Disclosure: I'm Ben, and I built the Apify tool I mention near the end. It charges per result. The DIY part below works with free stuff only.

If you sell to new UK companies, or you're an accountant looking for first-year clients, the official register is the right place to start. You don't need a Companies House API key for a basic leads list. Companies House publishes a monthly snapshot of live companies, for free, under the Open Government Licence v3.0.

The file is called the Free Company Data Product. It's a snapshot of companies on the register, compiled to the end of the previous month and usually published within 5 working days of month end. It's not a same-day feed. A company incorporated this week shows up in the next snapshot.

What is in the snapshot

I used the 1 November 2026 file, compiled through 31 October 2026. It ships as 7 zip parts (most around 69 to 70 MB, the last part about 50 MB). Inside are CSVs of live companies. A full pass of that snapshot had 5,711,779 rows.

A few gotchas I hit:

  • Some CSV header names have a leading space, so don't assume CompanyName is spelled exactly like that until you trim the headers.
  • Incorporation dates are dd/mm/yyyy, not ISO.
  • Each company can have up to four Standard Industrial Classification (SIC) codes. If you filter on a prefix such as 62 (computer programming, consultancy and related activities), keep the row if any of those four codes starts with 62.
  • "Live" includes some companies in liquidation or with a proposal to strike off, so I usually keep only Active.
  • Dormant shells often show accounts category DORMANT or SIC 99999. I drop those unless I want shells on purpose.

The snapshot has the registered name, company number, status, type, incorporation date, registered office (including postcode and post town), SIC codes, accounts dates, and a previous-name list. It doesn't include officers or filing history. Those need the official Companies House API and a free key from the developer site.

A small filter you can run yourself

Once you have a CSV part unzipped, a short Node filter looks like this:

import { createReadStream } from 'node:fs'; import { parse } from 'csv-parse'; const after = '2026-10-01'; // latest month of the 1 Nov 2026 snapshot const sicPrefix = '62'; const postcodeArea = 'M'; // Manchester area; this should not match ME or ML function iso(d) { const m = String(d || '').trim().match(/^(\d{2})\/(\d{2})\/(\d{4})$/); return m ? `${m[3]}-${m[2]}-${m[1]}` : ''; } function area(pc) { const t = String(pc || '').toUpperCase().replace(/\s+/g, ''); const m = t.match(/^([A-Z]+)/); return m ? m[1] : ''; } const rows = []; const parser = createReadStream('BasicCompanyData-2026-11-01-part1_7.csv').pipe( parse({ columns: (h) => h.map((x) => String(x).trim()), relax_quotes: true, relax_column_count: true }) ); for await (const r of parser) { const inc = iso(r.IncorporationDate); if (inc < after) continue; if (String(r.CompanyStatus).trim() !== 'Active') continue; const sics = [r['SICCode.SicText_1'], r['SICCode.SicText_2'], r['SICCode.SicText_3'], r['SICCode.SicText_4']] .filter(Boolean) .map((s) => String(s).trim()); if (!sics.some((s) => s.startsWith(sicPrefix))) continue; if (postcodeArea && area(r['RegAddress.PostCode']) !== postcodeArea) continue; rows.push({ name: r.CompanyName, number: r.CompanyNumber, inc, town: r['RegAddress.PostTown'], postcode: r['RegAddress.PostCode'], sic: sics[0], }); } rows.sort((a, b) => b.inc.localeCompare(a.inc) || b.number.localeCompare(a.number)); console.log(rows.slice(0, 20)); 

That's the idea, not a complete production script. Header names can vary slightly, you have seven parts to walk, and a full download is hundreds of megabytes. When I scanned all 5,711,779 rows that way, the run took about 26 minutes.

What the numbers looked like

For the latest month in that snapshot (incorporated on or after 1 October 2026, through 31 October 2026), with dormant companies excluded:

  • SIC prefix 62: 6,841 matches. The 20 newest were all incorporated on 31 October 2026 and all had accounts category NO ACCOUNTS FILED. The first few were BOOKEDTIK LTD (17554142, Ipswich, 62012 - Business and domestic software development), MACRODESK LTD (17554155, London) and LITTLE TREASURES INTERACTIVE LTD (17554158, Coventry).
  • Postcode area M (Manchester): 2,094 matches. The newest included DIRECT SPEC MOTORS LIMITED (17554162, Manchester, M16 9HW).

Those counts came from real runs against the official 1 November 2026 snapshot. October 2026 alone had 66,457 incorporations in the file.

If you sell accounting or formation follow-up, NO ACCOUNTS FILED plus a next-accounts due date is a useful flag. If you sell software, start with SIC 62 or a tighter code such as 62012. If you work a city, use a postcode area (M, EC, SW) or a post town (Manchester). Area M shouldn't match ME or ML.

The faster way

I got tired of downloading the full snapshot for every search, so I turned this into a tool on Apify: UK Companies House: New Company Leads, Officers and Filings. You can also open it in the Console with the example input.

The form is pre-filled with SIC 62, companies incorporated in the latest month of the snapshot, dormant companies excluded, and 20 results. That prefill run took 3.2 seconds, scanned the 6,850 SIC-62 rows for October (6,841 after the dormant filter), and saved the same 20 companies I listed above. A Manchester postcode-area search of that same month took 4.8 seconds and found the 2,094 matches. A later 5-company check of the published build finished in 2.2 seconds.

Typical last-year searches don't download the whole register. They read a precomputed index of the last 13 months (835,082 companies in this snapshot, from October 2025 through October 2026). If you search with no date bound, or you go older than that window, the tool falls back to a full snapshot scan. That path is slower, and there's a $0.10 fee for that fallback only.

It costs $0.002 per company saved, which is $2 per 1,000. Companies your filters remove are free. Officers and recent filings are optional and need your own free Companies House API key. New Apify accounts get $5 of free usage every month, so a small test run is covered.

The data is official public register data, licensed under the Open Government Licence v3.0. Use it for targeted outreach, not bulk spam, and follow UK GDPR and PECR.