A company publishes a sustainability report claiming fifty thousand tonnes of CO2 equivalent. Here's the thing — that number could be off by a factor of three depending on whether they used spend-based estimates or actual supplier data for their supply chain. And right now, June twenty twenty-six, that gap isn't just an accounting curiosity. It's the first full reporting cycle under the EU's Corporate Sustainability Reporting Directive for thousands of companies, and the SEC's climate disclosure rule is facing its first real enforcement tests. The difference between what you report and what's actually happening has never been more legally consequential. So the prompt asks: what is carbon intensity, how do organizations actually measure their Scope one and Scope two emissions, and when Scope three data from suppliers doesn't exist — which is most of the time — how do they fill the gap? And what does a business actually need to do to start collecting meaningful environmental data?
The honest answer to that last part — what they need to do — is surprisingly mundane. It's spreadsheets, it's PDF invoices, it's calling suppliers who don't pick up the phone. The mechanics are boring and difficult in equal measure, which is exactly why most coverage skips over them.
Let's start with carbon intensity, because it's the metric that makes comparisons possible. If I tell you a steel plant emitted a million tonnes of CO2 last year and a software company emitted fifty thousand, the steel plant looks worse. But the steel plant might be generating ten billion dollars in revenue and the software company five hundred million. Carbon intensity normalizes that — emissions per unit of economic output, typically kilograms of CO2 equivalent per million dollars of revenue, or per kilowatt-hour generated, or per ton of product shipped. It lets you ask: per dollar of value created, how dirty is this business?
Right, and this is where the first big misconception lives. Carbon intensity is not the same thing as a carbon footprint. A company can reduce its carbon intensity — getting more efficient per dollar of revenue — while its absolute emissions go up, because revenue is growing faster than efficiency gains. A tech company might double its data center footprint, increase total emissions by forty percent, but if revenue tripled, the intensity metric actually improved. That's not cheating, it's just math. But it means intensity is a useful benchmarking tool, not a measure of whether a company is doing enough in absolute terms.
It's the miles-per-gallon of corporate emissions. Your fuel economy can improve while you drive twice as far.
And just like fuel economy ratings, the methodology for calculating it matters enormously. Which brings us to the three scopes. The Greenhouse Gas Protocol, which is the accounting standard everyone uses, divides emissions into Scope one, two, and three. Scope one is direct — your own smokestacks, your fleet vehicles burning fuel, refrigerant leaks from your air conditioning. Scope two is purchased energy — the electricity, steam, heating, and cooling you buy. Scope three is everything else in your value chain, upstream and downstream. The definitions are straightforward. The measurement is where things get genuinely messy.
The question at the heart of this isn't "are companies lying." It's "are they even capable of knowing.
Most companies want to report accurately — there's reputational risk, there's regulatory risk, there's investor pressure. But accurate measurement is expensive, technically difficult, and in many cases depends on data from organizations you don't control and who have no obligation to help you. The result is a system built on proxies and estimates, where two companies with identical physical emissions can report numbers that differ by an order of magnitude, entirely legally, because they chose different accounting methods.
Walk me through Scope one. If I'm a logistics company with a hundred delivery trucks, what do I actually do?
Scope one covers direct emissions from sources you own or control. For most companies that aren't heavy industrial, this means combustion — natural gas for heating, diesel or gasoline for fleet vehicles, and refrigerant leaks from HVAC systems. There are two ways to measure it. Method one is direct monitoring — continuous emissions monitoring systems, or CEMS, which are basically sensors in the smokestack that measure what's actually coming out. These are accurate, sometimes down to within two to three percent error, but they're expensive. A CEMS installation can run hundreds of thousands of dollars per stack. You mostly see these at power plants, cement kilns, large refineries — facilities where the volumes justify the cost and where regulators require them anyway.
For everyone else?
Method two is calculation-based, using emission factors. You take the volume of fuel you burned, multiply it by a standard emission factor, and that's your number. For your logistics company with a hundred trucks, let's say they burn two hundred thousand gallons of diesel annually. The EPA's AP-forty-two emission factor for diesel, updated in twenty twenty-four, is ten point two one kilograms of CO2 per gallon. Multiply that out, you get two thousand forty-two tonnes of CO2. That's the number that goes in the report.
