For people who find companies for AI data buyers. Firmographics are the first filter in any sourcing list: industry, size, location, age, ownership and status. This guide explains the standard fields, how they line up with the rules buyers publish, where providers get them, how to test a provider before you pay, and the layer classic firmographics leave out: what data a company is likely to hold.
Already have a target? Qualify it against buyer rules in about 20 seconds.
Firmographic data describes a company the way demographic data describes a person: the plain facts that let you sort organizations into groups.
Firmographics answer six questions about a company: what it does, how big it is, where it is, how old it is, who owns it and whether it is still operating. Sales and marketing teams use those answers to build territories, score accounts and size a market. A team that sources companies for AI data deals uses the same answers for a narrower job: deciding which companies clear a buyer's published bar before anyone spends an hour on them.
Most providers deliver a firmographic record as one row per company, keyed on a name, a web domain and an internal ID. The fields below appear in almost every product, under slightly different names. Each tile notes what the field means for a sourcing list, not only what it means for a sales team.
A classification code, often NAICS or SIC in the US, or the provider's own taxonomy, with a primary industry and sometimes secondary ones. It decides which buyer thresholds apply and what records the work produces.
Usually a band, such as 11 to 50 or 51 to 200, sometimes a point estimate. Ask whether it counts full-time staff, all workers or people who name the company on a profile. The difference decides whether a firm clears a rule.
Reported for public companies and modeled for almost everyone else. Useful for ranking, but for a private firm it is an estimate built from other fields, so treat it as a hint rather than a fact.
Headquarters address, country and state, sometimes other offices. Language is rarely a standard field, yet one buyer lists primarily English. It is usually inferred from the website or the country.
The year the company was founded or registered. A rough proxy for how many years of records exist, and a poor one when a company migrated systems, changed its retention policy or was re-formed after a merger.
Private, public, investor or private equity backed, nonprofit, or a subsidiary with a named parent. Ownership decides who signs a data deal, which can be a parent company far from the team you know.
Active, dormant, dissolved, merged or acquired. It is the field most often out of date, and one of the most important when buyers also look at companies that are winding down.
The web domain, registry numbers and the provider's own company ID. The domain is the most practical key for joining sources, because almost every operating company runs a website.
AI data programs publish who they want, and most of their rules read like firmographic filters. The fit is close but never exact, and the gaps are where sourcing lists go wrong. Rules below are as published, checked 7 October 2026.
| Published rule | Program | Firmographic field | Where the standard field falls short |
|---|---|---|---|
| 30+ employees (its referral posting says 30 to 200) | micro1 | Employee count | A band such as 11 to 50 straddles the line. A count built from online profiles can undercount a firm whose staff rarely keep one. |
| 20+ full-time US office employees | Mode | Employee count plus employee location | Total headcount includes contractors, part-time staff and people abroad. Mode's rule counts full-time US office employees only. |
| Accounting firms 10+, law firms 6+ | Mode | Industry plus employee count | A firm coded as general professional services loses the lower threshold. Check the industry of small firms by hand. |
| US first, then other Western markets | micro1 | Headquarters country | Headquarters is not always where the team or the records are. A US holding company can own a team that works elsewhere. |
| Primarily English | micro1 | Rarely a field; inferred | Inferred from website language or country. A bilingual firm, or a US firm whose records are partly in Spanish, needs a manual look. |
| Several years of records the company owns | Mode | Founding year | Founding year says nothing about records still held. A system migration, a retention policy or an acquisition can cut usable history to a few years. |
| Records the company owns | Mode | Ownership and parent linkage | A subsidiary may not control the systems its team uses. Parent linkage tells you who to ask, not who owns the records. |
| Mature operations, documented processes, modern software tools | micro1 | No standard field | Not a firmographic question at all. It needs a view of systems and documented knowledge, covered further down this page. |
| Any vertical | Grepped | None | No size, country or years minimum is published, so firmographics cannot rule a company in or out. |
Every target that misses a published rule costs time you cannot get back. These are the amounts programs publish, as published, checked 7 October 2026. They are ranges and referral terms, not offers.
micro1 pays a referral after onboarding plus a minimum revenue threshold, at its sole discretion and with clawbacks, and its terms forbid sharing payouts with companies or posing as micro1's partner. Mode publishes "Earn $50K per referral" and Grepped "Refer for another $10K". Read each program's terms, summarized on data referral programs, before you introduce anyone.
No provider observes companies directly. Every record is assembled from a handful of source types, each strong on some fields and weak on others. Knowing which source sits behind a field tells you how far to trust it.
