Methodology and sources
Last updated
This page explains where our data comes from, how we prepare it and how we handle corrections. Individual reports add their own method notes.
Our sources
The table lists sources used in published pages. We do not describe a source as in use until a report, table or tool cites a file we have downloaded.
| Source | What we use | Licence |
|---|---|---|
| Companies House | Official statistics on incorporated companies, and the annual register-activities file for SIC codes on the register | Open Government Licence v3.0 |
| Office for National Statistics (ONS) | Business Demography reference tables, mid-2024 population estimates, UK business activity, size and location for 2025 and 2026, Business Insights and Conditions Survey Waves 165 and 166, and regional GDP for 1998 to 2024 (published 23 September 2026) | Open Government Licence v3.0 |
| Insolvency Service | Monthly company insolvency statistics, including the industry workbook | Open Government Licence v3.0 |
| Department for Business and Trade | Business Population Estimates 2025, and large-business payment practices statistics for 2025 | Open Government Licence v3.0 |
| Bank of England | Daily Bank Rate (series IUDBEDR), for the late payment calculator, and monthly interest rates on new loans to small and medium sized enterprises (CFMZ6LD, CFMZJ3L, CFMZJ3U) with the monthly average of Bank Rate (IUMABEDR) | The SME series is republished as a dataset. The daily Bank Rate file is kept with our sources |
Attribution
Where we use public sector information, we include this statement:
Contains public sector information licensed under the Open Government Licence v3.0.
Where we use Companies House data, we also state: "Source: Companies House data, analysed by Apexium Studio", with the date we downloaded it.
How we prepare the figures
- Download the official file. Each published number comes from a statistics release we have saved under
data/sources, with the page URL, publication date and the date we downloaded it. We do not type figures in from a news story or a search snippet. - Rebuild from that file.
scripts/build-official-data.pywrites the incorporation, insolvency and Pulse CSVs.scripts/build-batch-content.pywrites the later births, deaths, survival, size and register CSVs.scripts/extract-sme-loan-rates.pywrites the SME loan-rate CSVs.scripts/extract-regional-gdp.pywrites the 2024 regional GDP CSV. The site reads those CSVs. It does not recalculate them at build time. - Keep unlike series apart. An incorporation (Companies House) is not an ONS business birth. A monthly average of Bank Rate is a different series from the daily official rate. Real regional GDP growth is a different measure from GDP per head at current prices, and from a count of business births. A seasonally adjusted insolvency total is not the same as a not seasonally adjusted count. A SIC record on the register is not a company, because one body can have several codes. Business Population Estimates include unregistered sole traders. An insolvency count by industry is not a rate. A VAT and PAYE enterprise is not a local unit, and neither is the Companies House register. Payment-practice medians cover large reporters only. BICS is a weighted voluntary survey and is labelled official statistics in development. Reports say which one they are using.
- Company types. Incorporation charts use the Companies House row "All companies". Limited liability partnerships are a separate row in the same release and are not added in.
- Places. The local league table uses the ONS local-authority rows for business births. It does not use registered-office postcodes, and it is not a table of incorporations.
- Revisions. Official statistics can be revised. The page states the publication date and the next scheduled release when the publisher has announced one.
- Aggregates only. We publish counts and rates, never lists of named directors or other personal data.
- Rounding. Percentages and the Pulse are rounded half up to one decimal place. The method for each series is on the page that uses it, and the Pulse method is on the Pulse page.
Rounding and comparisons
Percentages are rounded half up to one decimal place. Year-on-year comparisons use the same period a year earlier. Where a publisher has changed its method, the page says so rather than treating the series as unchanged.
Corrections policy
If we get something wrong, we fix it quickly and say so. Corrections appear at the end of the page with the date and what changed. Significant errors are also noted at the top. To report an error, email research@apexiumstudio.com with the page link and what you think is wrong. We aim to reply within five working days.