Then reality intervenes.
Because what if they're using a five percent biodiesel blend? The emission factor changes. What if they're idling in city traffic versus highway driving? The actual combustion isn't identical. What if they're using the IPCC's emission factor instead of the EPA's? The IPCC default for diesel is about seven point four kilograms per gallon for CO2 alone, but when you add methane and nitrous oxide converted to CO2 equivalent, it varies. Small factor choices — different databases, different assumptions about fuel composition — create five to fifteen percent variance in the final number. And that's Scope one, which is supposed to be the easy one.
The one you own and control.
The one where the fuel is sitting in your own tank. And you can still be off by fifteen percent depending on which government table you looked up.
Scope two is where the accounting choices get even more creative.
Scope two is purchased electricity, steam, heating, and cooling. The GHG Protocol's Scope Two Guidance from twenty fifteen gives companies two options. Option one is location-based — you take the average grid emission factor for the region where you consume the power. In New York, using eGRID twenty twenty-four data, that's about zero point zero zero zero four one tonnes of CO2 equivalent per kilowatt-hour. You multiply your electricity consumption by that number, done. It reflects the actual physical reality of the grid you're plugged into.
Option two is where the fun begins.
Option two is market-based. This lets companies use emission factors tied to specific energy purchases — renewable energy certificates, or RECs, in the US, guarantees of origin in Europe. If you buy a REC from a wind farm, you can claim that wind farm's emission factor for your electricity, which is essentially zero. The problem is that RECs are unbundled from the physical electrons. A company in New York can buy cheap unbundled RECs from a wind farm in Texas — and I mean cheap, sometimes under a dollar per megawatt-hour — and claim one hundred percent renewable energy for their New York office, even though the actual electrons powering their building come from the local natural gas plant.
You're not buying renewable energy. You're buying the right to say you did.
You're buying the environmental attribute, which is a legal construct. And this isn't some loophole — it's explicitly allowed under the GHG Protocol. The CDP's twenty twenty-four disclosure data shows roughly sixty percent of Fortune five hundred companies use market-based reporting for Scope two. The result is that two identical office buildings in New York, using the exact same amount of electricity from the exact same grid, can report Scope two emissions differing by a factor of ten. One reports location-based and gets the real grid number. The other reports market-based with RECs and gets something close to zero.
The carbon accounting equivalent of putting a different sticker on the same car.
Both stickers are legal. The market-based method was created with good intentions — to incentivize renewable energy purchasing and to give companies credit for directly supporting clean energy projects. And when companies buy bundled RECs, which are tied to a specific power purchase agreement where they're actually contracting for the physical delivery of renewable energy, it's meaningful. But the unbundled REC market has become, in many cases, a paperwork exercise. You can buy enough RECs to claim carbon neutrality for a few thousand dollars, regardless of your actual consumption.
Before we even get to Scope three, we've got Scope one varying by five to fifteen percent based on factor selection, and Scope two varying by an order of magnitude based on accounting method. What does the underlying data collection actually look like? Because someone has to gather all this information.
This is where the spreadsheet reality hits. For Scope one, a company needs fuel purchase records — invoices from fuel suppliers showing gallons or therms purchased, ideally broken down by facility. They need fleet mileage logs if they want to do activity-based calculations rather than just fuel-purchase-based. They need refrigerant recharge invoices to account for fugitive emissions from air conditioning — and refrigerants have global warming potentials thousands of times higher than CO2, so even small leaks matter. For Scope two, they need utility bills for every facility, with kilowatt-hour consumption and tariff details, and they need them in a consistent format across potentially hundreds of locations.
Most mid-sized businesses don't have a centralized system for any of this.
A CDP survey from twenty twenty-four found that forty-three percent of companies still use manual data entry for at least one scope. Data lives in spreadsheets across different facilities, sometimes in paper invoices or PDFs that someone has to manually extract numbers from. Facilities in different countries use different units, different billing cycles, different languages. A multinational manufacturer might have thirty facilities across fifteen countries, each sending monthly utility data in different formats to a sustainability manager who's compiling everything in Excel. The first year of data collection is essentially an archaeology project.
