Firmographics for B2B ICP:
The 7 Filters That
Predict Closed-Won

Firmographics are the backbone of an ICP — the objective company attributes a rep can verify to gauge fit — but most founders use the wrong ones. They build firmographic filters around attributes that sound important (industry, revenue, headcount) without checking whether those attributes actually predict which deals close, ending up with an ICP full of impressive-sounding criteria that describe their market without predicting their wins. The firmographics that belong in an ICP are not the ones that describe a company most completely; they are the ones that most strongly separate the deals you win from the deals you lose. This guide is about the firmographic filters that actually predict closed-won — the seven that, across B2B, most reliably distinguish good-fit accounts from poor-fit ones — and how to find which of them matter most for your specific business. The goal is firmographics that predict, not firmographics that merely describe.

The distinction between predictive and descriptive firmographics is the whole game. A descriptive firmographic tells you what a company is (it is a 500-person manufacturing firm in Ohio); a predictive firmographic tells you whether it will buy and succeed (companies at this growth stage with this trigger close at three times the rate). Many firmographics are descriptive without being predictive — they characterize a company accurately but do not correlate with whether it becomes a good customer — and an ICP built on descriptive-but-not-predictive attributes aims your team using criteria that feel rigorous and do not actually sort fit from non-fit. The seven filters below are chosen because they tend to be predictive across B2B, but the deeper lesson is to always ask of any firmographic: does this attribute actually separate my wins from my losses, or does it just describe companies? Only the predictive ones earn a place in the ICP.

Firmographics: The Checkable Layer

Firmographics are to companies what demographics are to people — the objective, external attributes that describe an organization: industry, size, revenue, growth, location, structure, model. Their value in an ICP is that they are checkable: a rep can verify a company's firmographics quickly and objectively, which makes them the practical filter for fast account qualification. You cannot easily verify a company's internal pain or readiness at first glance, but you can verify its industry and size — so firmographics are the front-line screen that lets a rep rapidly assess baseline fit before investing in deeper qualification. This is why firmographics are the backbone of the operational ICP: they are the attributes you can actually filter a list on, score an account against, and check in seconds. The buying triggers and pain matter enormously, but firmographics are what make the ICP applicable at the speed and scale real targeting requires, which is exactly why getting the firmographic filters right — predictive, not merely descriptive — is so important.

The Vanity Firmographic Trap

The most common firmographic mistake is filtering on attributes that are impressive or intuitive but not actually predictive. Revenue is the classic example: founders filter for companies above some revenue threshold because bigger feels better, when in fact deal-closing may correlate far more with growth stage or a specific trigger than with raw revenue — so the revenue filter screens on the wrong thing while feeling rigorous. Industry can be a vanity firmographic too, when a founder assumes their product is "for healthcare" and filters on it, when the real predictor cuts across industries (companies with a specific operational characteristic, regardless of sector). The vanity trap is seductive because the impressive attributes feel like serious criteria, and an ICP full of them looks sophisticated — but if they do not separate wins from losses, they are decoration that actively misaims the team. The antidote is empirical: for every firmographic you are tempted to filter on, check whether it actually correlates with your wins, and keep only the ones that do. Sophistication in firmographics is not a long list of impressive attributes; it is a short list of genuinely predictive ones.

The 7 Filters That Tend to Predict

Across B2B, seven firmographic filters most reliably predict closed-won — though which matter most varies by business.

• 1 · Industry / vertical. When your product fits some industries far better than others, industry is strongly predictive — but verify it is real fit, not assumption, and watch for cross-industry predictors hiding underneath.
• 2 · Company size. Often predictive because the problem you solve becomes acute only in a certain size band — too small and the pain is absent, too large and you are outmatched. Find your band empirically.
• 3 · Growth stage and trajectory. Frequently more predictive than size: fast-growing companies feel scaling pains acutely and buy to solve them, while flat ones often do not. Trajectory beats absolute size for many products.
• 4 · Business model. How a company makes money shapes whether your product fits its operations — a predictor founders often overlook because it is less visible than industry or size.
• 5 · Technology stack / tooling. What a company already uses signals both technical fit (will you integrate) and readiness (do they buy tools like yours) — especially predictive for SaaS and technical products.
• 6 · Operational maturity / budget authority. Whether a company has the maturity to implement and the structure to buy — a predictor of whether a good-fit company can actually become a customer rather than stall.
• 7 · Geography / market. Sometimes predictive due to regulation, language, go-to-market reach, or market characteristics — though often less so than the others, and worth including only where it genuinely separates wins from losses.

