Claude for Climate | Episode 3 of 7
Why suppliers don’t answer emissions data requests, and five Claude prompt patterns that change the economics of asking.
Rob Aldrich · Technology & Sustainability Executive · Sydney, Australia
robaldrich.com · LinkedIn · @googlenut
In Episode 2, I walked through why Scope 3 data is fragmented across procurement systems, utility bills, and supplier spreadsheets. This episode is about the harder half of that problem: most of your Scope 3 data isn’t in your systems at all. It’s in your suppliers’ systems, and they’re under no obligation to hand it over.
The numbers frame it starkly. Scope 3 emissions account for around 75% of a typical company’s footprint, and in sectors like financial services and capital goods the figure regularly exceeds 90%, according to CDP. Yet when companies send supplier data requests, many receive responses from fewer than a quarter of their suppliers. In MIT’s 2025 State of Supply Chain Sustainability report, which surveyed more than 1,200 professionals across 97 countries, roughly 70% of respondents named lack of supplier data as the single biggest barrier to tracking Scope 3.
So the largest slice of your footprint depends on the cooperation of companies you don’t control, and three quarters of them aren’t cooperating.
Why suppliers don’t respond
It’s tempting to read low response rates as apathy. It’s mostly economics.
The incentives point the wrong way. From a supplier’s perspective, your data request is unpaid work with unclear benefits. Their sustainability team, if one exists, answers to their customers’ deadlines only when a contract depends on it.
Survey fatigue is real and getting worse. A mid-sized manufacturer selling to twenty large customers can receive twenty different questionnaires asking overlapping questions in incompatible formats. The EPA has explicitly flagged this, encouraging shared questionnaires precisely because duplicated surveys are suppressing response rates across the board. CDP’s Supply Chain program exists partly to solve this: more than 380 buyer members now request data from over 24,000 suppliers a year through one standardised questionnaire.
Most suppliers can’t answer even if they want to. Only 38% of businesses are currently measuring their Scope 3 footprint at all, according to an EcoVadis and IBM-commissioned survey. Many smaller suppliers have never calculated Scope 1 and 2, let alone Scope 3. When they do respond, the data often arrives as manual estimates in a spreadsheet compiled by someone with no GHG accounting training. MIT’s survey found 66% of companies still track Scope 3 in spreadsheets.
And when data does arrive, it doesn’t match. Different system boundaries, different emission factors, different base years. The divergence problem runs so deep that one academic study found the correlation between aggregated Scope 3 estimates from ISS and Trucost, two of the largest data providers, was just 16%. If the professionals can’t agree, expect your suppliers’ self-reported numbers to disagree too.
I’ve lived both sides of this. A good example of how much small choices matter came from a supplier assessment during my time at AWS. A data centre team was planning an extension and evaluating concrete suppliers, one in Virginia, one in Oregon. Jump on Electricity Maps and you can see the difference in grid emissions between those regions in real time. The cost was the same. The lead time was the same. The emissions powering each plant were not. The suppliers who were diligent in providing that data benefited from it, simply because they chose to include the detail.
At Cisco, in the early days of what was then called “Green IT,” getting anything from suppliers was a grind. Across more than 400 suppliers, fewer than 30% provided what we considered credible data. That forced us into making assumptions from historical data, and every assumption is a small erosion of the number you eventually publish.
It’s worth naming what filling those gaps costs. In my experience running sustainability functions, roughly 85% of the budget went into producing a single PDF: the annual report. Under the hood, that money paid for data analysts who spent all their time normalising and massaging inputs. It put the whole team on a treadmill. Ten months crunching data, a sprint with marketing to present it properly, submit, breathe, and start again in two or three months. That treadmill is why so many sustainability leaders struggle to get proactive. Claude changes that equation, handing back budget and time that can go toward actually leading change rather than tallying up what the rest of the business is doing. Comment below if you’d like a deep dive on that shift.
What engagement looks like when it works
There’s evidence the flywheel turns once suppliers start disclosing. CDP’s data shows 26% of first-time supplier respondents report setting climate targets, rising to 57% among repeat respondents. Suppliers engaged through CDP Supply Chain members reported saving 70 million tonnes of CO2e attributable to that engagement. Disclosure isn’t just measurement. It’s the front door to reduction.
The problem has never been whether supplier engagement works. It’s that doing it well is labour-intensive: segmenting hundreds of suppliers, tailoring requests to each one’s maturity, interpreting whatever comes back, and chasing the gaps. That’s exactly the kind of work most sustainability teams, typically two or three people, cannot do at scale.
This is where Claude changes the cost structure.
The Claude workflow: five prompt patterns
The pattern from Episodes 1 and 2 holds: Claude is useful not because it knows your suppliers, but because it can reason across frameworks, adapt tone and technical depth per audience, and process inconsistent inputs at volume. Here are the five patterns I use.
1. Segment before you send
Give Claude your supplier list with spend data and sector codes, and ask it to apply the GHG Protocol’s prioritisation logic: rank by estimated emissions materiality, not just spend, and propose an engagement tier for each. Your top 20 suppliers by emissions get direct, personalised outreach. The long tail gets a lighter standardised request or stays on spend-based estimates for now.
