Supplier Scope 3 emissions, explained.

What supplier Scope 3 data collection is: replacing an estimate of what you buy with what your suppliers measured

The yearly supplier data cycle in five steps, from request to signed figure

1. Screen and rankEstimate all categories, find where the tonnes sit, and pick the suppliers that matter. In the example, 240 of 1,860 suppliers cover 82% of purchased goods spend.
2. AskSend the climate data request in July, through a disclosure system, a questionnaire or a data exchange.
3. Read, check and swap inRead each submission, match it to the purchase line, test it, and replace the estimate only where it holds up. Ask a precise question where it does not. This step is what this guide is about.
4. Separate and write upSplit what changed into real change and method change, and write the method notes behind each figure.
5. Sign and defendThe head of sustainability approves the figures, and the assurance provider, an outside auditor of these figures, samples them and asks for method and source.

The people, companies and figures in this example are fictional.

Four surprises in Scope 3 supplier engagement, starting with data worse than an average

Scope 3, product carbon footprint, cradle-to-gate, method change: the vocabulary of supplier carbon data

Fourteen terms are defined in the one-page summary. Tonne (t) means a metric ton, the unit carbon accounts use, and CO2e means carbon dioxide equivalent.

Spend-based vs supplier-specific emissions, with hybrid and average-data between them: four ways to count one purchase, and why a dollar is a weak measure

In shortThe GHG Protocol names four methods for purchased goods. They differ in what you know: the supplier's product, the supplier's company, the material, or only the money. Most companies mix them, and the mix is what a reader wants to see. A spend estimate is a fair place to start and a poor place to stay, because it moves with price and with the industry code someone picked.

The GHG Protocol's guidance sets out the four methods for Category 1.2 The supplier-specific method takes a cradle-to-gate product inventory from the supplier. The hybrid method uses what the supplier knows and fills the gaps with secondary data, meaning published or database averages. The average-data method multiplies the mass you bought by an industry-average factor. The spend-based method multiplies money by a factor per unit of money. Sphera's 2025 survey of 315 sustainability professionals found 65% of companies combining these approaches, and 15% relying on spend alone, down from 30% in 2024.10

The same dollar across a drug maker's purchases: how far spend factors sit apart

Spend factors come from input-output tables, which divide a whole economy into industries. The set that many US companies use publishes its factors by industry code. Here is what one 2024 dollar carries in its "without margins" column, for things a drug maker buys.3

A code for finished medicines is a poor fit for a late-stage chemical intermediate, and so is a bulk organic chemicals code. Two careful analysts could choose either. The factor may be defensible, yet it tells you about an industry's average dollar in one country. It says nothing about your supplier's plant in another.

How one inventory splits across the four methods

Laying an inventory out by method shows how much of the number you know. Veltara's fictional FY2025 inventory shows the pattern: known in the categories where suppliers have data, averaged in travel, commuting and waste, and still mostly money in capital goods.

ESRS E1, the CSRD Omnibus, California's SB 253 and the GHG Protocol revision: what each asks of Scope 3 in November 2026

In shortThese rules moved again and again in 2026. The CSRD now reaches far fewer companies. The EU standards were rewritten and apply from financial year 2027. California's first report falls due November 10, 2026, with Scope 3 proposed for 2027. The GHG Protocol's Scope 3 revision is still a draft, with public consultation slated for Q2 2027. Date every statement you make about any of them.

What you may demand of a supplier under the CSRD value-chain cap, and how to ask for a product footprint anyway

In shortA supplier with up to 1,000 employees is a "protected undertaking". For CSRD reporting, you may not require more from it than a short list of datapoints, and that list holds company-level figures with no product's footprint among them. You may still ask. If you do, you must say which part goes past the limit and that the supplier may decline. Estimates remain acceptable for what you cannot get.

