Officer and secretary tracking
Workbooks captured appointment concentrations, dated counts, registered-office patterns, service-provider links and review notes.
SCC Companies House analyses public corporate records to identify contextual patterns for professional review. A high-volume address, a repeated officer, a filing delay, a missing classification or a charge pattern is not treated as a conclusion in isolation. The methodology is comparative, time-aware and subject to verification.
OriginsManual tracking
Officer and secretary appointment counts, registered-office clusters, correspondence-address patterns and linked notes were originally maintained in workbooks.
Technical bridgeAnaconda/Jupyter
Local Python notebooks were used for early bulk-data handling, address extraction, matching tests and repeatable analysis outside a production system.
Current stateScheduled runs
GitHub, Colab and Drive now support monthly population screening, weekly monitoring, historical comparisons, manifests and validation artefacts.
Origins
A long-running research process.
SCC Companies House did not begin as a finished website. Its foundations were built through manual public-record research: spreadsheets for company and officer review, tabs for address clusters, date-stamped appointment counts, correspondence-address comparison, entity notes and archived public context material.
The historical workbook evidence is important because it shows the methodology existed before the current platform. Early workbooks used Companies House officer-appointment hyperlinks, dated appointment-count columns and formula-assisted extraction. For public documentation, these are described anonymously rather than reproducing officer IDs, individual names or operational lead material.
Anonymised workbook pattern
=HYPERLINK("https://.../officers/[redacted-officer-id]/appointments", "appointment count")
=IMPORTXML("https://.../appointments", "appointment-count element")
This workbook-led phase was already longitudinal: the point was not only to collect a list, but to ask what changed, when it changed, and whether the same officer, secretary, service-provider or address patterns recurred over time.
Local analysis
Python, Jupyter and Anaconda.
Before the current cloud and GitHub workflow, much of the larger-scale experimentation was carried out locally in Python using Jupyter notebooks within Anaconda. Those notebooks provided the bridge between manual review and repeatable processing: bulk extracts could be loaded, filtered, grouped, compared and exported while preserving the logic in a rerunnable form.
The local notebook stage is a material part of the project history. It shows the methodology did not depend on a single tool. The same core questions persisted across spreadsheets, notebooks and the current automated platform: how can fragmented public records be preserved, compared and interpreted in context?
Archived source material
Public articles, account files, Companies House commentary, court-related material and company-specific workpapers were preserved in structured folders.
Anaconda notebooks
Python/Jupyter workflows tested repeatable extraction, filtering, grouping, matching and address-cohort analysis before production automation.
Companies House extracts
Bulk files and address-specific spreadsheets increased the scale of analysis and highlighted the need for validation and provenance controls.
Colab, Drive and GitHub
Cloud notebooks, version control, Drive archives and GitHub Actions made scheduled runs and archived outputs practical.
SCC Companies House
Monthly, weekly and historical dashboards now convert large public-record inputs into structured review-priority intelligence.
Methodology
Contextual review-priority indicators.
The methodology is intentionally multi-factor. SCC Companies House does not treat any single signal as determinative. High-volume registered-office addresses, repeated officers, missing SIC codes, filing delays, charges or account-status changes can all have ordinary explanations. They become useful only when compared against baselines, time-series movement and other public-record context.
Address concentration
Address analysis compares density with behaviour. The central question is whether filing, officer, sector, charge or temporal patterns deviate from expected address-type norms.
Officer and secretary patterns
Repeated appointments, role movement, correspondence-address patterns and service-provider structures are treated as network indicators for review prioritisation.
Filing and accounts chronology
Accounts, confirmation statements, filing categories and recency are used as chronology indicators. They are not treated as proof of trading status, misconduct or financial distress.
Charges and lender exposure
Charge records and lender groupings are summarised as contextual signals. Public pages avoid presenting named company-lender relationships as conclusions.
Industrial classification
Missing or generic SIC information is treated as a data-quality and screening-context issue, not as evidence of wrongdoing.
Longitudinal movement
Monthly snapshots and weekly monitoring preserve point-in-time movement, allowing remediation, deterioration and new review-priority cohorts to be compared.
Sources and public context
Why the subject matter matters.
SCC Companies House uses official Companies House data as the primary source base. Public reporting and official enforcement material are used only as background context for why registry quality, false-address reporting, filing integrity and corporate-transparency controls matter.
Examples of public context include BBC reporting on companies registered to residential addresses without householders' knowledge and official Insolvency Service material describing public-interest liquidation proceedings involving false filings and company formation patterns. These sources help explain the public-interest background; SCC Companies House does not import their allegations into its own public outputs.
Provenance and reproducibility
Preserving how each result was produced.
A useful public-record intelligence result is one that can be revisited. SCC Companies House therefore treats provenance as part of the product: source period, run identifier, generated timestamp, source hashes, code version, validation status, exported payloads and archived artefacts should be retained wherever practical.
Source lineageCompanies House public records, monthly bulk snapshots, targeted API enrichment and protected workpaper references.
Processing recordLocal notebook history, cloud notebook runs, GitHub workflow logs, run IDs and generated timestamps.
Validation artefactsValidation workbooks, manifests, row-count checks, redaction checks, smoke tests and generated reports.
Historical snapshotsPoint-in-time monthly outputs are retained so later analysis can distinguish current data from what was known at the time.
Public/protected splitPublic summaries are restrained; operational identifiers and detailed workpapers are held in controlled environments.
Technology evolution
How the original method became scalable.
Excel and Google SheetsManual tracking, appointment-count columns, hyperlinks, formulas and review notes.
Anaconda / JupyterLocal Python notebooks for early bulk-data experiments and repeatable extraction.
Google ColabCloud notebooks for larger processing, validation and reporting.
Google DriveEvidence library, archived outputs, manifests, PDFs, reports and protected workpapers.
GitHubVersion control, workflow history, site generation and deployment scripts.
GitHub ActionsScheduled monthly and weekly automation, site deployment and content-safety checks.
AI-assisted developmentAccelerated coding, debugging, synthesis and methodology drafting under human direction.
Limitations
What SCC Companies House does not do.
- It does not make allegations of misconduct.
- It does not determine insolvency, criminality, creditworthiness, investment suitability or legal liability.
- It does not replace direct record verification, professional judgement or human investigation.
- It does not treat address concentration, officer repetition, filing delay or missing classification as conclusive in isolation.
- It does not publish protected workpapers, named investigative leads, raw operational notes or full identifiers on public pages.
Summary
The project in one sentence.
SCC Companies House is the formalisation of years of manual and notebook-based Companies House research into a reproducible, auditable and context-aware public-record intelligence framework.