600 requests per 5 minutes is not enough for your batch job
Your job needs thousands of companies. The official API is designed for retrieving individual records, so a set-sized question becomes a queue of requests, rate-limit sleeps and 429 retries.
What is happening
Companies House provides an advanced-search endpoint for set queries by SIC, location, status, type and dates. It is the right free first choice when its 5,000-row maximum covers the job.
For per-company profile calls or a result beyond that maximum, requests for large rate-limit uplifts do not change the underlying mismatch.
Rate-limiter libraries, retry loops and scheduled overnight jobs can make the integration polite, but they do not turn a record-shaped interface into a set-shaped one. Your batch job is outside the API's intended shape, not necessarily badly written.
What it disrupts
- A simple list refresh becomes a long-running job with state, backoff and retry handling.
- One partial failure can leave an inconsistent batch that needs resuming or rebuilding.
- AI agents inherit the same fan-out problem when every organisation requires a separate tool call.
Why it happens
- The official profile endpoints are optimised for current information about a known company, while advanced search covers bounded set queries.
- A batch question asks for every company matching a condition, which is a query over a set rather than a sequence of lookups.
- The official bulk snapshot answers the scale problem, but moves filtering and database work onto the user.
Checks to make now
- Try the official advanced-search endpoint first when the job can be expressed as SIC, location, status, type and date filters and fits within 5,000 rows.
- Decide whether you need a larger monthly register baseline or genuinely real-time filing changes.
- Express the job as SIC, registered-office geography, company status, company type and incorporation-date filters where possible.
- Use the official live or streaming APIs for time-critical deltas that the monthly snapshot cannot provide.
Where DataScoop fits
DataScoop queries its own copy of the official monthly snapshot, currently 5,695,465 records imported on 2026-08-25. One filtered request replaces per-company fan-out for list-shaped work.
The free tier returns 25 rows for integration work. Pro at £19 per month supports paged JSON, full streamed CSV exports and the same query through MCP.
Where it fits, and where it does not
Good fit
- Batch, list and segmentation jobs beyond the advanced-search result maximum whose required fields exist in the basic company snapshot.
- Developers and AI agents that benefit from one set-shaped query with explicit filters.
Not the right fit
- Minute-by-minute filing changes, live company accounts, officers or persons with significant control.
- A transparent proxy around the official API; DataScoop maintains and queries its own dated snapshot instead.
Proof and boundaries
The source is a monthly snapshot, not the live filing stream. Combine it with an official live source when freshness within the month matters.
JSON is paged and subject to documented plan limits. A Pro CSV export streams the complete matching result.
Only company-level organisational fields are carried. Officers and persons with significant control are not ingested.
There is no performance or customer outcome claim. The proof is the public schema, documented limits, free query tier, OpenAPI specification and MCP tool surface.
Questions people ask
Is DataScoop real-time?
No. The Companies House source is the official monthly basic-company snapshot. Use an official live or streaming API for changes that must be known within minutes.
Does DataScoop proxy the Companies House API?
No. It maintains its own queryable copy of the official bulk snapshot, so DataScoop queries do not consume your Companies House API allowance.
How many rows does one call return?
The free tier returns 25 rows. Pro JSON is paged, while a Pro CSV request streams the complete matching set. The current limits and parameters are published in the DataScoop documentation.
Does it include company officers?
No. DataScoop deliberately carries company-level organisational fields only.
Ask the register-sized question once
Check the filters and response shape in the documentation, then use the free tier to test your integration against real rows.
Read the query documentation