construction output

Construction output is the measure of work actually done in a period — not orders booked or permits granted, but value and volume delivered on site. It is the demand signal material suppliers, equipment fleets and contractors price against.

The cataloged reference is Construction Industry Statistics (Output, New Orders, Materials) from ONS. Its granularity is "Monthly, by sector (new housing, infrastructure, industrial, commercial, R&M), public vs private split", and its description lists "output volume and value by work type, new orders, output price indices, revisions triangles, and index weights".

Delivery is one ~540 KB XLSX per release containing 22 tables, with roughly 140 archived vintages retained per dataset — enough to reconstruct past estimates, not just current ones.

Permits tell you what was approved; output tells you what happened. A materials-demand forecast keyed to permits alone inherits projects that stall, so pairing both series separates intention from delivery.

The public/private split changes the customer. Infrastructure spending behaves differently from private housing across rate cycles, and a dataset without that split cannot distinguish them even when totals match.

Revisions are part of the product here. Because ONS retains ~140 archived vintages, you can measure how much early estimates moved — which matters when your model trains on data that was itself provisional at collection time.

1. Demand the work-type breakdown. New housing, infrastructure, industrial, commercial and R&M should each be separately addressable; aggregate-only feeds hide the sector you care about.
2. Check the public/private dimension. The ONS release carries it natively; ask vendors to prove their extract preserves it rather than collapsing it.
3. Test revision transparency. The release includes revisions triangles and index weights — if your provider drops them, you lose the ability to audit estimate stability across the ~140 retained vintages.
4. Verify geographic scope matches your market. This series covers Great Britain; US-facing work needs a different source, so confirm geography before comparing prices.
5. Confirm format fit. XLSX-only delivery suits analyst workflows; automated pipelines need conversion, so check what else the vendor offers before committing.

Where to see it in context: building-products data, construction-engineering data.

• building-material-price-index — The input-cost counterpart: output volumes mean most when deflated against material inflation.
• building-permits — Approved authorizations that lead output; the pair separates projects filed from work delivered.