Methodology
How KEYSTONE. collects, processes and publishes data
The principles and procedures behind every indicator, dataset and publication. They are the rules that govern the passage from raw public statistics to the harmonised data we publish.
Principles
Our work rests on four guiding principles: rigour - only official, verifiable sources, with every data point documented; transparency - every indicator carries its source, reference date and the processing it has undergone; comparability - units, nomenclatures and reference periods are harmonised so figures read consistently across countries and over time; reproducibility - collection and processing procedures are documented and versioned, so any published result can be verified or replicated.
Sources
All KEYSTONE. data comes from two authoritative public sources. For multilateral trade we use UN Comtrade; for European trade at 8-digit Combined Nomenclature granularity, Eurostat Comext. Both report imports, exports and re-exports at HS/CN code level, and we cross-check one against the other. The full source catalogue is published on our data sources page.
Classification and nomenclatures
For natural stone we apply three mutually-harmonised systems. The Harmonised System (HS) is the multilateral backbone, with Chapter 25 covering crude products (2514 slate; 2515 marble and other ornamental calcareous stones; 2516 granite, porphyry, basalt, sandstone) and Chapter 68 covering worked products (6801–6803). The European Combined Nomenclature (CN) extends these codes to 8 digits, distinguishing for example rough blocks from sawn slabs. The Statistical Classification of Economic Activities (NACE Rev. 2.1 / CAE Rev. 3) - mainly divisions 08 and 23 - frames industrial production. We maintain internal concordance tables between HS, CN and NACE, available with every publication.
Harmonisation and processing
Raw statistics go through a standard pipeline before entering our databases. Tonnes and square metres are converted to a common unit (tonnes for trade flows; square metres for finished products where applicable). Values are converted to EUR and USD using annual average European Central Bank rates, and published both at current and at constant prices (deflated by the construction price index). Country codes follow ISO 3166-1 alpha-3. For trade we apply mirror reconciliation, cross-checking data reported by the exporting country against the importing country's figures and flagging discrepancies above 10%.
Validation and quality control
Every dataset passes through three stages before publication: automated checks for integrity (unintended negatives, abrupt period jumps, non-existent codes, duplicates); analytical review against historical series and independent sources (for example, Eurostat versus UN Comtrade); and a final editorial review by an analyst with sector experience. Published indicators carry, where applicable, a quality note: confirmed, provisional or KEYSTONE. estimate.
Update cycle and known limitations
International trade is updated monthly, with a typical two- to three-month lag; annual-series dashboards once a year. Existing outlooks are refreshed annually, and new outlooks are released as they are completed. Every data page shows the latest update and the next scheduled one. Known limitations are disclosed explicitly: uneven statistical coverage across producing countries; 6-digit HS granularity that does not distinguish finishes or colours; unit values that are sensitive to product-mix shifts and should be read as indicators, not market prices; and, in the most recent series, KEYSTONE. estimates where partial data requires modelling - these are always flagged.
Source data availability
Every trade dashboard is built from two public institutions — UN Comtrade and Eurostat Comext — and we publish each series exactly as those institutions release it. The availability and timeliness of the underlying figures depend entirely on each country reporting its trade to UN Comtrade and Eurostat: KEYSTONE. does not collect this data first-hand and does not control when, or whether, a given country's figures are published, and accepts no responsibility for delays, gaps or revisions on the institutions' side. Reporting lags vary widely between countries — many report within two to three months, but for some the most recent months or years can be delayed by more than six months. The latest periods shown may therefore be incomplete and are subject to revision as the source institutions receive and consolidate each country's submissions.
Citation and transparency
All data published by KEYSTONE. carry a primary source, a reference date and, where applicable, a link to the original dataset. Open-access data may be reproduced, downloaded and reused provided the source is cited, in the following format: "KEYSTONE. – Minerals Data Observatory (year). [Dataset or publication title]. Accessed [date]." For methodological enquiries, please use our contact page.