Data Sources & Methodology

Source data

All company data comes from Companies House, the UK registrar of companies, used under the Open Government Licence v3.0. We do not blend in third-party or estimated datasets.

Extraction pipeline

Financial figures are extracted from filed accounts in a strict order, and each period is labelled with the method used:

• iXBRL tagged data — machine-readable tags in the filed accounts document. Highest confidence.
• PDF text extraction — layout-aware parsing of PDF-only accounts. Labelled as PDF extraction with a confidence level.
• AI-assisted reading — used only when the first two methods fail, metered under plan fair-use limits. Always labelled, and low-confidence results are withheld rather than shown as fact.

PDF or AI-extracted figures never overwrite tagged iXBRL data. Where a filing cannot be read reliably, we say so and link you to the original document.

Calculated indicators

• EBITDA is calculated as operating profit plus depreciation and amortisation. It is not a reported figure and is labelled as calculated.
• Margins are calculated as the relevant profit measure divided by turnover for the same period.
• Solvency indicators (net asset change, liabilities-to-assets, cash cover) are derived from balance-sheet lines when a profit & loss account was not filed.

What we never do

• Missing values are shown as missing — never as zero and never silently estimated.
• Contradictory source data is surfaced, not silently resolved.
• AI never fabricates financial values; it can only comment on figures already extracted and shown to you.

Verification

Every company profile links to the original filings on the Companies House website. Before relying on any figure, verify it against the source document.