New company data for lenders, brokers and fintechs
How can a lender assess a company that is only weeks old? A company formed last month has no accounts, no filing history and no credit file worth the name.
Measured on 20,000 real companies that our model had never seen, followed for three years, including the ones that closed. How we know
The first accounts arrive up to 21 months later. Salenovo scores new companies on day zero from what does exist: formation channel, address type, sector, director history on the register, and then the early trading signals (a verified website, registered premises, an ICO registration, a first filing).
What you get
Signals before accounts
Most new companies never trade: about three in five, even since the 2024 fee rise. Knowing which third is which, on day zero, is the difference between a book of start-up loans that performs and one that doesn't.
Formation channel
A company formed at a mass-formation address with a 21% historical trading rate is a different proposition from one formed at its own trading premises. We know both.
Verified, dated, sourced
Every signal carries its source and date: the page the website was verified from, the council register the premises came from.
National, by API
For lenders the product is a feed, not a local list: every incorporation, scored, with signals updated as they arrive.
What one decision is worth
One avoided default on a £25,000 start-up loan is worth more than a year of the national feed; one performing loan written to a company a thin credit file would have declined is the upside. The score separates the two at day zero.
Why a ranked lead is worth more than a raw one
Half of new companies don't last, and most never trade. Of the 1,178,421 companies formed in the UK at their own address between January 2022 and September 2023, 49% were no longer on the register three years later and only 35% had ever filed trading accounts. That's the Business Survival Index, council by council. A list of every new company is mostly companies that will never trade, and half of it will be gone within three years.
We tested it. We took 20,000 companies formed in 2022 and 2023, ranked them with a model that had never seen them, and followed every one for three years, including the ones that closed. 89% of the new companies we ranked highest were still registered three years later. For all new companies it was 49%, and for the ones we ranked lowest 27%. Over those three years a company we ranked highest spent 27 months trading. The average new company spent 12. That's 2.3 times the value per lead. See the full comparison, and what it's worth to you.
Why 49% here and 51% in the Business Survival Index: this test is a random sample of every company formed in 2022 and 2023, including those registered at formation agents' addresses, which close more often. The index counts companies at their own address only. Both are measured the same way, three years after each company was formed.
Salenovo's score is built on outcomes like these: companies followed from the day they were formed to whether they went on to trade. Tested on companies it had never seen, the top band (A) went on to trade 67% of the time and the bottom band (E) 15% (method and held-out results). So the top of each week's list is the companies most likely to be trading next year and the year after: clients who need their accounts done every year, not a name that's gone by the time you call.
A worked example: Chelmsford
UK last week: 15,020 incorporated, about 6,209 expected to trade, 17% at mass-formation or virtual-office addresses. See the Chelmsford page, see a real week's list, or compare any area.
Against what you do today
| Raw register dump | Companies House by hand | Salenovo | |
|---|---|---|---|
| What arrives | Every incorporation, unscored | Credit-reference file: thin or empty for a new company | Day-zero score plus early trading signals, refreshed |
| Basis | None | Filings that don't exist yet | 30,000 companies with known outcomes; on held-out companies formed since the 2024 fee rise, band A traded about 67% of the time and band E about 15% |
| Delivery | CSV | Per-company lookup | Weekly file or API |
Business-to-business email to a limited company or LLP is permitted under PECR without prior consent, provided you say who you are and offer an opt-out. Every row says where the address came from. No director names, no home addresses.
Your first week
- The weekly national file delivers every incorporation with its band, entity type, formation channel and address type; a daily feed URL or API can be set up for a founding customer who needs it.
- Join on company number to applications and to your book; the band is a day-zero prior where no filing history exists.
- Use the early signals as they arrive: verified website, registered premises, ICO registration, first filing.
- Watch the running-accuracy table on the method page: each scored month is reported as its outcomes become known.
- Feed outcomes back if you can; the model is refit as cohorts mature.
Two short messages that work (an email and a letter) and the one line that matters for your trade are on the templates page.
Questions
How accurate is the day-zero score?
On held-out companies formed since the 2024 fee rise, band A (the top fifth) traded about 67% of the time and band E (the bottom fifth) about 15%, against a base rate of about 39%. The full method and the running accuracy as cohorts mature are published on this site.
What early signals do you track?
A verified website (with the company's registered number found on it where possible), a registered food premises, an ICO data-protection registration, a resolving domain, the formation channel and address type, and register events as they happen.
Does the list contain personal data, and what am I allowed to do with it?
No director names, no dates of birth, no nationalities, no home addresses and no personal email addresses: the list is about companies, not people. That matters for you as well as for them. When an organisation gets personal data from somewhere other than the person, UK GDPR says it has to tell them who has it, why, on what basis and where it came from, within a month or at first contact. A bought list of named directors starts that clock for the buyer. A list of companies never starts it. What you get instead is the registered office laid out for a letter, the company's own website once we have checked it is theirs, and the generic mailbox shown on it (info@, hello@), never a named person's.
So can I email these companies?
Yes, if they are limited companies or LLPs. The ICO treats those as corporate subscribers under PECR, so you may email them without prior consent as long as you do not conceal who you are and you give a valid address to opt out. Sole traders and ordinary partnerships are different, and need consent or the soft opt-in; everything on our list is a limited company or an LLP. Two things are yours to do whatever the list says: stop the moment somebody asks, because the right to object to direct marketing is absolute, and screen numbers against the TPS and CTPS registers before you ring. The ICO fined one company £160,000 in May 2026 for calling registered numbers.
Where does the data come from, and can a company be removed?
The Companies House public register, the Office for National Statistics, the Food Standards Agency, the Environment Agency and police.uk, all under the Open Government Licence, plus each company's own public website, fetched by a crawler that identifies itself and obeys robots.txt. Every row carries its source and the date. Any company can ask to have its contact details, or its whole record, taken out: salenovo.com/remove, honoured within five working days. Our own privacy notice is at salenovo.com/privacy.
Method and accuracy: how the numbers are made. Incorporations from the Companies House public register, mapped to councils with the ONS Postcode Directory; "expected to trade" from the historical outcomes of 2.5 million companies formed in 2022, 2023 and 2024, by formation address and sector; websites verified by fetching the company's name-matching domain and checking it against the company; food premises from the Food Standards Agency's register. All Open Government Licence. Every row in the paid list carries its source and date.
For advisory work: the growth list
The weekly list is new companies. The other list is established companies (every A and B one three or more years old) at the moment the public record says they are scaling: hiring, growing turnover, raising money, borrowing, winning contracts or taking on premises, each with the reason in one line. It is the growth list: established companies that are scaling: £250 a month for a region or £1,000 for the UK, monthly, stop at any month's end, and the first two weeks are free. The weekly list and the growth list side by side.
Other uses: Accountants & bookkeepers · Insurance brokers · Web, design & marketing agencies · Hospitality & food-service suppliers · Commercial property & serviced offices · B2B suppliers: telecoms, payroll, EPOS, merchant services · Councils, LEPs & business improvement districts · Business owners