Incorporation records and annual filings held by states and national registries.
What a company says about itself on its home, about, careers and contact pages.
Company profiles claimed and edited by the company, trade association member lists and professional network pages.
Public company reports, nonprofit tax filings and other mandatory disclosures.
Press releases and reporting on acquisitions, mergers, funding, closures and rebrands.
Values a provider calculates from other fields when no source states them, common for revenue and headcount.
Providers combine these sources and resolve them into one company record. That merge is where most of the quality lives: matching a registry entity to the right domain, deciding which headcount wins when two sources disagree, and linking a subsidiary to its parent. For large public companies the sources mostly agree. For a 40-person private firm they often do not, and that is exactly the size range AI data programs publish rules for.
Firmographic data reaches buyers through a few kinds of business, each built for a different first customer. Knowing the category helps you predict what a product does well before you sit through a demo.
Company and contact databases built for sales teams, usually priced per seat. Firmographics are the backbone, and much of the value sits in people and their contact details, which a company-level sourcing team may not need.
Built around credit and risk decisions. Typically strong on legal entities, ownership trees and financial standing, and priced for finance and procurement teams.
Collect official filings across states or countries and resell them in one format. Authoritative on legal status, thin on what a company actually does day to day.
Build company records from websites and other public sources, sold per lookup or in bulk. Often broad on domains and small firms, with fields limited by what sites publish.
Fill missing fields on records you already hold, usually by domain. Convenient inside a sales workflow, and limited to what their underlying sources contain.
Industry lists, association member files and one-off datasets. Can be deep in one vertical, with uneven freshness and licensing terms that vary file by file.
Most searches for the best B2B firmographic data providers come from sales teams, and the products reflect that: the value is in contacts, buying intent and CRM sync. A team that sources companies for AI data deals needs something else. It needs to know whether a company clears a size, country and history rule, whether the company still operates, and whether its work produces the records buyers take. Contact data matters less than you might think, because introductions in this market work through people you already know and through each program's own referral process, with the company's consent.
That difference changes how you read a provider's pitch. Coverage of the largest companies, intent signals and seat counts are sales features. For sourcing, the questions are narrower and harder: how reliable is the size band for a 25-person firm, how quickly does a closure or an acquisition show up, and can you see where each value came from.
A demo shows the companies a provider covers well. A test set shows the ones you need. Run these six tests on any provider before you sign, using companies whose real facts you already know.
Pick a few dozen companies you know well: small private firms close to the 20 and 30 employee lines, a firm acquired last year, one that closed, one that rebranded, one subsidiary, and a few accounting and law firms. Write down the true values before you look at the provider's records, so the product cannot anchor your judgment.
Count how many of your test companies appear at all, then how many have a size value that is more than a default band. Buyer programs publish rules at 6, 10, 20 and 30 employees, so coverage of firms under 200 people matters far more to you than coverage of the largest companies.
Ask when each field was last confirmed, not when the record was last touched. A record updated last week can still carry a headcount from years ago. A provider that keeps a date on each field lets you set your own cutoff and drop stale values.
Look up the acquired and closed companies in your set. Does the provider show them as active, dormant, acquired or dissolved, and how long after the event did it change? A list that treats a closed company as active sends introductions to an empty inbox.
For any value that decides a rule, ask where it came from: a registry filing, the company's own page, a self-reported profile or a model. A headcount with a source and a date is worth more than a confident number with neither.
Check whether you may use records to qualify companies for third parties, store them, or pass a company's details to a buyer program. Many data licenses restrict resale and sharing. Then compare price models: per seat, per record, per lookup or annual credits, and what happens to credits you do not use.
Twelve questions that separate a product built for your job from one built for someone else's.
Send these in writing before the demo, so the answers come from people who know the data rather than from a general pitch. Then check two or three of the answers against your own test set. A provider that cannot say whether a headcount was reported or modeled is telling you something useful about every row it sells.
The questions lean toward small private companies, status and evidence, because those decide whether a sourcing list holds up. Coverage of large public companies is rarely the problem: their facts are published, and every provider has them.
The last two questions are about your own risk. A license that forbids sharing company details with third parties may not suit a broker whose whole job is introducing companies to buyer programs. And contact data about people brings privacy obligations that company-level data does not.
None of these questions has a single right answer. A provider that says "modeled, and we label it" is more useful to you than one that cannot say.
Two companies can share every firmographic field and still be worth very different amounts of your time.
Take two US firms in the same industry code, each with 80 employees, founded in 2008, privately owned and active. The first runs its work through a ticketing system, a CRM, a written knowledge base and a code repository with years of history. The second runs on email and a shared drive, and replaced its main system two years ago. A sales database scores them the same. An AI data buyer does not.
