Companies House data, pulled on a schedule
Companies House publishes every UK company filing for free, which sounds like it solves the problem until you try to use it at scale. The web interface is built for looking up one company at a time. The bulk downloads are enormous and arrive in a shape nothing else reads. If you want incorporations from last Tuesday in three SIC codes, filtered by region, landing in your CRM every morning, you are writing code. That is the piece I build.
- Starting at
- from £1,200
- Refresh
- Daily, or hourly for a watchlist
- Access
- public api
What comes out of it
- Company number, name, status, and incorporation date
- Registered office address and SIC codes
- Officer appointments, resignations, and dates of birth
- Filing history with document links
- Charges, insolvency notices, and mortgage records
What makes Companies House awkward
Rate limiting is 600 requests per five minutes, which sounds generous until you walk the filing history of 40,000 companies. The fix is a job queue that respects the window, checkpoints its position, and resumes rather than restarting after a failure. The second problem is that filings arrive as PDFs and iXBRL, so extracting numbers from accounts means parsing the structured tags where they exist and falling back to document parsing where they do not.
Companies House runs a real public API with documented rate limits, so the interesting work is not evasion, it is the data model. Officer records are not deduplicated across companies. The same director appears as several entities with slightly different birth dates and spellings, and any product that counts directors has to resolve that.
What people actually use this for
The most common request is lead generation with a filter that means something. New incorporations in a SIC code, in a region, above a certain share capital, checked every morning and pushed into a CRM before a salesperson has opened their laptop. Volume is not the hard part. Precision is: a list of every company incorporated yesterday is noise, and a list of the eleven that match your customer profile is a pipeline.
The second use is monitoring rather than acquisition. Accountants and lenders want to know when a client files late, appoints a director, or has a charge registered against them. That is a watchlist with alerting, not a bulk pull, and it wants to run hourly rather than daily.
The third is enrichment. You already have a customer list and you want it backed by filed accounts, company age, and officer history so your risk scoring rests on something filed at a registry rather than typed into a form by the customer.
How the pipeline is built
A scheduled job walks the API within its rate limit, writes raw responses to storage before parsing anything, and only then transforms into your schema. Keeping the raw layer matters more than it sounds: when you discover in month four that you want a field you did not extract in month one, you reprocess history instead of refetching it.
Change detection runs on a hash of the fields you care about, so a company that files nothing generates no downstream noise. Alerting fires on transitions rather than states, which is the difference between one useful notification and a daily email nobody opens.
Delivery is whatever you already use. A Postgres table, a warehouse, a CRM via its API, or a dashboard if there is nowhere sensible to put it yet.
What it costs and how long it takes
Workflow automation starts from £1,200 for a bounded Companies House API extraction with one agreed filter and destination. Scheduled monitoring, storage, watchlists and handover requirements are priced against the actual scope; the full monitoring system is not included in that entry price.
A monitoring system with watchlists, change detection, document parsing, and enrichment against your existing records is quoted separately after agreeing filing coverage, parsing requirements and acceptance checks.
Because this source is a documented public API rather than a website being scraped, maintenance is lighter than most. The API changes rarely and with notice. Most clients here do not need a maintenance plan, which is not something I can say about many sources on this site.
Where the line is
Official public API. No terms are being worked around.
Companies House data is published under the Open Government Licence and the API is offered for public use with a free registered key. This is about as clean as data sourcing gets: no terms are being circumvented and no login is being shared. The care needed is on the other side. Officer records contain personal data covered by UK GDPR, so if you are storing directors and their addresses you need a lawful basis and a retention position. I build the retention rules in rather than leaving them to you.
MathersLabs is not affiliated with or endorsed by Companies House. Work is scoped to publicly accessible data and to compliance with applicable terms and law, and feasibility is assessed per engagement before any fixed price is agreed.
Related work
SignalSauce
250 qualified leads a month, fully automated
Built an end-to-end intelligence pipeline that ingests from 5 curated news sources twice daily, scores every announcement against a custom algorithm, and enriches the ones that pass with 21 structured data points per company.
Questions people ask first
Is your source on this list?
Tell me what data you need and where it lives. I will tell you whether it is feasible, what it would take, and whether it is worth building before you commit to anything. Working hours overlap with US Eastern and Pacific.