Like trying to reconstruct a dinosaur from a drawer full of receipts.
Half the receipts are in a language nobody at headquarters speaks. So that's Scope one and two on paper versus in practice. But here's where it gets messy — Scope three, where you're dependent on data from organizations you don't control.
Scope three is the supply chain. The fifteen categories under the GHG Protocol, but the heavy hitters are Category one — purchased goods and services — and Category four, upstream transportation. Everything your suppliers do before the product reaches you, plus everything that happens after it leaves.
The fundamental problem is that you need emissions data from your suppliers, most of whom don't measure their own emissions, and even if they do, they may not want to share that data with you. A CDP study from twenty twenty-five found that only eighteen percent of suppliers provide primary emissions data to their customers.
The other eighty-two percent of the time, the buying company has to estimate. And they have three methods for doing that.
Walk me through them.
Method one is the spend-based method. This is the simplest and crudest approach. You take the amount of money you spent on a category of goods or services, multiply it by an environmentally-extended input-output emission factor. These EEIO factors come from databases that model the average emissions intensity of different economic sectors. So if you spent a million dollars on electronic components, and the EEIO database says the electronics manufacturing sector emits zero point three two tonnes of CO2 equivalent per dollar of output, your estimated Scope three emissions for that purchase are three hundred twenty tonnes.
You're not measuring what your supplier actually emitted. You're measuring what the average supplier in that sector probably emitted, scaled to your dollar spend.
It requires no supplier engagement at all — just your own accounting data, which you already have. The downside is the error margin. Spend-based estimates typically have error bars of plus or minus forty to sixty percent. The variance comes from multiple sources — the EEIO models themselves are based on sector averages that may be years out of date, the dollar spent doesn't account for price variations between suppliers, and two suppliers in the same sector can have radically different emissions profiles depending on their energy mix and processes.
It's the carbon equivalent of guessing someone's weight based on their grocery bill.
That's actually a perfect analogy. You know they bought food, you know roughly what kinds of food people in that demographic buy, but you don't know if they ate it all, shared it, or fed it to their dog. Method two is the average-data method. Instead of using dollars spent, you use physical units — kilograms of steel purchased, liters of chemical feedstock, number of electronic components. You multiply those physical quantities by industry-average emission factors per unit. This is better than spend-based because it removes price variation. A kilogram of steel has roughly the same embodied carbon regardless of what you paid for it. But it's still an industry average, not your supplier's actual emissions.
Method three is the one everyone wants but almost nobody has.
Supplier-specific method. Actual primary data from your actual suppliers about their actual emissions. This requires your suppliers to measure their own Scope one and two emissions, calculate the share attributable to the products they sell you, and report that data in a consistent format. It's the gold standard, and as we just noted, only eighteen percent of suppliers do it.
The reality for most businesses starting out is: you've got five hundred suppliers, three of them have emissions data, and your sustainability report is due in six months. What do you actually do?
You start with spend-based for everything. That's year one. You pull your accounts payable data — every dollar you spent with every supplier — map each supplier to an EEIO category, multiply, and you have a number. It'll be wrong, possibly by fifty percent or more, but it's a starting point. The act of producing that number is itself valuable because it forces the organization to look at its supply chain through an emissions lens for the first time. You discover things like "we spend fifteen million dollars a year on aluminum castings and that single category appears to be driving thirty percent of our estimated Scope three emissions.
Which is where the hotspot analysis comes in.
Year two to three, you identify the hotspot categories — the twenty percent of suppliers driving eighty percent of emissions. For most manufacturers, it's materials — steel, aluminum, chemicals, electronics — plus logistics. You focus your supplier engagement efforts on those twenty to thirty suppliers. You send them questionnaires, you request primary data, you offer assistance.
Most of them still don't give you anything.
A few do. In the typical pattern, you might get three out of twenty hotspot suppliers to provide primary data in year two. The rest either don't have it or won't share it, so they remain on spend-based estimates. Year three to four, you switch to a hybrid model — primary data for the suppliers who gave it to you, spend-based for the long tail of small suppliers where the effort of engagement isn't worth the improvement in accuracy.
