Thousands of new companies are registered in the UK every week (when I checked on 15 Nov 2026, the register showed about 14,000 incorporations in the previous seven days). For accountants, banks, insurers, web agencies and SaaS sales teams that is a useful list of research leads: a freshly incorporated company may still be choosing its providers, though that is a hypothesis, not a given.
The good news: you don't need to buy that list. Companies House, the UK company register, offers it for free through an official API. Here's how to pull it yourself with a short Python example, plus a no-code option if you just want a CSV.
1. Get a free API key
- Create an account at the Companies House developer hub (
developer.company-information.service.gov.uk). - Create an application (choose Live).
- Create a REST API key. No domains or IPs needed.
The key is used as the username for HTTP Basic auth, with an empty password. The limit is 600 requests per 5 minutes per application, which is enough for the examples here.
2. Query the advanced search
The endpoint you want is /advanced-search/companies. It filters by incorporation date, SIC code (industry), status, company type and location, and returns up to 5,000 companies per request.
import csv
import os
from datetime import date, timedelta
import requests # pip install requests
API_KEY = os.environ["COMPANIES_HOUSE_API_KEY"]
since = (date.today() - timedelta(days=7)).isoformat()
params = [
("incorporated_from", since),
("company_status", "active"),
("sic_codes", "62012"), # business and domestic software development
("sic_codes", "62020"), # IT consultancy
("location", "London"),
("size", 5000),
]
r = requests.get(
"https://api.company-information.service.gov.uk/advanced-search/companies",
params=params, auth=(API_KEY, ""), timeout=30,
)
if r.status_code == 404: # documented answer for "no matching companies"
items = []
else:
r.raise_for_status() # any other error stays visible
data = r.json()
items = data.get("items", [])
if int(data.get("hits", 0)) > len(items):
print(f"Warning: {data['hits']} matches, only {len(items)} returned. "
"Page with start_index or narrow the filters.")
with open("new_companies.csv", "w", newline="", encoding="utf-8") as f:
w = csv.writer(f)
w.writerow(["number", "name", "incorporated", "sic", "town", "postcode"])
for c in items:
a = c.get("registered_office_address", {})
w.writerow([c["company_number"], c["company_name"], c.get("date_of_creation"),
" ".join(c.get("sic_codes", [])), a.get("locality"), a.get("postal_code")])
print(len(items), "companies saved")
That's it: a snapshot of companies matching these filters that were incorporated in the last 7 days. Run it again tomorrow and you get an overlapping list; to see only new ones, store the company numbers you already have and skip them. In my test on 15 Nov 2026, this exact query returned 616 London software/IT companies.
3. Things that will trip you up
-
Dormant shells. Filter out SIC
99999(dormant) and74990(non-trading). Otherwise some of your results are shells that are unlikely to be reachable customers. (Holding companies have their own SIC codes, such as64209, and are not the same thing.) - Formation-agent addresses. Many companies use the address of their formation agent or accountant as the registered office. The postcode is often not where the business actually operates.
-
Filing deadlines can be a research signal. The company profile endpoint (
/company/{number}) returnsaccounts.next_accounts.due_onandaccounts.next_accounts.overdue(the olderaccounts.next_dueis deprecated) plusconfirmation_statement.next_dueandconfirmation_statement.overdue. An overdue filing is a reason to look closer; it does not prove that the company wants to buy anything. - Personal data. Directors' names come from the officers endpoint, and a registered office can be a private home address. Under UK GDPR and PECR, outreach to individuals needs a lawful basis, transparency and an opt-out; legitimate interest is one possible basis, not a blanket permission, and depending on the channel and recipient, PECR may require prior consent. Company-level data is the safer starting point. This is not legal advice.
-
Pagination.
sizegoes up to 5,000, andstart_indexlets you page. Companies House staff have stated that the search stops at 10,000 results per filter set (forum answer, July 2025), so narrow by date or SIC if you need more. - Empty results. A search without matches returns HTTP 404 instead of an empty list. The script above handles that case.
4. No-code option
If you don't want to maintain a script, I packaged the same logic as a ready-made tool on Apify: UK Companies House Search. (Disclosure: I built it.) It adds SIC descriptions, filing deadlines and overdue flags, and an "only new" mode that remembers company numbers for a daily feed. It leaves out directors and officer data, but registered office addresses are included. You pay per company returned. Ready-made examples like "New Tech Companies in London This Week" can be run as-is.
The Python above works without it.
Wrap-up
Companies House offers a very useful free business dataset. With one API key and one endpoint, you can pull an industry-filtered list of recently incorporated companies without scraping.
What are you using new-company data for? I'm curious which industries people filter for.
Contains public sector information licensed under the Open Government Licence v3.0.
Top comments (0)