Eurostat PRODCOM — Production Statistics for Building Product Industries

The EU's official answer to who made what, and how much did it sell for. PRODCOM - PRODuction COMmunautaire - is Eurostat's survey of manufactured output covering mining and quarrying, manufacturing and materials recovery (NACE sections B, C and E), organized around a product list of about 4,000 distinct goods at 8-digit level.

That coding is the quiet genius of the thing. The first four digits of every code place it in a NACE class, the first six in a CPA heading, and only the last two are PRODCOM-specific - so a single number reads as an industry, a product family and a sellable good at once, and most codes carry a complete Combined Nomenclature reference that lets production reconcile line-for-line against international trade data.

For building products, the relevant headings cover bricks, cement, ceramic tiles, doors, windows and insulation, sitting alongside the broader chemicals, paper, metal and rubber categories this industry shares with packaging, automotive and electrical manufacturing. Variables collected are sold production in value and quantity, production under subcontracted operations - broken out as its own measure starting with reference year 2021 - and total actual production. Within Datadory's catalog of 1,744 datasets, where the mean quality score is 7.81, this record scores 8/10 on the strength of its verified field structure and unusually clean product taxonomy. Get a sample of this dataset cut to your headings.

One observation per product heading per member state per reference year, with the measure set attached:

# prodcom panel -- one observation per product heading x country x year
prodcom_code     : 23311010   # 8 digits: 2331=NACE class, 233110=CPA heading
geo              : DE         # reporting EU member state
time             : 2024       # reference year

# measures carried on every heading
sold_production_value    : EUR millions, national currency converted to EUR
sold_production_quantity : physical units defined for the heading
subcontracted_production : collected separately from reference year 2021 onward
actual_total_production  : total production irrespective of sale channel

# the same spine repeats across ~4,000 headings x member states x years
# bricks, cement, ceramic tiles, doors, windows and insulation all resolve to
# their own 8-digit codes inside NACE sections B, C and E

# scale check from the source's own news line:
# wind turbines, reference year 2024 -> EUR 10.4 billion in sold production value

Read the anatomy rather than the digits. The code does triple duty - 2331 places the row in a NACE industry, 233110 in a CPA product family, and the trailing pair pins the exact good - which is why a single extract serves an industry analyst and a SKU planner at once. The four-measure block matters just as much: value without quantity invites price inflation to masquerade as growth, sold without total hides utilization, and since reference year 2021 the subcontracted split means outsourced production no longer vanishes between the plant and the sale. The wind-turbine line is the survey's own news item - EUR 10.4 billion in sold production for reference year 2024 - a fair scale check on what one heading's value column carries.

Seven fields anchor the record, definitions verified against Eurostat's published documentation during the August 2026 research pass. Where the survey describes a variable but publishes no worked example cell, the dictionary leaves the example empty rather than decorating it with an invented figure.

Identity trio - PRODCOM code, geo and time key every row. The code's internal structure (4 digits NACE, 6 CPA, 8 PRODCOM) means joins to industry classifications need no lookup table; string-slicing does it. See what a PRODCOM code is.

Measure quartet - sold production value in EUR millions and sold production quantity in the heading's declared unit carry the commercial reading; subcontracted production and actual total production complete the picture from reference year 2021 onward. See how sold production statistics work.

Additional fields on request: the heading-level product-list reconciliation, the Combined Nomenclature crosswalk, NACE section B and E panels, revision-calendar alignment per member state and confidentiality-suppression flags. Name them at sampling and they arrive as typed columns, documented against delivered extracts - see NACE Rev.2 codes.

• Geography: EU member states, with participation varying by country and by statistical confidentiality - where too few producers report, cells are suppressed rather than published. That suppression is a feature of official statistics and a trap for anyone summing the table naively.
• Temporal: annual series back to 1995 in the database, giving three decades of member-state output that span two full construction cycles. The latest full reference year lands each July - 2024 data is already in the record - and a revision calendar built from national calendars is published alongside.
• Granularity: annual, one cell per 8-digit product heading per member state, across roughly 4,000 headings in NACE sections B, C and E. Nothing is pre-aggregated away; the brick-level view survives to the bottom of the hierarchy.

Set against the wider Datadory catalog - average quality score 7.81 across all 1,744 datasets - this slice scores 8/10, carried by disciplined field documentation and a product taxonomy that doubles as an industry classification. It sits inside our best building products datasets ranking.

