Firmographic data is the set of structured attributes that describe a company as an organization: size, industry, geography, legal structure, ownership, age and funding stage. It is the company-level counterpart to demographic data about people, and it is the raw material for account selection. When a team defines an ideal customer profile, it is mostly defining firmographic boundaries plus evidence from existing customers, which is exactly how Salesforce frames an ICP as an organization-level description distinct from a person-level persona (Salesforce).
The practical problem is not knowing what firmographic data means. It is that many firmographic fields look simple but hide structure that determines whether your account selection works. A headcount without a date, a country without knowing whether it is headquarters or service territory, or a record that silently points at a subsidiary instead of a parent will all pass a schema check and still produce a bad target set.
Core firmographic attributes and what each one actually encodes
| Attribute | What it should encode | Common trap |
|---|---|---|
| Legal name and entity | The specific legal entity, not the brand | Brand names span multiple entities |
| Headcount | Reported value, observation date, and derived range | Undated values; treating headcount as revenue |
| Industry | Classification code plus free-text descriptor | One company, many plausible codes |
| Headquarters location | Where the company is operationally based (registered domicile can differ) | Confused with where it sells or operates |
| Service or operating geography | Markets the company sells into or serves | Often missing entirely, HQ used as a stand-in |
| Company age | Founding year of the entity in question | Parent founding year applied to a new subsidiary |
| Ownership structure | Independent, subsidiary, PE-backed, public | Scoring a subsidiary as a small independent company |
| Funding stage | Latest known round with date | Stale rounds presented as current stage |
Two of these deserve expansion because they cause the most selection errors.
Headquarters versus service geography
Headquarters location answers an organizational question: where the company is operationally based, which is not necessarily its registered legal domicile. Service geography answers a commercial one. A logistics company headquartered in Rotterdam may serve customers across four continents; a restaurant group registered in Delaware for legal reasons may run its operating headquarters and every location in Texas. If your territory rules filter on headquarters country, you will include companies that cannot buy in your region and exclude companies that can. Store them as separate fields, and when service geography is unknown, record it as unknown rather than copying the HQ value into it. That copy is one of the easiest ways for this distinction to be silently destroyed.
Headcount needs a date and a range, and it is not revenue
Headcount is an observation, not a constant. A value of 180 observed in January and a value of 240 observed in October are two different facts, and the difference between them is itself a useful signal. Store the reported number, the date it was observed, and a derived range band (for example 101 to 250) used for segmentation. The range absorbs normal reporting noise; the dated exact value lets you see direction.
One discipline worth stating plainly: headcount is not a proxy for revenue and should not be scored as one. A 40-person trading firm and a 40-person nonprofit have nothing in common financially. If revenue matters to your selection, source revenue or a defensible revenue band as its own field, and let it be unknown when it is unknown.
Subsidiaries and entity resolution
Account selection quietly fails when the record you scored is not the entity you meant. A subsidiary with 60 employees inside a 12,000-person parent can look like a mid-market prospect while procurement, budget and vendor approval all sit at the parent. The reverse also happens: you disqualify a "small" record that is actually the buying entity for a large group.
The workable fields are ownership type (independent, subsidiary, parent, PE-backed, public), a parent identifier when one exists, and a note on which entity holds buying authority when that is knowable. You will not resolve every hierarchy, but recording that a record is a subsidiary, even without full detail, prevents the worst misreads. This kind of structural gap is a typical target for waterfall data enrichment, where multiple sources are consulted in sequence until the entity relationship is confirmed or explicitly marked unresolved.
A field ledger: recording meaning and date, not just value
A useful pattern is to treat each firmographic field as a small ledger entry rather than a bare value: what was observed, when, from what kind of activity, and by which process. This mirrors the structure of the W3C provenance model, which separates entities, activities and responsible agents (W3C PROV-O); the ledger described here is an illustrative implementation of that idea, not a standards claim.