These seven are the usual suspects, but the point is not to filter on all seven — it is to find which of them actually predict your closed-won and build your firmographic filter from those.

How Specific Should Each Filter Be?

A practical question once you know which filters predict: how precisely should you define each one? Too loose and the filter does not discriminate (defining size as "SMB to enterprise" filters nothing); too tight and you exclude good-fit accounts on arbitrary boundaries (defining size as "exactly 200–250 employees" cuts off fits at 199 and 260 for no real reason). The right precision is whatever the data supports as genuinely predictive — if your wins cluster sharply in the 100-to-500-employee band and drop off outside it, that band is the right precision; if the predictive range is broad, the filter should be broad. The error in both directions comes from setting precision by intuition rather than evidence: founders either hedge with vague ranges that do not filter or impose false precision that excludes good accounts. Let the win/loss data set the boundaries — define each filter as precisely as the data justifies and no more, so the filter captures the real predictive range without arbitrary cutoffs. This is also why firmographic filters should have soft edges rather than hard walls in scoring: an account just outside a predictive range is not a hard no, just a lower score, which avoids mechanically excluding accounts that sit near a boundary the data drew approximately.

The same logic applies to how many filters to use. More filters mean more precise targeting but also more accounts excluded and more complexity in applying the ICP. The right number is the few that genuinely predict — usually two to four firmographic filters carry most of the predictive power, and adding more beyond those tends to exclude good accounts for marginal precision gains. Resist the urge to filter on all seven dimensions just because you can; filter on the few that move your win rate, defined as precisely as the data supports, and let the rest go.

Why Combinations Beat Single Filters

The most predictive firmographic targeting usually comes not from any single filter but from a combination, because fit is often defined by the intersection of attributes rather than any one alone. "Mid-size companies" might convert at an average rate, and "fintech companies" at an average rate, but "mid-size fintech companies that recently raised" might convert at three times the average — a combination far more predictive than any of its parts. This is why analyzing firmographics one dimension at a time can understate their predictive power: the real signal lives in the interaction, where a specific combination of size, industry, stage, and trigger identifies a sharply-defined sweet spot. The practical implication is to look for the combination that defines your best-fit accounts, not just the individual predictive dimensions — your ICP's firmographic core is often a specific intersection ("companies that are X and Y and Z") rather than a list of independent filters. Founders who find this combination gain a targeting precision that single-dimension filtering cannot match, because they are aiming at the specific kind of company where all their predictive attributes coincide.

Finding the combination requires looking at your wins as whole profiles rather than as separate attributes — asking "what does the complete picture of my best accounts look like?" rather than "which single attribute predicts best?" Often a clear archetype emerges: a specific kind of company defined by several attributes together, which converts dramatically better than the broader population. That archetype, the intersection where your predictive firmographics coincide, is the sharp center of your ICP — and targeting it directly is far more effective than filtering on each dimension independently and hoping the overlap works out.

Finding Which Filters Matter for You

The seven filters are candidates, not a checklist — your job is to discover which ones genuinely predict your wins, which you do by analyzing your won versus lost deals across each dimension (exactly the closed-won analysis that finds an ICP from data). For each of the seven, ask: does this attribute separate my wins from my losses? Often you will find that two or three of the seven are strongly predictive for your business while the rest are noise — and your firmographic filter should be built from those two or three, not all seven. A common discovery is that the firmographics founders assumed mattered (revenue, industry) are weaker predictors than ones they overlooked (growth trajectory, tech stack), which is precisely why the empirical check matters: intuition about which firmographics predict is frequently wrong, and only the win/loss data reveals the truth. Build your firmographic ICP from the few filters your own data proves predictive, weight them by how strongly each separates wins from losses, and leave the descriptive-but-not-predictive attributes out — they only add noise. A short list of genuinely predictive firmographics is worth more than a comprehensive list of impressive ones, because the short predictive list actually sorts fit from non-fit while the long descriptive one just looks thorough.

A final caution: predictive firmographics drift over time, so the filters that predicted your wins last year may not predict them next year as your product, market, and competition change. The firmographic analysis is not a one-time exercise but a periodic one — re-run the win/loss analysis as your data grows and your business evolves, and update which filters you trust. A firmographic ICP frozen at an earlier stage slowly decays into describing the customer you used to win rather than the one you win now, which is the same staleness trap that afflicts ICPs generally. Keep the firmographic filters tied to current win/loss reality, and they stay predictive; let them ossify, and they quietly become vanity firmographics of a different kind — once-predictive attributes that no longer are, carried forward out of habit.

FAQ: Firmographics for B2B ICP