What Data Manufacturers Need to Collect for an EPD
Author:
Luke Boivin

Know what to collect for your EPD, who should provide it and how to keep the evidence ready for review. For manufacturers supplying data-center projects, that means bringing product, facility, supplier and logistics records together before a customer deadline turns missing information into urgent rework.
A good data request starts with the applicable Product Category Rule. The PCR defines the declared or functional unit, life-cycle stages, data-quality rules and reporting requirements. The checklist below covers common manufacturing inputs, but the PCR and program operator determine the final scope.
What Goes Into an EPD Data Collection Checklist?
Before requesting spreadsheets, write down the scope.
Confirm:
The product or product family
The configurations, sizes and materials included
The manufacturing facilities represented
The reporting period
The declared or functional unit
The applicable PCR and program operator
The life-cycle stages required
Any intended grouping, averaging or scaling approach
This prevents teams from collecting data that cannot be matched to the product boundary. It also exposes decisions that need input from the LCA practitioner or program operator before the request spreads across the company.
The first data-quality control is a clear scope. A precise spreadsheet cannot repair a product boundary that was never settled
Product Data Connects the Catalog to the Model
The product team usually holds the information that defines what is being assessed.
Common inputs include:
Product names, model numbers and descriptions
Bills of materials with quantities and units
Component masses and material specifications
Size, capacity, rating or performance variants
Packaging materials and quantities
Drawings, specification sheets or product data sheets
Production volumes for the reporting period
Rules connecting product configurations to component quantities
For product families, avoid sending only one representative bill of materials without explaining how the rest of the family changes. The practitioner needs the relationship between configuration and input, such as component mass by size or a material substitution by option.
Check that totals reconcile. The bill of materials, product mass and packaging records should either match or carry a documented explanation for the difference.
Facility Data Describes How the Product Is Made
Operations and facility teams supply the primary data behind manufacturing.
Depending on the PCR and system boundary, the request may include:
Electricity and fuel consumption
On-site renewable generation
Process-specific energy where it is available
Water use
Production output
Scrap, yield and rework
Waste quantities and treatment routes
Direct process emissions
Refrigerants or other relevant releases
Shared-facility allocation information
Maintenance or consumable materials that meet the inclusion rules
The reporting period matters. Record the dates covered by every source and explain any mismatch. A facility total from one year and production volume from another can create a confident-looking but unreliable intensity.
Also document the source. A utility invoice, meter export and manually prepared estimate are not equivalent evidence. Each may be usable, but the model should preserve what the value came from.
Supplier Data Explains Upstream Materials and Components
Procurement often owns the information that is hardest to retrieve late in the project.
Useful records include:
Supplier and manufacturing location
Material or component specification
Supplier-specific EPD or product carbon footprint
Recycled-content declaration
Certification or test documentation
Purchased quantity and unit
Country or region of origin
Supplier data period and methodology
Known changes during the reporting period
Not every supplier will provide primary environmental data. When secondary datasets or proxies are required, document the selection and why it represents the purchased material. A verifier needs to understand the choice, not just see a database name.
Keep supplier evidence connected to the exact material line in the bill of materials. A folder full of PDFs is not yet a usable data system if nobody can tell which document supports which input.
Transport Data Needs Origins, Distances and Modes
Transport is frequently estimated too loosely.
For inbound materials and components, gather:
Origin and destination
Distance
Transport mode
Load or shipment information when available
Material mass
Any warehousing or intermediate stops required by scope
Use actual routes or supplier records where practical. If distance must be estimated, record the method, such as mapped road distance between named locations. Avoid unlabeled round numbers that cannot be reproduced later.
The product’s outbound transport, installation, use and end-of-life data may also be required, depending on the PCR. Treat those as scope-specific work rather than assuming every EPD stops at the factory gate.
Waste and Co-Products Need Clear Destinations
Waste data is more than a total mass.
For each relevant stream, capture:
Material type
Quantity
Hazardous or non-hazardous classification
Recycling, recovery, treatment or disposal route
Transport to treatment when required
Evidence source
Whether the material is a waste, by-product or co-product
The classification can affect allocation and end-of-life modeling. Flag ambiguous streams early rather than choosing a treatment silently.
Organize Evidence for Review, Not Just Calculation
UL Solutions recommends organizing supporting documents so records can be located when verification questions arise. A practical evidence structure connects each model input to:
A stable input name
Value and unit
Product, process or facility scope
Reporting period
Source owner
Source file or URL
Collection date
Calculation or conversion applied
Assumption or limitation
Approval or review status
Use consistent filenames and preserve original files. If a value is converted, keep both the original quantity and the conversion. If several sources were combined, record the formula.
This structure makes internal QA easier before verification begins. It also prevents the LCA report from becoming the only place where an important assumption is explained.
Assign Owners Across the Company
EPD data collection is cross-functional.
A typical responsibility map looks like this:
Sustainability or product compliance: scope, PCR coordination and overall evidence tracking
Product and engineering: product definitions, bills of materials and configuration rules
Operations: facility energy, output, yield, waste and process data
Procurement: suppliers, materials, recycled content and upstream evidence
Logistics: routes, modes, distances and shipment information
Finance or data owners: reconciliations, reporting-period controls and source access
LCA practitioner: data template, modeling decisions, gaps and QA
Verifier: independent review after the LCA and draft EPD are prepared
Name one owner for each requested field and one person responsible for resolving conflicts between sources.
Run a Readiness Check Before Modeling
Before the model is finalized, confirm:
The product boundary and reporting period are settled.
Units and mass balances reconcile.
Facility totals connect to production output.
Every important input has a source or a documented proxy.
Supplier and transport assumptions are reproducible.
Allocation choices are identified for practitioner review.
Files are accessible and linked to the values they support.
Known gaps have an owner and next action.
A complete-looking workbook can still fail this check. Readiness means the evidence is traceable and internally consistent, not that every cell is filled.
Use the Checklist for a First Data Request
Start with one priority product family and one reporting period. Send each team only the fields it owns, with the expected units and a request for the source record. Agree how gaps will be flagged before the files return.
For an illustrative cooling-equipment family, engineering might own component masses and capacity options, operations the facility energy and production totals, procurement the major component sources, and logistics the inbound routes. A missing component mass is a different problem from a missing supplier-specific environmental dataset. Record them separately so the practitioner can choose the appropriate next step.
Keep three states visible: supplied and checked, supplied but needing clarification, and not available. A blank cell should not silently become zero. Where an estimate is needed, retain its owner, method and limitation alongside the value.
Make the Next Update Easier Than the First
The best EPD data collection process leaves a reusable evidence system behind. When a supplier, facility process or product configuration changes, the team should be able to identify the affected inputs and update the model without reconstructing the project history.
CarbonGraph helps manufacturers organize source evidence, structure reusable product-family LCA models and preserve the assumptions behind verification-ready outputs.
Want to review the data you already have and identify the gaps? Book a walkthrough.