A minimal per-field ledger:
| Column | Example |
|---|---|
| Field | headcount |
| Value | 240 |
| Observed date | 2025-10-14 |
| Meaning note | Global full-time, per company careers page |
| Collection method | Automated web observation |
| Confidence | Medium |
The "meaning note" earns its keep. Two records can both say headcount 240 while one counts global full-time staff and the other counts one office including contractors. Selection built on values without meaning is selection built on coincidence. For a deeper treatment of tracking where field values come from, see data provenance.
Handling missing data honestly
Missing firmographic fields are inevitable. The failure mode to avoid is the silent-zero problem: a scoring rule that treats an empty headcount as failing the size criterion, so accounts you never evaluated get disqualified as if they were evaluated and rejected. Three rules keep this honest:
- Distinguish unknown from failing. An account missing service geography has not failed a geography test.
- Set a coverage threshold. A suggested rule: an account needs at least four of six core fields observed before it can be qualified or disqualified. Below that, it routes to enrichment, not to a decision.
- Date-expire stale values. A headcount observed three years ago is closer to unknown than to current. Pick an expiry window per field, treat it as an operating choice rather than a measured decay rate, and demote expired values to unknown-stale.
Worked example: a small eligibility table
Illustrative example with a hypothetical target profile: companies with 101 to 500 employees, serving the North American market, independent or parent entities, founded before 2021. Each of the four criteria is worth 1 point if met, 0 if failed, and marked U if unobserved. Eligibility requires 3 or more points with at most 1 unknown.
| Account | Headcount band | Serves NA | Entity type | Founded pre-2021 | Score | Unknowns | Decision |
|---|---|---|---|---|---|---|---|
| A | 101-250 (1) | Yes (1) | Independent (1) | 2017 (1) | 4 | 0 | Eligible |
| B | 251-500 (1) | Unknown (U) | Parent (1) | 2015 (1) | 3 | 1 | Eligible, enrich geography |
| C | 51-100 (0) | Yes (1) | Subsidiary of 8,000-person parent (0) | 2019 (1) | 2 | 0 | Not eligible at this entity; check parent |
| D | Unknown (U) | Unknown (U) | Independent (1) | 2022 (0) | 1 | 2 | Insufficient data; route to enrichment |
The arithmetic is exact: A scores 4 of 4; B scores 3 with one unknown and stays eligible under the rule; C scores 2 and fails, but the subsidiary flag triggers a parent-entity check rather than a permanent disqualification; D has two unknowns, so no decision is made at all. Account D is the one a silent-zero system would have rejected without ever evaluating it.
Note the tradeoff in the coverage rule itself. A strict rule (no unknowns allowed) shrinks your addressable set and inflates enrichment cost. A loose rule (any two fields suffice) admits accounts on thin evidence. Tune the threshold to the cost of a wrong inclusion in your specific motion, and revisit it as your field coverage improves. Firmographic eligibility is also only the first gate; behavioral evidence such as intent data and downstream lead scoring belong after the structural filter, not mixed into it.
FAQ
What is the difference between firmographic and technographic data? Firmographic data describes what a company is: size, industry, geography, entity structure, funding stage. Technographic data describes what a company uses. Firmographics typically define eligibility; technographics refine fit and timing within the eligible set.
Should I store exact headcount or a range? Both. Keep the exact reported value with its observation date as the record of truth, and derive a range band for segmentation and routing. The range is stable against reporting noise; the dated exact values reveal growth or contraction between observations.
How should missing firmographic fields affect account scoring? Score only what was observed, mark missing fields as unknown, and require minimum coverage before qualifying or disqualifying an account. Records below the threshold route to enrichment. Treating missing values as zero disqualifies accounts you never actually evaluated.
Build the company layer before the selection layer
Account selection is only as good as the firmographic records underneath it: dated headcounts, separated HQ and service geography, resolved entities, and honest unknowns. AstroFabric's autonomous agents take an objective and a target profile, discover matching companies, enrich the firmographic fields with dates and confidence attached, and deliver structured records into your CRM or operational sheets. Start with the data infrastructure for prospecting overview, or sign up and run your first objective-to-dataset build against your own eligibility rules.