The rule sits in the amended Accounting Directive. A reporting company "shall not require" a protected supplier to provide information beyond the voluntary standards, and a contract clause saying otherwise "shall not be binding". Where a request goes past the limit, the company "shall ensure" the supplier is told which information exceeds it and that it has a "statutory right to decline". A company that stays within the limit is "deemed" to have met its value chain duty.8

The list itself is Annex II of Delegated Regulation 2026/1560, and the simplified ESRS says it applies to non-EU suppliers too.9,5 The list applies to financial years beginning on or after January 1, 2027, and national laws are due by March 19, 2027.9,8

Four things the cap leaves open for Scope 3 teams

Footprints, CDP responses, questionnaires, VSME reports and freight files: what each document can and cannot prove

In shortFive kinds of document arrive, and they answer different questions. A product footprint is the only one built for a single purchase line. A CDP response gives a company total, with no key to divide it by unless you ask. A short report from a small supplier may cover Scopes 1 and 2 only. Knowing what each one lacks tells you what to ask next.

How to read a supplier product carbon footprint: boundary, reference period, allocation, primary data share and verification

In shortFive fields decide whether a footprint can replace an estimate. Boundary must be cradle-to-gate for Category 1. The period must be close enough to your year. Allocation between co-products follows a set order. Primary data share says how much of the figure the supplier measured. Verification says who else has looked. PACT sets rules for each, and its verification roadmap runs in two stages, to 2030 and beyond.

The Partnership for Carbon Transparency, or PACT, is an initiative hosted by the World Business Council for Sustainable Development, and version 3 of its method is dated April 2025. Companies "shall assess the primary data share". From 2027 they also assess a data quality rating. If upstream values are unknown, they apply the worst case: 0% primary data and a rating of 5.11

Allocation, worked: why a vial's footprint depends on who else buys the tubing

When one furnace makes glass for vials and tubing sold to others, you have to split its emissions. PACT's order is to avoid allocation if you can. If you cannot, compare the economic value of the co-products: when the most valuable is worth more than five times the least valuable, allocate by value; otherwise use a physical measure where one fits. "In doubt, mass allocation should be prioritized".11 Here is why it matters, on illustrative numbers.

How to state the share of Scope 3 resting on supplier data without overstating it

The word "primary" invites overstatement. A footprint with a 71% primary data share is 29% secondary data inside. A company total, allocated to your purchases, is supplier data that you have divided by a key of your own. So "61% on supplier data" and "61% primary" are different claims. Veltara's fictional 3,112 tonnes from Kestrel contains about 2,210 tonnes that the supplier's document calls primary (3,112 × 71%).

ESRS's 2023 text speaks of "primary data obtained from suppliers"; the Scope 3 Standard of "data obtained from suppliers or other value chain partners".4,6 The GHG Protocol's draft adds that data type "is not by itself an indicator of data quality or data accuracy".7 State what your percentage counts. If you can, publish the product layer and the company layer separately. In the example they are 26% and 35% of the total.

Swapping a spend estimate for a verified footprint, line by line, and booking 3,900 tonnes as a method change

In shortThe arithmetic takes one line: kilograms bought times the supplier's figure. The accounting takes more care. The fall goes on a separate line from real reductions, your target and base year have to be treated consistently, and the share of supplier data moves too. Here is one swap, in full, from this guide's fictional case.

Real change against method change: what moved between two inventories

A year-on-year comparison mixes two things. Purchases grew, and suppliers cut emissions: that is real change. Better data replaced worse data: that is method change. In the example the reported total fell 25,800 tonnes. Only 7,700 of that is real.

A swap can also push the number up. In the example, one supplier of recovered solvent reports a footprint 410 tonnes above its estimate, because its recovery runs on gas. A program that only ever shows falls should make a reader suspicious.

Base years and targets: what the method change does to your reduction goal

If you hold a Scope 3 target, the Scope 3 Standard requires a policy for recalculating your base year, the year you track progress against, and says to recalculate "when significant changes in the company structure or inventory methodology occur".6 You set the threshold for "significant" yourself, so write it down before the swaps begin. Otherwise the base year stays on spend estimates while the current year moves to supplier data, and your apparent progress is partly an artifact. The opposite failure is restating every year, which leaves readers no fixed point.

Testing a chemical supplier's footprint against process mass intensity, peer ranges and the estimate it would replace

In shortA footprint can be well formatted and still wrong. Drug chemistry gives you tests that need no plant visit: how much material the route uses per kilogram of product, how much solvent is recovered, what similar routes report, and how far the figure sits from the estimate. A large gap in either direction earns a second look. Passing shows only that a figure is plausible.

