourced, and fingerprinted. Eight fields per truck, every truck.

• Production inputs — water, propane, diesel, grid power, pyrolyzer uptime. All metered, all logged.

• Lab-verified chemistry — independent labs run carbon, hydrogen, and nitrogen panels on every batch of bio-oil and biochar we ship.

• Co-product inventory — every tote of bio-oil and bag of biochar tracked from production to its final fate in the field.

• The journey to injection — trucks and railcars weighed at every transfer on trade-certified scales, sparging losses subtracted, pH confirmed, every mile of transport counted against the tonne.

• Injection Site  emissions — every gallon of diesel, every kWh from the grid, all of the consumables required to run the injection site.

• Embodied emissions — LCA documents created for every piece of capital equipment we deploy, amortized and netted against removals.

3,300 data points a month

The schema is the floor. The volume is what fills it.

Across a typical month, our MRV system touches more than 3,300 distinct data points to substantiate the period's removals. Every delivery of biomass that comes into our facility, every container of bio-oil and bio-char that leaves it, every kWh of electricity and gallon of fuel we use is captured, logged, tracked, and quantified in terms of GHG impact. 

And that's the conservative count. We didn't include every replicate sub-measurement on the lab side, every QA-checklist tick on every batch, or every line item on our spare-parts inventory. Add those in and the real number is meaningfully higher. 

If you’re familiar with this blog, you probably know we don’t just do this by hand. We’re working with drones that track biomass inventory, extensive automation across our distributed sites, and streamlining clunky procedures. Plus, if you want to dig into the technology that powers all this number crunching, just keep on reading. But before we get to the “How” of it, we’ll take a tip from Nietzsche and start off with the Why.

Why this matters

There's a temptation in CDR to wave at rigor and move on. Methodology PDF on the website, third-party logo in the footer, done.

We think that's how trust in this market gets eroded. Carbon removal isn't just a claim. It's a chain of evidence — and the chain is only as strong as the weakest measurement.

That's why every tonne we sell goes through:

• An independent registry methodology (Isometric)

• Charm internal QA

• A third-party validation and verification body (350 Solutions)

• A buffer pool against reversals

• Public documentation of every assumption, EF source, and uncertainty discount

This way, we keep the physical reality of CDR tied to the data and vice versa. No biomass = no bio-oil = no carbon removal. And no measurement = no tonne.

The number we want to keep growing

If we're being honest, 175 is not where we want this to land long-term. It's where we are today. As Charm scales — more sites, more product lines, more pathways to permanence — the schema grows with us. Biochar in soil adds a different measurement set than bio-oil in former oil reservoirs. Each new pathway adds its own evidentiary requirements.

Our job is to keep the chain unbroken as the system gets bigger.

So next time you see a number on a CDR purchase agreement — a tonne, a kilotonne, a million-tonne offtake — know what's underneath it. A scale ticket. A lab certificate. A meter reading. A BOL signed by a driver in Texas at 4am. And a whole team of people working around the clock to verifiably turn the tide on climate change tonne by tonne.

Now a reasonable person might ask, “why do you subject yourselves to this?” Well, it’s because that's what proving a tonne actually takes. That proven tonne shows that we are having the impact that our customers pay for us to make, and that gives us motivation to keep pushing on one of our species biggest problems every day.

However, that leaves open a critically important question: How does this scale? Rigor and verifiability are non-negotiables when it comes to building and maintaining trust in CDR. But if it becomes an impediment to hitting the gigaton scale, we still haven’t gotten where we need to go. 

We agree, and that’s exactly where our tech stack comes in. Speak “nerd” and enter!

How we built the chain

A year ago, a lot of this was a person and a spreadsheet.

Now most of it isn't.

Our MRV stack today is a network of Python harvesters and orchestrators sitting on top of the Google Workspace APIs (Gmail, Drive, Sheets, Docs), a BigQuery-backed biomass throughput query, Mangrove’s dMRV platform, and the Isometric Certify API. Together they cover the boring 90% of MRV work and let us focus on the interesting 10% where real judgment calls are necessary.

Vendor invoices are a good example. Each fuel and gas vendor — bulk propane, cylinder propane, diesel, gasoline cards, liquid argon — has a dedicated harvester. The harvester searches Gmail for the vendor's invoices over the period, parses the PDFs (line items, gallons or cubic feet, dates, freight miles), files the source PDF into the matching Drive receipts folder, and returns paste-ready rows for the calc sheet. If a period comes back empty, it flags the gap rather than guessing.

For lab data, we run two parallel filers — one for bio-oil certificates of analysis, one for biochar — these match each sample's CLIENT ID back to the originating batch UID and file the PDF into that batch's testing folder. Carbon, hydrogen, and nitrogen values flow into the calc sheet automatically; the tote-to-batch mapping is enforced at file time so we never have an orphaned sample.

Month-end close is run by an orchestrator that scaffolds the next period's folder structure from a JSON config, copies prior-period sheets, clears the data-entry cells, and spawns the harvesters in parallel. It produces a dry-run reconciliation report — totals across calc sheets, gaps, anomalies — that we review before any data gets written.

On the registry side, we've automated the most painful loop in MRV: rebuilding a verifier-ready GHG Statement from per-removal results. Isometric has built this out from their end — Certify (Isometric’s CDR reporting platform) now exposes both a REST API and an MCP server — so a script pulls each removal's mass, carbon content, net and gross CO₂e, per-component emissions and uncertainty, sorts the components into process, site, and reduction categories, and writes a fully formatted sheet into the reporting package.

And critically — none of this skips a verifier. Every harvested invoice is filed as the original PDF. Every extracted value is reconcilable to the source. Every formula in every calc sheet uses cell references back to an annually updated canonical set of standard  emission factors, so when an emission factor updates, the change propagates and the lineage stays intact. Even before hitting a verifier, all outputs have to satisfy our “4+ Eyes” rule: Every GHG entry is reviewed by the person putting it together, a digital QA agent who checks completeness and computational accuracy, and a final human review of those findings. We don’t use AI and automation to replace human judgment, but to enable it at scale by allowing the MRV team to focus on real judgment calls, assurance of accuracy, and system optimization while delivering high-integrity CDR at scale. 

The point isn't the tooling. The point is that automation lets us push rigor up as we scale up. Every hour we save on data plumbing is an hour we spend on the harder questions: Is this emission factor still the right one? Why does this scale ticket disagree with the BOL by 40 kg? How can we eliminate manual entry tasks for operators and replace them with streamlined measurement practices that are both easier and more accurate?

That's where the trust gets built: Creating scalable systems that allow buyers, verifiers, and our own internal assurance teams to see how and why any given Charm tonne credited maps to at least a tonne of CO2e durably retired from the atmosphere. We’ve got the receipts.