A company that registered on Monday has no accountant, no insurance broker, no payroll provider, no bank relationship, no website and no marketing agency. Four weeks later it has most of those, chosen under time pressure from whoever reached it first.
That window is the most valuable one in B2B sales, and the list of who's in it is public.
Three feeds, three different kinds of company
State registries. Every US state maintains a business registry, though not always under a Secretary of State — Arizona runs it through its Corporation Commission, Michigan through LARA, Virginia through the State Corporation Commission, and Alaska, Hawaii and Utah have no Secretary of State at all. Many publish the registry as open data on a Socrata portal, the platform behind a large share of municipal open-data sites. New LLCs and corporations appear within days of filing. Connecticut is the standout: nearly every new company comes with the email address the founder gave the state.
Municipal licence feeds. Cities publish business licences as they're issued. This catches a different population — the restaurant, the salon, the contractor, businesses that need a licence to open rather than a corporate entity to exist. Owner names and phone numbers appear here far more often than in state data.
SEC EDGAR Form D. When a company raises private capital under Regulation D, it files a notice with the SEC within fifteen days of the first sale. That filing is public and structured, and it names the amount raised, the date of first sale, the number of investors, and the executive officers with a company phone number.
Form D is the highest-intent feed of the three. A company that just closed a round is about to hire, buy software and outsource functions, and it has money on the day you call.
Here's an actual Form D record from a run this month:
{
"type": "signal",
"signal": "FUNDING_ROUND",
"sourceName": "SEC EDGAR Form D",
"companyName": "AI Planning Logic Inc",
"cik": "2090528",
"entityType": "Corporation",
"jurisdictionOfInc": "DELAWARE",
"yearOfIncorporation": "2022",
"incorporatedWithinFiveYears": true,
"address": "16192 COASTAL HIGHWAY",
"city": "LEWES", "region": "DE", "postalCode": "19958",
"contactPhone": "+1707888XXXX",
"industry": "Other Technology",
"industryTags": ["Technology & Software"],
"dateOfFirstSale": "2025-12-05",
"fundingAmountSold": 240000,
"investorCount": 3
}
$240,000 raised from three investors, a Delaware corporation under five years old, in technology, with a phone number on the filing. That's a qualified lead assembled entirely from a government record. (The phone is masked here; it's public in EDGAR.)
One caveat visible in that very record: dateOfFirstSale is December 2025, while the filing itself is recent. Form D is filed within fifteen days of the first sale, but amendments and late filings both exist, so read the sale date rather than assuming the filing date is the event date.
Why this isn't as easy as "the data is public"
Every source has a different schema. Connecticut's field names have nothing to do with Chicago's, which have nothing to do with EDGAR's XML. Normalising forty sources into one row shape is the actual work.
Registries are full of things that aren't new businesses. Biennial statements, renewals, amendments, name changes, dissolutions and reinstatements sit in the same feed as formations. A naive pull of "recent filings in New York" returns mostly companies that have existed for a decade. The filing type has to be filtered, and the vocabulary differs by state.
The same business appears more than once. A company files with the state and then takes a city licence. Two rows, one business. Deduplication has to survive punctuation, "LLC" versus "L.L.C.", and suite numbers written four ways.
Dates lie. Future-dated filings, typo'd years, and registries that publish the processing date rather than the filing date. If freshness is your entire value proposition, you have to defend it.
Contact fields are unverified. Registries publish what the filer typed. Nobody checks it. A raw registry feed contains dead numbers and misspelled email domains at a meaningful rate, and each one costs you sender reputation or a wasted dial.
That last one is worth solving properly: every phone in this feed is validated against the North American Numbering Plan and every email domain gets a live MX lookup plus disposable and role-address flags, before the row reaches you. A number in a reserved exchange or a domain with no mail server is labelled, not sold to you as a lead.
Volumes, measured
Actual output over a seven-day window:
| Feed | Volume | Contact coverage |
|---|---|---|
| New local businesses, 9 cities | ~190 | 100% with phone, ~30% with owner name |
| New Connecticut companies | ~860 | 100% with email, 99% with NAICS code, 95% with a usable address |
| Funded startups (Form D) | ~250 | 94% with company phone, 100% with an officer's name |
About 1,300 qualified new businesses a week from three presets. Not a database of everyone — a feed of who is new.
Who this is for
Anyone selling to new businesses. Accountants, bookkeepers, insurance brokers, payroll, POS systems, business banking, commercial cleaning, signage. The entire category of vendor a business chooses once, in its first month, and then keeps for years.
Agencies and web developers. A newly registered company with no website is the least ambiguous lead in existence.
B2B SaaS selling to funded startups. Form D plus a five-year incorporation flag plus an amount raised is a better qualification filter than anything in a generic lead database, and it's fresher.
Commercial real estate and equipment. New licences in a city are the leading indicator for space and fit-out demand, by industry and by ZIP.
Market researchers. Formation rates by state, city and industry give you the composition behind the headline number — the Census Bureau publishes Business Formation Statistics weekly, but not broken down the way your question needs it.
Running it
US New Business Leads ships with three presets so the first run needs no configuration:
{ "preset": "funded_startups", "maxResultsPerState": 100 }
Custom targeting takes states, cities, industries, a date floor, and contact-quality filters:
{
"states": ["CT", "NY"],
"registeredSince": "2026-08-01",
"requireContact": true,
"contactQuality": "deliverable",
"monitorMode": true
}
monitorMode is what turns it from a scraper into a feed: it remembers what it has already delivered and returns only businesses you haven't seen. Put it on a daily schedule and each run is that day's new companies, deduplicated against every previous run.
From the API:
curl -X POST "https://api.apify.com/v2/acts/lergassy~us-business-filings/runs" \
-H "Content-Type: application/json" \
-H "Authorization: Bearer YOUR_APIFY_TOKEN" \
-d '{"preset":"funded_startups","maxResultsPerState":100}'
Every row carries a lead score from 0 to 100 built from freshness, contact completeness, address quality and entity type. Sort by it, or set a floor and never look at the rest.
What you don't get
Not every state. Coverage follows which registries publish machine-readable open data, and several large states don't. Where a state is missing, it's missing because the data isn't published, not because it's hard to scrape.
No decision-maker names beyond what's filed. State filings name a registered agent, sometimes an organiser. Form D names executive officers. Neither gives you a VP of Engineering.
No revenue, headcount or tech stack. This is a formation feed, not an enrichment product. It tells you a company exists and is new. What it does is a separate lookup.
No email for most municipal records. Cities publish phones far more reliably than emails. Connecticut is the exception that makes email-first outreach viable at all.
The thing that makes it work
Freshness is the entire product. A newly-registered-business list that's six weeks old is a list of companies that already chose a vendor, and it's worth close to nothing.
Which is why this belongs on a schedule rather than in a one-off run, and why deduplication across runs matters more than volume in any single one. Reaching a business in week one is a different conversation from reaching it in week seven.
Actor: apify.com/lergassy/us-business-filings
Related: Email & Phone Verifier runs the same checks over any contact list you already have, including the output of another actor.
Top comments (0)