Filter elements wear out on schedule. That single fact underwrites the whole vertical: replacement programs, media subscriptions, service contracts — recurrence built into physics.
It is also one of the smallest independent universes we have screened: 109 eligible US companies, in a landscape crowded with the brand-sites of global groups.
Scarcity changes the sourcing math, and the specimen shows how.
Filters load up and must be replaced. That single physical fact makes filtration one of the most naturally recurring verticals in industrial services.
Elements exhaust on schedule; media needs replacement; centrifuge wear parts have fixed service lives. An installed base is an annuity — and companies publish their replacement programs openly.
Global groups have consolidated aggressively for decades, operating through portfolios of acquired brands. Many brand-sites keep founder histories and local addresses with no obvious corporate parent on the homepage.
A large fraction of credible-looking “independent” filtration companies are subsidiaries — disclosed two clicks deep. Our specimen exclusion: “proud to be part of a global industrial group.”
The smallest of our ten specimen verticals. A list that misses thirty has missed more than a quarter of the market; every mis-included subsidiary wastes a meaningful share of outreach capacity.
We start from 100M+ classified domains sorted through 700+ industry categories, and run LLM analysis of every plausible candidate against your written thesis with quoted evidence.
When the universe is 109 companies, the difference between what a database happened to capture and reading everything is not marginal — it is most of the value.
The classic false positive in filtration is not the reseller — it is the acquired brand running an independent-looking site.
A founding story from 1958. A coherent specialty product line. Nothing above the fold suggests a corporate parent.
About page final paragraphs, press-release archives, privacy policy corporate-entity language. Documented group ownership zeroes scores; the quoted disclosure ships with its URL.
109 eligible US independents; representative companies shown in the specimen format with every classification quoted — including the group-owned brand sites we caught and excluded.
Open the Industrial Filtration & Separation specimen →From the 15-signal framework, scored 70/20/10 across Mandate Fit, Outreach Suitability, and Transition Context — with group ownership zeroing the total.
The load-bearing signal. Replacement-element programs, scheduled media changes, and standing consumables relationships — we quote the program language and distinguish real consumables recurrence from generic “service” claims.
Manufacturer, service house, or catalog reseller — the vertical's decisive three-way split. We weigh manufacturing evidence against service evidence against resale evidence; hybrids are recorded as their evidenced mix.
Food and beverage, pharmaceutical, municipal water, metalworking coolants, dust and fume — each end market carries its own buyer logic. We extract markets from case studies and application pages, not footer lists.
The brand-site gate, applied with quotes from about pages, news archives, and legal fine print. In a 109-company universe, each subsidiary correctly excluded protects a measurable fraction of outreach.
Which filtration OEM lines a company carries and on what stated terms. Line-card language is quoted exactly — for service-led theses, OEM service authorizations are among the strongest published assets a small filtration company holds.
Drawn from the published specimen — a live US screening run, anonymized.
Family-associated; twelve locations; food, beverage, and pharmaceutical end markets. “Strong positioning in high-performance decanter centrifuges and separators for solid-liquid separation across critical sectors… with a recurring revenue model.” The full signal extraction sits behind that sentence, quote by quote.
Family-associated, Pennsylvania, two locations; industrial, municipal, and military customers; ISO 9001:2015 certification captured as exact claim text. The evidence chain includes the company's own family-business language — succession context as published fact, not inference.
“Homepage lacks the obvious category keywords — keyword-driven databases would likely miss or misclassify this company.” Kentucky; founder-led; recurring revenue model evidenced.
Fume and pollution-control work is filtration by physics and thesis, but not by homepage vocabulary — the exact gap between keyword coverage and full-web coverage.
The site is excellent. The about page settles it: “the company is proud to be part of a global industrial group.” Quoted, linked, excluded, scores zeroed.
In a universe of 109, catching this class of brand-site is not hygiene — it is a material fraction of the entire screening value.
Three extractable signals across the whole site. The specimen flags it with a note instead of guessing — and delivers it, because the insufficient file is where a buyer's own market knowledge occasionally recognizes a real company hiding behind a bad website.
All ten counts use identical screening architecture — filtration sits second from the bottom. Exhaustiveness is achievable and therefore mandatory.
Filtration's companies do not share a language. The word “filtration” appears on perhaps half the sites a filtration thesis should include.
Triage collects every site whose category, service language, or application vocabulary intersects separation physics — over-collecting on purpose, because a missed independent in 109 is expensive.
Application pages, program language for consumables recurrence, line cards for channel position, about pages and legal fine print for ownership disclosures. Fifteen signals, quoted, sourced.
Manufacturer, service house, or reseller — decided from multiple pages, not one. Hybrids recorded as their evidenced mix with quotes.
Where the excluded population outnumbers the eligible, the who-owns-whom file becomes industry intelligence. See the false-positive guide.
Some facts are not published and therefore never estimated. The complete inventory is on the standards page.
Element margins, contract terms, revenue, EBITDA — never estimated.
Installed-base size, customer concentration — not published, not guessed.
A brand acquired last quarter reads as independent until it discloses — one honest reason monitoring re-runs exist.
Where evidence is too thin, the company lands in the insufficient-evidence file — visible, separate, never blended in.
Global groups continue acquiring — the eligible independent population is 109 and declining. Small universes reward craftsmanship over throughput.
The evidence format makes a two-day exercise of what would otherwise take months. Genuinely all — no sampling.
Consumables-led hybrids first, service houses second, specialist manufacturers third — with end-market columns cutting across.
Quote each company's own published programs. At this scale, template mail has no excuse — and the response-rate difference shows it.
Shows which global groups own what. Custom ICP re-runs (included) handle refinements: liquid-liquid only, dust and fume only, specific geographies.
Each re-run delivers the delta — new entrants, disappeared domains, fresh acquisitions among your 109. From €18,000/year per thesis. See thesis-universe mapping.
The filtration specimen ships the same day you ask. Send the thesis and we scope the full universe, or your custom cut of it.
Request the specimen report