Is a factory really the biggest polluter?

Whenever we pass a bunch of factories emitting smoke, what’s the first thought that comes to mind? That it’s causing so much pollution.

What if I told you that the factory belonging to a car company only amounts to something like 2% of that car’s entire carbon footprint?

So where does the other 98% come from?

Upstream and downstream. Parts being manufactured in different countries and shipped to that factory. And even at the end of the car’s life, when you’re done driving it, emissions keep adding up. Recycling the parts, dumping the car as waste, all of it.

Today’s deep dive is about how all this emission tracking works and why we need it. Is it really helpful, or is it just a tangled web of corporate jargon?

Whose emissions are these?

If we’re going to track carbon emissions, we have to know which ones belong to you and which don’t. How does a company decide that?Let’s use a roommate analogy. You share an apartment and your name is on the lease. Are you responsible for a share of the trash everyone generates, or only the trash you generate?

The GHG Protocol, the rulebook for carbon accounting, gives companies three ways to draw this boundary.

The first is operational control. If you’re the one making the chore chart and implementing changes, you’re responsible for tracking the footprint.

The second is financial control. Your name is on the lease but you don’t live there, your friend does. You’re still responsible for taking the trash out. Same with the carbon footprint.

The third is equity share. If you own 30% of the flat, you’re responsible for 30% of the footprint that comes with it.

Now there’s a flaw. Say you own 50% of a polluting steel mill and your partner runs it. If you pick operational control, your 50% stake disappears from your books because you don’t run the place. Pick equity share and it would’ve counted. (A company has to pick one approach and stick with it across the whole business, but still.) Does that mean the emissions actually disappear? No. They just get pushed out of your Scope 1 and 2 and into Scope 3.

Scope 1: the smoke from your own chimney

“Scope” here just means different buckets for tracking a company’s emissions.

Scope 1 is direct emissions, like the ones from the factory. You have operational control, so they’re yours. Mathematically they are represented as :
Emissions (tCO2e) = Fuel Volume × Heat Content (HHV) × Emission Factor (EF) × GWP

Fuel volume is how much fuel was physically burned. Heat content is how much energy is packed into that fuel. The emission factor tells us how much gas is released per unit of energy burned. Finally  global warming potential or GWP , which is basically a score for how much heat a gas traps compared to CO2. Methane, for example, traps around 28 times more heat than CO2 over 100 years, so 1 kg of methane counts as 28 kg of CO2e.

Scope 2: the coal plant you never see

Scope 2 is the indirect stuff. The electricity, heat, steam or cooling you buy. Take your office. It runs on electricity, right? And that electricity is coming from a power plant, maybe burning coal, in a totally different place from where you’re sitting.

The GHG Protocol lets companies report Scope 2 in two ways, location-based and market-based.

Location-based looks at the average carbon intensity of the grid you’re plugged into. If your local grid is 40% coal, your footprint reflects that.

Market-based looks at what you bought, using financial instruments like renewable energy certificates (RECs).

Think of the grid as a water bowl. Everyone pours water in, everyone draws water out. A company wants a green office, so it buys certificates from a wind farm in Texas, a few states away. That wind farm’s electricity goes into the Texas grid like normal. The company gets a certified receipt and says, “we now run on green electricity,” even though physically the office is drawing from its own local grid, which is 50% coal.

To me this sounded like an accounting scam. It isn’t. Once electricity is on the grid you can’t track electrons. The certificate is the only way to track who paid for the clean part. And the GHG Protocol designed it this way on purpose. Without that demand, the Texas wind farm would struggle to fund more turbines.

But another question kept coming to my mind: doesn’t this create double counting?

Say a grid runs on 20% wind and 80% coal. Ron builds a data center on it and buys the wind farm’s certificates to cover his electricity. Now he claims his data center runs on green electricity. Once those certificates are issued and used, they’re retired. Nobody else can claim that wind again.

Now Bob has a factory on the same grid. He bought no certificates. If Bob simply got the grid average, he’d be claiming a share of the same wind Ron already claimed. So he doesn’t. Bob gets the residual mix, which is the grid’s mix after all the retired certificates are taken out. The wind is gone from his mix, so his electricity now counts as 100% coal. His reported emissions go up.