Buyers pay for records of real work that never appeared on the open web: SOPs and knowledge bases, support tickets and their resolutions, CRM histories, project files and code with its history. What matters is depth, continuity and how well the records connect, not how big the company looks from outside. The full picture is on what data AI labs want.
That is the layer a firmographic record does not have. It can tell you the company is the right size in the right place. It cannot tell you whether the company holds anything a buyer would take.
None of the right-hand questions has a standard firmographic field. Answering them used to mean reading each company's website by hand, one target at a time.
The Data Asset Score rates any company from 0 to 100 for the data AI buyers want. Each result comes with a grade, the data the company likely holds, its history and an activity status: active, winding down or acquired, parked, or unreachable.
The score is built on our index of 102 million domains, 99.99% of the active internet, with domain history. The same lookup works for a 15-person law firm and a public company, as long as the company runs a website, and the history shows companies that were active before and are not anymore.
Every score looks at the nine factor groups above. We publish the groups; how they combine is our own. What you act on is the output: the score and grade, the likely data assets, the history and the status.
Use it next to whatever firmographic source you already have. Firmographics answer "does this company clear the rule?" The score answers "is this company likely to hold records a buyer wants, and is it still operating?" Together they turn a long list into a ranked shortlist.
Status deserves its own note. Buyers of wind-down and archive data exist: in August 2026 Google agreed to pay $10M for bankrupt Spirit Airlines' internal data and micro1 filed a $12.5M competing bid, with approval not confirmed as of 7 October 2026. We do not sell a wind-down list. Every score simply shows the company's activity status, so a company winding down or acquired is flagged wherever it appears.
Scores are estimates from public signals. They are not valuations, not offers, and not proof that a company wants to sell. Only a buyer can price data, after it sees a manifest and samples.
Fictional company and values, shown only to explain the fields a result contains. Not a real result, a valuation or an offer.
The order matters. Firmographics are cheap to filter on and rule out most companies quickly. The data layer then ranks what is left, and the check against buyer rules confirms the best targets with the company's own words.
Last checked: 8 October 2026. Every plan uses the same index and returns the same fields. The difference is volume, lists and export.
For a broker who scores shortlists from a personal network or a firmographic export, and wants the data layer on every target before an introduction. Scores companies only; lists start on Pro.
Choose BasicAdds company lists for 20 US sectors, always current, through the API as JSON, up to 100 companies per call with paging.
Choose ProFor sourcing teams that score whole firmographic segments every month. Adds the one-file bulk CSV export of a whole list, and new segments on request.
Choose ScaleA business that sells descriptive facts about companies: industry, employee count, revenue, location, founding year, ownership and operating status, usually keyed on a company name and a web domain. Providers assemble records from registries, company websites, self-reported directories, filings, news and their own estimates, then sell them per seat, per record, per lookup or as a file.
Employee count, country, industry, founding year and operating status, because buyer programs publish rules on them. As published, checked 7 October 2026, micro1 lists 30+ employees, US first and primarily English, and Mode lists 20+ full-time US office employees (accounting firms 10+, law firms 6+) and several years of records the company owns.
Not necessarily. Sales products are built around contacts, buying intent and CRM sync. Sourcing for AI data deals needs reliable size bands for small private firms, a current activity status and a view of what records a company holds. Test any provider on companies you already know, close to the published size lines.
Build a test set of a few dozen companies you know, most of them private firms with 10 to 200 employees. Count how many the provider has at all, how many carry more than a default size band and how many show the right status. Then ask whether each size value was reported, self-reported or modeled.
No. It adds the layer firmographics do not have: a 0 to 100 score for the data AI buyers want, the data a company likely holds, its history and its activity status. Use your firmographic source to filter on size and location, then score the shortlist by domain.
No. A score is an estimate from public signals. It is not a valuation, not an offer and not proof that a company wants to sell. Only the company can decide to sell, and only a buyer can price data, after it sees a manifest and samples.
No. There is no free plan and no trial. The Data Asset Score demo is free and limited per day, and the check against buyer rules is free. Paid plans start at $99 a month for 5,000 lookups, paid by PayPal or card, and the API key appears in your dashboard after payment.
No. Everything is company level: scores, likely data assets, history and activity status. We do not sell names, email addresses or phone numbers of people, and the check against buyer rules never contacts the company it reads.
Try any company in the free demo, check a target against published buyer rules, or pick a plan when you are ready to score shortlists through the API.