Disclosure: I'm Lindenbyte, the small studio that built the Apify Actor used below, and this post was
drafted with AI assistance. Companies House's own advanced search and free API work too; this is a shortcut.
A brand-new limited company needs an accountant, a bank, a website, insurance, software and an office
address, usually within weeks. If you sell any of those, the Companies House register is a free, official
list of businesses that did not exist last week.
How big is it? In the 7 days to 13 Nov 2026, 12,540 active companies were incorporated in the UK
(about 2,476 on 12 Nov alone). Nobody wants all of them. This post shows how to narrow that down to your
niche and area, and get a fresh list every morning.
1. Pick your SIC codes
Every UK company declares one to four SIC 2007 codes at incorporation. A few useful ones:
-
62012Business and domestic software development,62020IT consultancy -
56101Licensed restaurants,47910Retail via mail order or internet (e-commerce) -
41202Construction of domestic buildings,70229Management consultancy
The software pair (62012 + 62020) matched 1,188 active companies in that same 7-day window, and
218 on 12 Nov alone. So a "new UK software companies" list is a few hundred rows a day, not thousands.
Open the UK Companies House New Company Finder
and start with:
{
"incorporatedLastDays": 7,
"sicCodes": ["62012", "62020"],
"location": "London",
"excludeDormant": true,
"includeProfile": true,
"maxResults": 500
}
Each row has the company name and number, incorporation date, SIC codes with their descriptions
("62020 Information technology consultancy activities"), the registered office address split into
fields, and a link to the company's Companies House page.
2. Filter by area (and know what the address means)
There are two location filters:
-
locationis Companies House's own free-text filter."London"cut the 1,188 software companies down to 547. -
postcodePrefixesunderstands UK postcodes:Mis Greater Manchester (not ME),SW1is SW1A-SW1Y (not SW10),BS8 1is a single sector. It runs after the search, and dropped rows are not charged.
One honest caveat, visible in the data. I filtered the software companies to EC1V and EC2A (Old Street,
Shoreditch) and got 157 hits, but 150 of them used one of a handful of addresses: 128 City Road
(116 companies, counting its 124 and 124-128 variants) and 66 Paul Street (34). Those are well-known
company-formation and virtual-office addresses. A registered office is a legal address, not necessarily where anyone works.
If you sell something local (an office fit-out, a local accountant), add nameExcludes or drop known
formation-agent addresses downstream; if you sell something remote (software, bookkeeping), those
companies are still perfectly good leads.
excludeDormant drops companies registered as dormant (SIC 99999) or non-trading (74990). For new
companies it is a small slice (94 of the 12,540, under 1%), but they are almost never buyers, and
the dropped rows are free.
3. Add accounts dates if you sell to finance people
includeProfile: true makes one extra API call per company and adds the accounts and confirmation
statement dates. For a London IT consultancy incorporated on 8 Nov 2026 the profile said:
{
"accountsNextDue": "2028-08-08",
"accountsNextPeriodEnd": "2027-11-30",
"accountingReferenceDate": "30-11",
"confirmationStatementNextDue": "2027-11-21",
"lastAccountsType": null
}
That is exactly what an accountant or bookkeeper wants: the first confirmation statement is due in about
a year, the first accounts in about 21 months, and nobody has filed anything yet. All 25 London companies
in my sample had first accounts due 8-10 Aug 2028.
4. Schedule it daily and keep only new rows
In Apify Console, create a Schedule (for example 07:00 UK time) that runs the Actor with
"incorporatedLastDays": 2. Companies House can take a day to index new incorporations, so a 2-day
window overlaps and catches stragglers; remove the duplicates on your side by company number:
from apify_client import ApifyClient
import csv, os
client = ApifyClient(os.environ["APIFY_TOKEN"])
run = client.actor("lindenbyte/uk-companies-house-new-companies").call(run_input={
"incorporatedLastDays": 2,
"sicCodes": ["62012", "62020"],
"postcodePrefixes": ["M", "SK", "WA"],
"excludeDormant": True,
"maxResults": 1000,
})
seen_file = "seen_companies.txt"
seen = set(open(seen_file).read().split()) if os.path.exists(seen_file) else set()
new = [c for c in client.dataset(run.default_dataset_id).iterate_items() if c["companyNumber"] not in seen]
with open("new_companies.csv", "a", newline="") as f:
w = csv.writer(f)
for c in new:
w.writerow([c["incorporationDate"], c["companyNumber"], c["companyName"],
c["sicText"], c["registeredOfficeAddress"], c["companiesHouseUrl"]])
with open(seen_file, "a") as f:
f.writelines(c["companyNumber"] + "\n" for c in new)
print(f"{len(new)} new companies")
If you prefer no code, the dataset also exports to CSV/Excel or connects to Google Sheets, Zapier or Make.
The free alternatives (honestly)
- The Companies House advanced search page lets you filter by incorporation date, SIC code, status and location for free, by hand.
- The Companies House REST API is free too: register, get a key, and you get 600 requests per 5 minutes. Its advanced search stops at 10,000 results per query, so a week of all UK incorporations (12,540) needs to be split by date.
The Actor wraps that same official API and adds: no key needed (or bring your own), automatic date
splitting past the 10,000 cap (a 30-day pull of 30,000 companies took 17 seconds in my test), SIC
descriptions, a UK-aware postcode filter, a dormant switch, scheduling and exports.
What it deliberately does not do
It outputs company-level data only: no director or owner names, birth dates or home addresses, and no
emails or phone numbers. The optional officer/PSC enrichment returns counts only. You will need your own
way to reach the business (its website, LinkedIn page, or a letter to the registered office). Outreach to
UK businesses is covered by PECR and UK GDPR, with different rules for email, phone and post and for
limited companies versus sole traders, so read the ICO's guidance before you contact anyone.
Cost: $0.003 per run plus $0.0015 per company, and another $0.0015 per company with includeProfile.
The full 7-day list of 1,188 new software companies would be about $1.79; a daily London-only software
list is typically well under $0.50 a run.
Not affiliated with Companies House. Contains public sector information licensed under the Open Government Licence v3.0.
Top comments (0)