Companies that report location-based don’t get hit by this. They always use the plain grid average. That’s why companies have to report both numbers.

The residual mix is a beautiful accounting mechanism designed to drive real-world green investment. It rewards companies that switch and penalizes the ones that don’t. One caveat: whether buying certificates really leads to new wind farms is debated, and the GHG Protocol is revising these rules right now, including proposals like hourly matching.

And here’s the astonishing part. Scope 2 is often only 5% to 10% of a company’s footprint.

Scope 3: everything you don’t own or control

For many companies, 75% to 90% of the footprint sits in Scope 3, the value chain. It’s spread across 15 categories.

Imagine you’re running a bakery. Upstream is everything it takes to get ingredients to you, from fertilizer to tractors to growing the wheat. That’s Category 1, purchased goods and services. The truck bringing wheat from the farm to your bakery is Category 4, upstream transportation.

Then there’s downstream. Remember that customer who drove to your bakery at 11 in the morning? That car’s emissions are your Scope 3. Category 9, downstream transportation. And it doesn’t stop there. When he’s done with the cake and throws the box in a landfill, that releases methane. Also yours. Category 12, end-of-life treatment.

Category 11, use of sold products, is huge for industries like automotive. Every time a customer drives the car, the gasoline they burn lands in the carmaker’s Scope 3.

And what if a company doesn’t make a product at all? Banks, for example. Category 15, investments, covers the emissions of every company they invest in or lend to, and it often makes up 95% or more of a financial institution’s total footprint.

By now you can at least appreciate how hard Scope 3 is to track. How do you track data you don’t have, or don’t even know exists? And even if a supplier on another continent hands you numbers, can you trust them? Two big hurdles here: transparency and data fragmentation. 

How do we calculate Scope 3?

There are two main ways, spend-based and activity-based.

Spend-based is easier because it uses data you already have. You take how much money you spent in a category and multiply it by an EEIO factor, which maps the average emissions produced per dollar spent in that sector. Buy $1,000,000 of steel and the database says, on average, that’s X tonnes of carbon.

This is also the fundamental flaw. Say you buy 1 ton of steel and your supplier gives you a 20% discount. Same steel, same emissions as last month. But you paid less, so on paper your emissions dropped even though nothing physically changed. Flip it with inflation and on paper you’re suddenly polluting more for the exact same steel.

To fix this, standards are pushing toward the activity-based way, or supplier-specific. Don’t count dollars, count the actual stuff. Tons of steel, liters of fuel. You ask your supplier for LCA factors, basically how much carbon it took to make one ton of that material, from digging it out of the ground to the factory gate.

That’s the cradle-to-gate part Emissions (tCO2e) = Mass of Material (tons) × Cradle-to-Gate LCA Factor (tCO2e per ton)

But you can’t get an LCA factor for every screw, chip or service you buy. So we use a compromise. LCA factors for your biggest suppliers, like steel and microchips. Spend-based for the small stuff, like office equipment and services, so the distortion doesn’t mess up the whole calculation.

Why does any of this matter?

With rising concern about climate change, reporting carbon emissions is increasingly required by law. Like filing taxes every year.

In California there’s SB 253, the Climate Corporate Data Accountability Act. Any US company making over $1 billion that does business in California has to report, public or private. Scope 1 and 2 came first, with the first reports due in August 2026. Scope 3 kicks in 2027.

The EU has CSRD and ESRS E1, which runs on double materiality. It used to cover a lot more companies, but in early 2026 the EU cut it down to companies with over 1,000 employees and €450 million in turnover. Double materiality is the interesting part. It’s not just how climate change hits the company’s finances, it’s also how the company hits the planet. That’s a crucial step toward bringing climate into the finances of the global economy.

The next time you pass a factory with smoke pouring out, remember it’s only the part of the footprint you can see. The other 98% is sitting in a supplier’s furnace somewhere, a customer’s exhaust pipe or a landfill.