Process mass intensity, or PMI, is the mass of everything used to make one kilogram of product: reagents, solvents, water. It is the drug chemist's shorthand for how heavy a process is, and it varies enormously by type of medicine. Published work puts small molecules at a median of 168 to 308, biopharmaceuticals at about 8,300, and solid-phase peptide synthesis at about 13,000.30

McKinsey, in an article we could read only in excerpts, gives average material intensity of 70 to 433 kg per kg for active ingredients, and emission factors of 50 to 1,000 kg of CO2 per kg of active ingredient.31

Eight checks for a supplier footprint on a chemical intermediate

The thresholds in that table are the fictional company's choices. No rule sets a maximum footprint age or a minimum primary data share for a buyer. Published ranges are wide: 50 to 1,000 would pass almost anything, so your own suppliers' range on similar routes does most of the work. That range grows more useful every year you collect.

Common Scope 3 data errors: the wrong year, cradle-to-grave figures, double counted energy and reductions that were only method

In shortThe errors repeat. Old footprints, wrong boundaries, economic allocation, company totals with nothing to divide by, and unit mix-ups are what the checks exist to catch. The most damaging one is different in kind: a method change reported as a cut. We found no enforcement case on this exact point, and no public rate for any of these errors, so measure your own.

From 1,860 suppliers to 168 usable answers: where supplier data falls away

Coverage narrows at each step, and each step is a place to act. Watch the gap between the third bar and the fourth. A supplier that sent something unusable has already made an effort, and a precise question can turn it into usable data.

Public evidence that supplier emissions disclosure is still thin

CDP looked at more than 18,700 companies' 2022 responses and found nearly 60% reported on no supply chain category at all.32 That data is old, so read it as background. BCG and EcoVadis, writing in September 2025 about companies on EcoVadis's platform, report 24% disclosing upstream Scope 3 and 8% with a target to reduce it.33 A 2025 benchmark for automated extraction found about half of the sustainability reports it studied held no usable greenhouse gas data, with other indirect emissions "rarely complete".34

Why pharma suppliers are hearing about Scope 3 from health systems, big drug makers and auditors at the same time

In shortThe pressure comes from three sides. Health systems want lower supply chain emissions. Large drug makers have written to their suppliers with shared targets for 2027, 2028 and 2030. And auditors now sample these figures. The sector's footprint is large enough to attract attention, and a poor supplier data set blocks all three conversations at once.

The health-care buyers' targets, as of the November 2025 letter

In November 2025, procurement chiefs of the Sustainable Markets Initiative's Health Systems Task Force, large drug and health care companies, updated the targets their chief executives first set in 2023. They say the sector is about 5% of global emissions, more than half from the supply chain.13 NHS England, England's public health system and a sponsor of PAS 2090, reports 62% of its own emissions from its supply chain, over a quarter linked to medicines and prescribing.12,36

The initiative's supplier toolkit says individual customers may still ask for products beyond the top 10, and that a request "may replace existing product-level data requirements".28 PAS 2090:2025, published November 30, 2025, sets product category rules for pharmaceutical life cycle assessments.12 Medicine makers in the UK should also watch NHS England's plan for product-level footprints from 2028. A draft market survey, out for testing in June 2026, asks suppliers how they collect and report this data, and says life cycle assessments will not be needed for every product.37

How few companies disclose upstream Scope 3, and how few of those are on track

BCG and EcoVadis report that partnering with suppliers makes companies "nine times more likely" to set and deliver a Scope 3 target, and that upstream emissions average 21 times Scopes 1 and 2.33 Read the first as a correlation: companies that engage suppliers may be further along in other ways.

Reading supplier PDFs with AI: what automation does well in Scope 3 work, what no regulator has written, and what stays with people

In shortReading documents, matching purchase lines and running rule checks suit software. Judging a borderline footprint, deciding what to tell a supplier and signing the figure do not. No regulator has written an AI rule for Scope 3 work. Most AI rules aimed at drug makers govern other work, so the practical test is the assurance provider's question: show me the method and the source.