API, files, or your warehouse. Daily, weekly, or hourly.

Pick the channel your stack speaks; the rows arrive identical either way - codes resolved against the product list, units typed, suppressions flagged rather than read as zeros. Most teams take the 1995-forward history once as a backfill and keep each July's new reference year arriving on whatever cadence their models expect; cadence changes are a settings conversation, not a re-integration project.

Every delivery travels with the field dictionary above, the product-list reconciliation for your headings and the sample rows for validation. Get a sample of this dataset before anything recurring exists.

• Member-state market sizing - sold production value in EUR millions prices the addressable market for cement, tiles or insulation country by country while the quantity measure keeps price spikes from posing as volume growth; see market sizing use cases.
• Capacity and utilization signals - total actual production beside sold production exposes how much of what a country's plants make actually moves.
• Outsourcing watch - the subcontracted variable, new from 2021, tracks how much European building-products output has moved off producers' own books.
• Share-of-market tracking - one methodology across all reporting member states makes cross-border producer-share trends comparable; see supply chain mapping use cases.
• Production-to-trade reconciliation - Combined Nomenclature references let domestic production sit directly against import and export flows.
• Forecast anchoring - series back to 1995 give demand models three decades of official output including two downturns; see citation-grade research use cases.

Inside the same vertical, the USGS Mineral Commodity Summaries carry cement, crushed stone, sand and gravel with global reach but US-centric depth, and the Census MRTS construction series read American demand monthly. PRODCOM owns the member-state production grid.

Market researchers and consultants (relevance 3 of 3) get the citable denominator for any European building-materials study - official value and quantity per product per member state; see market researchers x building products. Investors and quant researchers (relevance 2) frame the demand side behind cement, insulation and fenestration equities with three decades of member-state output; see investors and quants x building products. Competitive intelligence and product teams (2) read the sold-versus-total gap and the subcontracted split as early tightening signals; see competitive intel x building products. Data scientists and ML engineers (2) get a stable 8-digit-coded panel that drops into pipelines without schema archaeology; see data scientists x building products. Journalists, academics and students (2) cite a named official EU survey whose methodology is published; see journalists and academics x building products. Start from the building products data hub for the rest of the shelf.

Within Datadory's twenty-four building-products records this one occupies the member-state production grid: who manufactured how much of which good, officially, every year, since 1995. The Eurostat Dissemination API construction production and permits record shares the household but counts activity - construction output indices and building permits - rather than product-level manufactured goods, so the two stack naturally: permits say what will be built, PRODCOM says what Europe's factories made to build it with. The UK ONS construction statistics record is the deepest single-country counterpart, with output, new orders and materials price data in monthly and quarterly cadence.

The natural division of labor runs geography against granularity: PRODCOM trades frequency (annual) for product resolution (8-digit codes nobody else publishes); construction-indicator feeds trade product detail for speed. Teams modeling the European building-materials economy typically hold both and reconcile them on the NACE spine.

Four things, stated plainly.

First, confidentiality suppresses real cells. Where too few producers report in a member state, values are withheld rather than published. Those gaps are part of the official record; deliveries flag them explicitly so nothing gets summed as a zero.

Second, value and quantity are different instruments. Value arrives in EUR millions converted from national currency; quantity arrives in the unit the product heading declares - tonnes, square metres, pieces. Summing across headings without honoring units is the classic way this table lies.

Third, the newest measures have the shortest histories. Subcontracted production is collected separately only from reference year 2021 onward, so any trend claim about outsourcing starts there, not in 1995.

Fourth, annual means July. The latest full reference year enters the record each July; a pipeline expecting mid-year freshness should say so at scoping, because the fix is a companion indicator feed rather than wishful polling. None of these wrinkles get smoothed over - the sample shows the terrain unchanged. Get a sample of this dataset shaped to your headings and years.

Because a 4,000-heading product list, a 27-country dimension, three decades of revisions and a unit system that switches per heading is exactly the bundle teams mis-assemble once and argue about forever. Datadory delivers it assembled: headings reconciled against the current product list, Combined Nomenclature links intact, suppression flags explicit, and samples cut to the products, member states and years your question names before any recurring delivery exists. Browse the vertical on the building products data hub, the building-products entries in our free building products datasets list, the Eurostat source profile, or get a sample of this dataset and see your own cut first.