Patagonia is the case study here.
Patagonia's twenty twenty-three sustainability report laid this out clearly. In twenty twenty, they were at one hundred percent spend-based for Scope three. By twenty twenty-three, they'd reached sixty-two percent supplier-specific data coverage. That's impressive, but it required hiring three full-time staff just for data collection and verification. Three full-time employees, three years of work, and they still have more than a third of their supply chain on estimates.
The lesson for a mid-sized manufacturer with five hundred suppliers and a sustainability team of one person — who also handles regulatory compliance and maybe HR — is that this is a multi-year project that will require resources they probably don't have budgeted.
That's the practical reality most coverage misses. The market has responded with software platforms — Watershed, Persefoni, Salesforce Net Zero Cloud — that automate data ingestion from ERP systems like SAP and Oracle, pull utility data through APIs, and provide supplier portals for data collection. But these platforms cost between fifty thousand and five hundred thousand dollars a year. For a mid-sized manufacturer with fifty million in revenue, that's a meaningful line item. For a small business, it's completely out of reach.
For companies without those tools, what's the actual starting point? Someone gets an email from the CEO saying "we need to report emissions data, figure it out." What does their Monday morning look like?
They open Excel. They create three tabs — Scope one, Scope two, Scope three. For Scope one, they start pulling fuel purchase records from accounts payable — natural gas bills for facilities, diesel receipts for fleet vehicles, refrigerant invoices for HVAC maintenance. For Scope two, they pull utility bills for every facility, extracting kilowatt-hour consumption and cost data. For Scope three, they pull the full accounts payable ledger — every dollar spent with every supplier over the reporting period. Then they map each supplier to an EEIO category using one of the free databases, usually the US EPA's USEEIO model or the environmentally-extended input-output tables from the Global Trade Analysis Project.
The first year's data will be terrible.
But that's fine. The first year's numbers are a baseline, not a target. What matters is that you've built the data collection muscle. You now know which facilities have good records and which ones sent you scanned PDFs from twenty nineteen. You know which suppliers are responsive and which ones ignored your emails. You know that your spend-based estimate says purchased aluminum is your biggest emissions category, so that's where you focus next year's engagement effort.
There's another wrinkle here that doesn't get enough attention. The GHG Protocol requires companies to recalculate their baseline emissions when their methodology changes. So if you start with spend-based in year one and switch to supplier-specific data for some categories in year two, you have to go back and restate year one using the new methodology for those categories, so you're comparing apples to apples.
That's not a minor administrative footnote. The Science Based Targets initiative did an analysis in twenty twenty-four and found that thirty-seven percent of companies had to restate their baseline emissions within two years due to methodology changes. More than a third of companies changed their historical numbers. That creates real problems — your year-over-year progress looks different, your targets may need to be revised, and if you've already published your baseline publicly, you now have to explain why it changed.
Which looks terrible, even when the explanation is perfectly legitimate.
It looks like you're moving the goalposts. The only defense is documentation — meticulous records of which emission factors you used, which data sources, which assumptions, and why you made each choice. The companies that handle this well treat their carbon accounting like financial accounting, with an audit trail for every number. The companies that handle it poorly have a single spreadsheet maintained by one person who might leave the organization.
We're entering an era where the stakes of getting this wrong are escalating rapidly. The EU's CSRD requires limited assurance on Scope one and two by twenty twenty-eight, and reasonable assurance by twenty thirty. Limited assurance is basically a review — an auditor checks that your methodology makes sense and there are no obvious errors. Reasonable assurance is a full audit, closer to what financial statements get. For Scope three, the timeline is further out but it's coming.
This is the open question that keeps sustainability directors up at night. How do you submit numbers you know are plus or minus fifty percent accurate to a reasonable assurance audit? The auditor is going to ask for evidence. For spend-based Scope three estimates, your evidence is an EEIO database and a spreadsheet mapping suppliers to categories. That's not going to satisfy a reasonable assurance standard. The liability risk is real — if your reported numbers turn out to be materially wrong and you've made public claims based on them, you're exposed to regulatory action, investor lawsuits, and reputational damage.
