CarbonSpec / Changes
What was specified, what was installed
The carbon consequence of a substitution is arithmetic anybody can do. Substitutions go wrong for a different reason: the change happened on site, in a week, between a subcontractor and a site manager, and nobody wrote down what it replaced or who approved it. By the time anyone asks, the specified product is in a superseded revision and the installed one is on a delivery note in a skip.
So this is a ledger rather than a calculator. An entry missing an approver or its evidence is not a lesser entry — it is an open one, and that difference is the whole point of a golden thread.
Substitutions
location, specified, installed, reason, approved by, date, evidence, carbon delta low kg, carbon delta high kg, cost delta £, O&M reference — leave any blank and it is reported as missing rather than assumed
The record
4 substitutions: 1 complete, 3 open. Net carbon effect of the recorded changes: -35.2 t to -24.8 t, across the entries that carry a figure — 2 do not, and those are missing from this total rather than zero in it.
At a gateway: 3 of 4 entries are missing an approver, evidence, or both. A pack assembled now would present those gaps as though they were the record. They are not: they are the parts of the record that were never made, and closing them is faster now than after handover — the people who know are still on site.
Fire: One substitution replaces a product with one whose class typically classifies worse for reaction to fire. Class figures are what a class usually achieves, not a classification — each needs the installed product's own evidence and a fire engineer's decision.
L3 east facade — Aluminium rainscreen A → Aluminium rainscreen B
closedLead time
Approved by A. Engineer on 2026-05-04, with evidence on file.
Fire, before carbon: Both products sit in the Aluminium products class, which typically classifies A1. That is a class-typical comparison and not a clearance: the installed product's own test evidence is still required.
Carbon -4200 to -6800 kg · Cost £-9,000
Core walls — C40/50 CEM I → C40/50 CEM III/A
openCarbon reduction
Recorded but open. Missing: Who approved it, Evidence for the installed product, Operation and maintenance information. This is not a lesser entry than a closed one — it is the entry somebody will have to reconstruct from memory later, and the reconstruction is what a golden thread exists to make unnecessary.
Fire, before carbon: Neither product could be placed in a material class, so nothing here can compare their reaction to fire. The installed product's own classification is what the record needs.
Carbon -31000 to -18000 kg · Cost +£2,500
Roof — Membrane spec A → Membrane spec B
openAvailability
Recorded but open. Missing: Who approved it. This is not a lesser entry than a closed one — it is the entry somebody will have to reconstruct from memory later, and the reconstruction is what a golden thread exists to make unnecessary.
Fire, before carbon: Neither product could be placed in a material class, so nothing here can compare their reaction to fire. The installed product's own classification is what the record needs.
Party wall cavity — Mineral wool slab → PIR insulation board
openThinner build-up
Recorded but open. Missing: Who approved it, Evidence for the installed product, Operation and maintenance information. This is not a lesser entry than a closed one — it is the entry somebody will have to reconstruct from memory later, and the reconstruction is what a golden thread exists to make unnecessary.
Fire, before carbon: The PIR and PUR rigid foam class typically classifies E for reaction to fire, against A1 for Mineral wool insulation. Hold this swap for a fire engineer's decision before it is accepted. A particular product can test better than its class, so this is a reason to ask, not a refusal — but a lower carbon figure does not answer it.
What each change cost per tonne avoided
1 swap pays for itself and is not an abatement cost at all — do those on the cost case and keep them out of the carbon budget, where they would compete with real trades. Ranked by the WORST case each swap could cost per tonne, not the midpoint. A midpoint is the one value in a carbon band with no claim to being right, and sorting on the optimistic end promotes the swap with the widest band rather than the cheapest one.
L3 east facade: Aluminium rainscreen A → Aluminium rainscreen B
Cheaper as well as lower-carbon — £9,000 saved, not spent. This is not an abatement cost at all and should be done on the cost case alone; putting it in a carbon budget makes it compete with things that are actually a trade.
Core walls: C40/50 CEM I → C40/50 CEM III/A
£81 to £139 per tonne avoided, depending where in the carbon band the truth falls. Decide on the upper figure: if the swap is worth it there, it is worth it everywhere.
What this does not decide
Nothing here says a substitution was acceptable, met the specification, or satisfies a gateway. Those are judgements made by named competent people against a product’s own evidence, and software that returned “compliant” would be offering to make them.
Nor can it see a substitution nobody logged. A complete-looking record says the record is complete as recorded; whether it is complete as to what happened on site is a different question, and no software can answer it.