Ten years ago, proposing to retain an unremarkable industrial building required an argument. The default assumption was demolition, and retention was a position you had to justify — usually on heritage grounds, which meant the building had to be special enough to defend.

That has reversed, and quickly. Carbon accounting did most of it, though rising demolition and disposal costs did more than anyone admits. The result is that we are now regularly asked to keep buildings nobody would have called significant: sheds, warehouses, an office block from 1978. Not because they are good, but because they exist, and existing turns out to be worth a great deal.

This is a better default. It also introduces a set of questions the heritage tradition never had to answer, because heritage work concerned buildings whose value was agreed in advance.

The building is not the question

The first mistake is to assess an existing building on its architectural merit. Most of them do not have any, and the assessment then becomes a slow way of arriving at demolition.

The useful assessment is structural and geometric. Is the frame sound and what will it carry? What are the floor-to-floor heights, and do they permit the services the new use requires? What is the plan depth, and can daylight reach the middle of it? Is the grid compatible with anything anyone wants to do now?

A mediocre building with 3.6-metre floor-to-floors, a 7.2-metre grid and a shallow plan is worth keeping. A well-composed building with 2.4-metre heights and a forty-metre depth may not be, however much anyone likes the elevation. This is an unromantic way to look at it and it produces better decisions than any judgement about quality.

The one thing that usually settles it: substructure. Foundations and basements are the most carbon-intensive and expensive part of any building, and they are almost always reusable even when nothing above ground is. If the answer to everything else is no, check the ground before concluding.

How much to keep is a separate decision

Retention is not binary, and treating it as such is the second mistake.

There is a spectrum from full retention with minimal intervention, through structural retention with a new envelope, to keeping the frame and substructure only, to salvage — dismantling for materials while accepting the building goes. Each is a legitimate answer and they carry very different costs and carbon outcomes.

At a row of dock sheds in Rotterdam we did almost the minimum: repaired the roofs, made them weathertight, left the interiors entirely unfinished, and reused the brick from three collapsed bays to patch the ones that stood. The embodied carbon came in at 96 kgCO₂e/m², the lowest figure we have recorded. That was not achieved by clever specification. It was achieved by not doing very much.

The money that would have gone into upgrading the sheds went into the yard outside instead, which is what the tenants actually use. This is the trade that reuse projects most often get wrong — spending the retention saving on bringing the retained building up to a standard nobody asked for, rather than on the parts of the project that determine whether it works.

Legibility, not disguise

When new work meets old, it should be identifiable as new. This is the one position I would defend without qualification.

Not shouting — the contrast does not need to be aggressive, and the fashion for deliberately jarring insertions has produced some tiresome buildings. But not pretending either. Matching brick, replicating a moulding, continuing a rhythm so the join disappears: this removes information from the building permanently, and it is done almost entirely for the comfort of the person approving it.

The practical test is reversibility. At a boathouse in Copenhagen, everything we inserted is a freestanding steel and glass structure that touches no historic surface; the gap between the two is the circulation route. Somebody could remove all of it in a fortnight and find the building we started with. That is a harder discipline than it sounds, because services want to be chased into walls and fixings want to go where they are convenient, and holding the line costs money in details.

The parts that are genuinely harder

Reuse projects carry risks that new-build does not, and pretending otherwise sets clients up badly.

Survey information is incomplete by definition. You will open something up during construction and find a condition nobody anticipated, and the contingency needs to reflect that rather than the ten percent that would be normal on a new build. We advise fifteen to twenty on anything involving significant existing fabric, and we have been glad of it more often than not.

Performance is a compromise. Internal insulation on a retained facade means accepting cold-bridging at slab edges, which a new build would not tolerate. At a Rotterdam silo we made exactly that trade to keep the exterior untouched, and the fabric performance is measurably worse than the new-build alternative would have been. It remains overwhelmingly the right decision on carbon, but it is a trade, and the modelling should say so honestly rather than quietly using optimistic assumptions.

Programme is less predictable. Discovery drives change, change drives delay, and the sequence cannot be compressed the way a new build's can.

The reason it is still worth it

None of those difficulties outweighs the arithmetic. Retaining the structure at that Rotterdam silo avoided around 4,100 tonnes of embodied carbon — more than the building will emit in operation across its expected life. No improvement in the fabric of the demolish-and-rebuild option would have closed a gap of that size.

The remaining objection is usually about yield: the conversion delivered twenty-six fewer apartments than the consented new-build scheme. That is a real cost and I would not dismiss it. But it was offset by removing demolition and ground remediation from the programme entirely, which saved the client eighteen months, and the units achieved a higher value each.

Reuse is not always the answer. It is now the answer often enough that demolition should require the written justification, rather than the other way round.