The practical advice for someone starting this journey — and I'm assuming a lot of our listeners are at companies that are just beginning — what's the roadmap?
First, start with what you have. Don't wait for perfect data because perfect data doesn't exist. Spend-based Scope three estimates are rough, but they create organizational awareness and they give you a number to improve from. The first year's numbers will be wrong — accept that, document your methodology, and treat it as a baseline.
Second, focus on the eighty-twenty. For most companies, three to five suppliers or categories drive the majority of emissions. Identify those hotspots and invest your limited supplier engagement resources there. The remaining eighty percent of suppliers can stay on spend-based estimates with acceptable error. You don't need primary data from the company that supplies your office paper.
Third, build for iteration. Document every choice — which emission factor database you used, which version, which assumptions about fuel blends, which Scope two accounting method. Assume that someone will need to reconstruct your calculations two years from now, either because you're switching methodologies or because an auditor is asking questions. The thirty-seven percent restatement rate from that SBTi analysis isn't a failure rate — it's a reflection that methodologies are evolving rapidly, and your systems need to accommodate that.
There's one more thing worth flagging. We've been talking about this as a compliance exercise, but there's a competitive dimension. The companies that build robust data collection systems now — even if the numbers are rough — are going to have a significant advantage when reasonable assurance requirements hit. They'll have years of historical data, documented methodologies, and established supplier relationships. The companies that wait until twenty twenty-eight to start will be scrambling.
The supplier dynamic is shifting too. As more large companies demand emissions data from their suppliers, the suppliers who can provide it will have a competitive advantage. We're already seeing this in industries like automotive and electronics, where OEMs are requiring suppliers to set science-based targets as a condition of doing business. The eighteen percent supplier response rate we see today will increase, not because suppliers suddenly care about climate disclosure, but because their customers are making it a condition of the contract.
The next frontier, and this is interesting, is real-time emissions tracking. The MIT Climate and Sustainability Consortium ran a pilot in twenty twenty-five that demonstrated near-real-time Scope one tracking for a steel plant using edge sensors — basically IoT devices on the smokestacks feeding data continuously rather than calculating emissions once a year from fuel receipts.
There are blockchain-based supply chain data sharing pilots, where emissions data gets passed along the value chain in a verifiable way. The vision is that when you buy a ton of aluminum, you get not just a material certificate but an emissions data packet that's been cryptographically signed by each entity in the chain. But scaling that to global supply chains — where you've got thousands of suppliers across dozens of countries with different regulatory environments and technical capabilities — that's a decade away at best.
We're in this awkward intermediate period. The regulatory requirements are accelerating, the methodologies are still evolving, the data infrastructure isn't there yet, and companies are being asked to report numbers they know are uncertain, under standards that will soon require audit-level assurance. It's like being told to build a bridge while you're still figuring out what materials you're working with, and the inspector's already scheduled.
Yet, the alternative — not measuring at all — is worse. Imperfect data drives action in a way that no data doesn't. Companies that measure their emissions, even badly, tend to reduce them. The act of measurement creates visibility, and visibility creates accountability. The system is flawed, but it's the system we have, and understanding its flaws is the first step to improving it.
And now: Hilbert's daily fun fact.
Hilbert: In the nineteen eighties, researchers studying bee waggle-dance dialects around Lake Baikal discovered that local honeybees performed a distinctive "figure-eight with a shimmy" variant, which they informally named the "Baikal Boogie" — though the term never appeared in any published paper and survives only in a single PhD student's field notebook from nineteen eighty-seven.
...right.
Here's the question we'll leave you with. As regulators move toward mandatory assurance of emissions data, and as the gap between reported numbers and physical reality becomes legally consequential, how will companies handle the liability of reporting numbers they know are plus or minus fifty percent accurate? The EU's timeline — limited assurance by twenty twenty-eight, reasonable assurance by twenty thirty — isn't that far away. And if thirty-seven percent of companies are restating baselines within two years under today's voluntary standards, what happens when those restatements carry legal penalties? That's the tension at the heart of this whole enterprise. Thanks to our producer Hilbert Flumingtop. This has been My Weird Prompts. Find us at myweirdprompts dot com or wherever you get your podcasts.