What automated reading gets wrong, and the controls that catch it

The authors of a 2025 benchmark on greenhouse gas extraction warn that users can place full trust in model output and overlook measurement errors that occur often.34 We found no independent error rate for reading supplier footprints specifically.

Four controls follow from that. Every value carries its page, so a person can check in seconds. A missing value stays missing, so nothing gets invented. The rule checks are plain rules with versioned thresholds, so an auditor can re-run them. And a person decides every swap above a size you set, with a record of who and why.

Who runs Scope 3 supplier engagement: sustainability, procurement and the auditor, and where their working days go

In shortA small sustainability team owns the number and procurement owns the suppliers. The hand-offs between them are where time disappears. Reading and checking take the largest share of the days, which is why many programs check a sample. Teams run it in house, with a consultancy, on a software platform, through a sector program, or by some mix of these.

The five roles behind a signed Scope 3 figure

How a supplier submission moves from inbox to inventory

Where a two-person team's 50 working days go on 240 suppliers

No one publishes this split. These are illustrative figures, an estimate for a cycle done by hand. Replace them with your own.

In-house spreadsheets, a consultancy, carbon accounting software or a sector program: how teams staff this work

From a 410 kg estimate to a verified 182 kg footprint: Kestrel's VT-114 line, read, checked, confirmed and signed

In shortThis is the example from the top of the page, inside the product. The agents read the footprint, find the purchase line and run the checks in view. Nadia Brenner, the Head of Sustainability, looks at the evidence and confirms. The estimate turns into supplier data in place, the change is filed as a method change, and she signs the category before the auditors arrive.

1

The home screen asks how much of the Scope 3 number is real

Nadia's home screen is the FY2025 inventory: 459,400 tonnes across ten categories, 59.8% on supplier data, against 18% in FY2023. Each category is one bar in four layers: dark green for supplier product data, lighter green for supplier company data, blue-gray for average data and hatched gray for spend estimates.

On the right, five supplier submissions have arrived. Kestrel's is first, and it shows the gap at a glance: 410 in the estimate, 182 in the footprint.

2

The agents read Kestrel's file and run the checks in view

Nadia presses "Check and swap in". The supplier data reader pulls 182 kg of CO2e per kg, cradle-to-gate, January to December 2025, mass allocation. The factor matcher finds the line: 17.1 tonnes bought for €38.6 million, now an estimate of 7,011 tonnes. The plausibility checker runs through its tests, step by step, each with its result.

One note appears, on biogenic carbon, which is not material for a synthetic route. The figure passes, and because the line is above the 250-tonne limit for automatic acceptance, it stops and waits for a person.

3

Every value and check points at a page in Kestrel's document

Before confirming, Nadia opens the evidence. On the left is the footprint itself, with each value highlighted. On the right, the effect of the swap, the eight checks and the values read. Move to a check and its source lights up: process mass intensity of 210 kg per kg, 78% of solvent recovered on site, 62% of electricity from a solar contract, limited assurance dated October 2, 2026.

Rahul Menon, who owns this supplier, sees the same screen, and can send a question from it.

4

Confirmed: the estimate's gray slice turns dark green and Category 1 falls 3,900 tonnes

Nadia confirms. Kestrel's 7,011-tonne estimate becomes 3,112 tonnes in the product data layer. Category 1 goes from 316,300 to 312,400 tonnes, Scope 3 to 455,500 tonnes, and supplier data from 59.8% to 61.0%.

A note appears under the bar: supplier PCF, PACT-aligned, filed as a method change. It will show on the Progress screen apart from real change, and it never counts toward the reduction target.

5

Nadia signs Category 1 at 312,400 tonnes for the sustainability statement

The other nine categories were approved on October 10. With Kestrel's line settled, she reviews the table and signs with her password. The record keeps her name, the time and the meaning of the signature: "I approve these figures for the FY2025 sustainability statement".

The figures lock, and the assurance file can go to Wexcombe & Pryce, the fictional assurance provider, read-only for 60 days.

