Why “almost identical” still matters
Tiny recipe shifts can nudge climate-change potential by two to eight percent, enough to push a product over a regulatory line in markets that reward the lowest-carbon option. Buyers compare EPDs side by side, so fuzzy averaging can cost real sales.
The 10 percent spread rule
Current PCR 2019:14 version 2.0.0 lets you group variants only when every impact indicator varies by less than ten percent (PCR 2019:14 v2.0.0, 2025). Break that limit and you must pick either a worst-case number or a fully parameterised model.
One average number: fastest route, limited flex
When variants fall inside the 10 percent band, declare the statistically weighted average. Verifiers like it because documentation is short and transparent, and sales teams like it because one PDF covers an entire family. Double-check that the average is based on current 12-month production data, not last year’s pipeline estimates.
Worst-case declaration: insurance for regulated bids
If one or two indicators poke above the ten-percent ceiling, declare the worst result across all variants. Yes, the numbers look higher, but you sidestep extra modelling and stay eligible for public tenders that ban mid-year revisions. The new General Programme Instructions 5.0.0 label this the "maximum impact" method and warn that it must be justified on the EPD cover page (GPI 5.0.0, 2024).
Want to speed up your EPD creation process?
Follow us on LinkedIn for insights on optimizing your EPD strategy to win more projects and unlock market opportunities.
Parametric variant EPDs: power when catalogs explode
An alternative is to build a single core LCA model and let software auto-update impacts as you toggle inputs. Smart EPD and IBU both accept parameterised files as long as every variant lands its own verifier-approved PDF (Smart EPD technical note, 2024). Up-front cost rises, but you gain the ability to issue fresh EPDs in days when a customer requests a bespoke length, coating, or add-on.
Allocation rules that rarely fail
- Mass allocation beats economic allocation for internal processes because market prices swing wider than material flow data.
- Keep recycled content allocation consistent across all variants. Mixing “cut-off” and “credit” methods in one family triggers red flags at verification.
- Document any by-product energy allocation choices. Auditors increasingly ask for this thanks to the EN 15804+A2 focus on net-zero targets (EN 15804+A2, 2024).
Data hygiene tips
Rounding can hide a three-percent difference in eutrophication potential. Always store cradle-to-gate results with at least four significant figures, even if the published table shows only two. Automatic export templates in most LCA tools respect this, but manual Excel copies do not, we learnt that the hard way.
Dollars and sense
A single representative EPD can cost roughly thirty percent less to produce than three individual ones according to a 2024 survey of program operators (EPD Australasia, 2024). Yet the bigger gain is speed: average-or worst-case grouping cuts scheduling time by four to six weeks, which can lock in spec wins before a competitor even hires a consultant. That comercial edge often outweighs the minor precision sacrifice.
Aim for clarity, not clutter
Choose the simplest option that keeps every stakeholder—auditor, sales lead, and carbon-curious architect—confident in the numbers. A clear rule-based approach now will save headaches and maybe a few grey hairs during the next PCR update. Remember, envionmental declarations sell only when people trust them.
