The cost drivers, ranked, and where the money is best spent
Ranked by how much they move the total on a typical project, the drivers run roughly as follows. First, the ground: foundation type and depth, any build-over agreement for a drain, and retaining structure on a sloping plot. These are invisible in the finished room and can absorb a fifth of the budget before anything is seen. Second, the structure: clear spans, open corners and anything carrying a storey above. Third, the envelope: the proportion of glass, its specification and its detailing, plus the roof build-up. Fourth, the services: underfloor heating, a relocated boiler or cylinder, the extract route, and any electrical work triggered elsewhere in the house. Fifth, the finishes and the kitchen itself — which is where owners concentrate their attention and where the marginal pound has the least structural consequence. That ordering suggests a strategy. Resolve the first two as cheaply as the design honestly allows, because money spent there buys nothing you can see: accept the intermediate column, keep the spans sensible, take the drainage survey early so the foundation is designed rather than guessed. Then spend deliberately on the third and fifth, which are what you actually experience: a smaller area of excellent glazing in exactly the right place, good daylight from above where the plan is deep, and a kitchen that is well made. The common failure is the reverse — an ambitious structural gesture that consumes the contingency, followed by value engineering that lands on the glazing specification and the joinery, so the money went into things nobody will ever notice and came out of the things everybody will. A second, quieter point: the cost per square metre falls as the extension gets larger, because the fixed costs of the ground, the structure and the services are spread further. That does not mean building bigger is better — daylight in the middle of the plan usually argues the other way — but it does mean that a very small extension is rarely good value, and that the honest comparison is between a modest extension plus a proper reorganisation of the existing plan, and a larger extension alone.