When a supplier's footprint fails: a 2022 vial figure held, a company total with no key, and the question each one earns

In shortThe submissions that fail show the checks at work. A submission that fails a blocking check is held, the estimate stays, and a question is drafted that names what failed and cites the page. A category manager reads it, edits it and sends it under their own name.

Category drill-downs, the 240-supplier list, real against method change, the disclosure table and the settings behind them

In shortEach screen answers one question a person in the cycle asks. Which lines make up this category? Who has answered? What really changed since last year? What goes in the statement, and what stands behind each figure? Who decides, and on what thresholds?

Purchased goods, capital goods and freight, line by line

Real change, method change and the ESRS disclosure, each on its own screen

Your thresholds, factor sets and cycle calendar, set by you

The Head of Sustainability signs every Scope 3 category; five agents prepare what she signs

In shortEach agent does one job and has one thing it may never do. Nothing goes to a supplier without a person, no figure that fails a blocking check gets used, and a method change never counts as a reduction. Roles, approvals and the audit trail are built in, for an enterprise supplier base and an external assurance review.

The five agents behind a supplier data cycle

Who confirms a swap, who sends a question, who signs the Scope 3 figures

The people in Veltara's fictional Scope 3 cycle

Supplier files and purchase lines going in, a signed Scope 3 table and assurance file coming out

Systems a Scope 3 rollout connects to

Design targets for the sample program, and the controls behind them

These are labeled design targets for the sample program, to be measured in your own run. They are goals and carry no promise for your data.

Botminds is SOC 2 Type II audited, with information security managed to ISO 27001. Supplier documents are kept for 10 years in the sample's settings. Records and approvals support your assurance process; the responsibility for the statement stays with your company.

The price of spend estimates and unchecked supplier data: analyst days, audit samples and progress you cannot show

In shortThe hours are the part you can price, and they are modest. The larger costs do not fit a calculator. You cannot show a supplier's real cuts. An auditor's sample can turn into a late rework. A customer asks for the footprint you do not have. Start by measuring the few things below, which takes about a week.

Timing ten supplier files and tracing ten Scope 3 figures to find your baseline

A supplier's real cut you cannot show, and other costs no Scope 3 calculator counts

What automating supplier data checks returns: analyst days and audit hours against the tool's cost, for your supplier count

In shortPut in your own numbers. Every default is an assumption to replace. On the defaults, hours alone do not pay for a $100,000 yearly tool at 240 suppliers; they begin to at about 645. For most drug makers the case rests on checking every submission, a trail that holds at assurance, and being able to show suppliers' real cuts. A small supplier list may do well with a better spreadsheet.

Analyst days, audit samples, full coverage: where Scope 3 automation can pay back

Suppliers, analyst days and audit samples: your Scope 3 cycle in numbers

How ready your team is for supplier data automation: purchase lines, acceptance rules, a base year policy and a named signer

In shortEight questions, scored 0 to 2, and one for reflection. A low score usually means writing down your acceptance rules and base year policy before buying anything. Your answers stay on this page.

A low score means writing your acceptance rules before anything else

What to ask Scope 3 software about footprints, allocation, the value-chain cap and method change, and a 14-week test

In shortAsk any vendor, including us, to show a held submission as readily as a passed one. Ask how it treats a figure below the estimate and a request to a protected supplier. Then replay last cycle's files against your own decisions, run one batch in parallel, and decide on four numbers.

Questions for any Scope 3 vendor, including us

Replay last cycle's supplier files, then run one batch in parallel

1. Choose 30 submissions from last cycle: mixed types, some you accepted and some you held, with your decisions recorded.
2. Write the rules: age, boundary, allocation, primary share, verification, gap, and who confirms.
3. Connect a purchase extract: one category, with quantities, and your approved factor sets.
4. Replay and score: compare values read, decisions and questions with your own record.
5. Decide on four numbers: extraction errors, decisions that differ, days per 100 suppliers, and minutes to trace a figure.

Reading errors, differing decisions and trace time: how to score a Scope 3 pilot

Supplier Scope 3 questions answered: primary data, spend-based estimates, ESRS E1, PACT, VSME and restatements

In shortDirect answers to what people search for, each one standing on its own. Status statements are as of November 3, 2026.