Prompt pattern: “Here’s my supplier list with annual spend and industry classification. Using spend-based emission factors, estimate each supplier’s share of my Category 1 emissions, rank them, and propose three engagement tiers with a recommended data request depth for each tier.”
2. Draft requests matched to supplier maturity
A steel producer with a published CDP score needs a different letter than a 40-person components supplier who’s never heard of Scope 2. Claude can generate both from one brief, adjusting technical vocabulary, explaining why you’re asking, and connecting the request to the commercial relationship.
Prompt pattern: “Draft two versions of a supplier emissions data request. Version A is for a supplier with an existing CDP disclosure: request their latest verified Scope 1 and 2 figures and product-level intensity data. Version B is for a small supplier with no measurement history: request only the activity data they already hold (fuel bills, electricity consumption, freight records) and explain in plain language how we’ll use it. Both versions should state our reduction target and why their data matters to it.”
That last element matters more than it looks. Specific, goal-connected requests get responses. Generic compliance requests get deleted.
3. Normalise what comes back
Responses arrive with mismatched boundaries, factors, and years. Claude can read each response, identify the methodology used, flag where it deviates from your inventory’s basis, and convert what’s convertible while marking what isn’t.
Prompt pattern: “Here are twelve supplier responses. For each: identify the reporting boundary, emission factor source, and base year. Flag any that use market-based Scope 2 where we require location-based. Produce a table of what can be incorporated directly, what needs conversion, and what needs a follow-up question, with the follow-up question drafted.”
4. Chase gaps without burning relationships
Follow-up is where response rates are won or lost, and it’s the first task teams drop when stretched. Claude can draft polite, specific follow-ups referencing exactly what’s missing from each supplier’s response, rather than re-sending the whole questionnaire.
5. Build supplier-facing guidance
The highest-leverage move is making it easier for suppliers to answer. Claude can turn your data request into a two-page plain-language guide per supplier segment: what each term means, where the data lives in their business, and a worked example for their sector. Companies that share resources with suppliers see measurably better data quality.
The limitations, stated plainly
As with every episode in this series, here is what Claude cannot do, because pretending otherwise would defeat the purpose.
Claude can’t fix the incentive problem. If a supplier has no commercial reason to respond, a better-written email won’t create one. Procurement leverage, contract clauses, and scoring suppliers on ESG alongside price do that. Claude scales the communication; it doesn’t supply the leverage.
Faster outreach can mean faster garbage. If your suppliers’ underlying numbers are untrained spreadsheet estimates, Claude will help you collect bad data more efficiently. Normalising inconsistent responses is not the same as verifying them, and Claude cannot audit a supplier’s meter readings.
Auditors will ask for provenance. Under CSRD, California’s SB 253, and Australia’s mandatory regime, Scope 3 figures need documented lineage. A Claude-assisted workflow must log what was requested, what was received, and what transformations were applied. That’s a process discipline, not a model capability.
Survey fatigue is a systemic problem Claude can make worse. If every buyer uses AI to send more, better-personalised questionnaires, suppliers drown faster. The genuinely sustainable answer is fewer, shared, standardised requests. Use Claude to make your asks lighter, not just more numerous.
The bottom line
Supplier engagement has always worked. It’s just been too expensive to do properly for anyone without a dedicated team. Claude doesn’t change whether suppliers should respond. It changes how much it costs you to ask well, interpret honestly, and follow up relentlessly. That used to be a headcount problem. Now it’s a workflow problem.
Next episode gets uncomfortable. Amazon’s carbon emissions are rising, driven by the AI data centre buildout, and the company’s defence rests on exactly the metric-selection problem I covered in Episode 1. I spent years inside Amazon’s sustainable buildings practice. Episode 4 is about what happens when the technology we’re using to solve climate reporting is also the technology driving emissions up.
Rob Aldrich has spent 25 years at the intersection of enterprise technology and sustainability. He is the co-creator of Cisco EnergyWise, former Global Sustainable Buildings Lead at Amazon Web Services, and former interim Chief AI Officer at SafetyCulture. He is the author of IP-Enabled Energy Management (Wiley/IEEE Press) and has conducted sustainability audits across more than 70 data centres worldwide.
robaldrich.com | linkedin.com/in/rob-aldrich | @googlenut
Sources
- Normative, “Scope 3 Supplier Engagement: Primary Carbon Data” – normative.io
- Zevero, “The Key Role of Supplier Engagement in Scope 3 Reporting” – zevero.earth
- MIT Sloan / MIT Center for Transportation & Logistics, “2025 State of Supply Chain Sustainability Report” – mitsloan.mit.edu
- Environment+Energy Leader, “Scope 3 Data Depends on Suppliers. Most Suppliers Are Not Ready.” – environmentenergyleader.com
- Dcycle, “CDP Supply Chain program: how buyers use it to engage suppliers” – dcycle.io
- CO2 AI, “Call to action: Product-level data is crucial to address Scope 3 decarbonization” – co2ai.com
- PLOS Climate, “Scope 3 emissions: Data quality and machine learning prediction accuracy” – journals.plos.org
- GHG Protocol, “Technical Guidance for Calculating Scope 3 Emissions” – ghgprotocol.org
- US EPA, “Questionnaire for Suppliers on Energy & Greenhouse Gas” – epa.gov