One industrial sweep classified 367,478 domains worldwide and ended with 702 eligible independent US machining and fabrication companies. Every inclusion and every exclusion ships with a verbatim quote and a source URL.
Precision machining is one of the last genuinely atomized industrial trades in the United States. The reasons it fragments are the same reasons conventional databases undercount it.
A machine shop's capacity is its spindle list. Growth means capital equipment, trained machinists, and qualified processes — none of which scale quickly. The result is thousands of durable, geographically bound businesses.
Tolerances, envelope sizes, and materials define each shop. A five-axis aerospace shop and a Swiss screw-machine house rarely compete for the same purchase order.
AS9100D, ITAR registration, ISO 13485 and NADCAP accreditations take years to earn. They lock in customers — and lock the industry into small, specialized units.
Work follows proximity: OEM clusters in New England, the Midwest, and the Southwest each sustain their own long tail of suppliers few national lists ever capture.
Machine-shop websites are written by engineers for buyers who already know the vocabulary. They describe materials, machines, and tolerances — not the category labels a keyword index expects to find.
A homepage that says “wire EDM to ±0.0001″ in hardened tool steel” never says “precision machining company.” Keyword filters skip it; a full-page read does not.
Stamping, fabrication, tool & die, and contract manufacturing shade into machining. In our specimen run, a fifth or more of confirmed fits lacked the category's obvious homepage keywords.
The screening target here is not any site that mentions CNC. It is an operating business whose revenue visibly comes from making parts to other people's drawings.
An independent manufacturer producing custom or low-to-mid-volume machined components to customer specification — CNC milling and turning, EDM, grinding, or fabrication — for external industrial customers, as evidenced by its own site text rather than directory metadata.
Distributors of machine tools, captive plants of OEMs, and prototype-only studios sit outside that definition. Each exclusion in a delivery states which boundary was crossed, quoting the page that proves it.
The first pass is triage: confirm the domain belongs to a live operating manufacturer, not a directory, a distributor, or a parked page. In machining, that pass alone removed nearly a third of the 25,000-domain US set.
The second pass reads the surviving sites in full — capability pages, history pages, certification lists — and extracts each of the 15 signals as a quoted claim. Nothing reaches a score without a snippet behind it.
These are the real counts from the machining sweep — not projections. The same funnel re-runs on any custom ICP you define.
Our 15-signal framework applies to every vertical; these six do the heavy lifting in machining. Each is captured as quoted site text, never inference.
AS9100D, ITAR, ISO 13485, NADCAP: in machining these are the clearest quality proxies a website offers. We capture the exact claim text, because “AS9100D certified” and “AS9100-compliant processes” are different facts.
Machining is dense with explicit generational language — “fourth-generation,” “family-owned since 1942.” Roughly half of confirmed industrial fits carry founder or family evidence in their own words. We quote it; we never infer it.
Blanket orders, kanban programs, vendor-managed inventory, and long-run production language separate contract manufacturers from one-off prototype work. It is the difference between a book of repeat business and a queue of quotes.
Defense, aerospace, medical device, and energy exposure is read from case studies and named end markets — not guessed from keywords. Concentration in regulated markets usually travels with the certification set above.
How many principals does the site actually name? A shop that lists a president and nobody else reads differently from one naming operations, quality, and finance leads. We record what is published, and only that.
“A wholly owned subsidiary,” “part of the group,” investor press pages — any of these zeroes the score. About 1 in 10 keyword-perfect machining candidates fails here, which is precisely why the check exists.
Anonymized for the public site; client deliveries carry the company name, the full evidence set, and source URLs. The complete 20-row set is in the machining specimen.
Precision CNC machining for defense, aerospace, medical device, and automotive customers. Massachusetts, 58,000 sq ft facility. 16/16 evidence snippets verified against site text.
CNC machining, EDM, and fabrication for medical manufacturing, defense, and industrial equipment. Ohio; ISO 13485 on the certifications page; 14/14 evidence snippets verified.
Stamping, welding, and assembly with machining capability across two Ohio locations. The homepage never uses the category's keywords — a keyword-driven database would likely miss or misclassify it.
A keyword-perfect machining site that a filter query would happily list. The deep read found the ownership line, and the roll-up check zeroed the score before it could reach a shortlist.
The number that should worry any list buyer: a fifth or more of confirmed machining fits never use the category's obvious keywords on their homepage. Whatever screening starts from keywords starts a fifth short — before the first call is made.
Machining roll-ups are well underway: about 1 in 10 keyword-perfect candidates in our run was already group-owned. Yet 702 eligible independents remained visible after every exclusion.
The practical question is not whether targets exist. It is whether your list distinguishes the independents from the already-acquired — with proof.
| Screening question | Leading company databases | Full-web screening |
|---|---|---|
| Where does coverage start? | Companies previously indexed and tagged | 367,478 category domains — the entire active web |
| Shops without category keywords | Largely invisible to filter queries | Full-page reads recovered 5 of 13 specimen fits |
| Group ownership detected? | Often stale or absent | Roll-up check zeroes the score; exclusion documented |
| Why is each company on the list? | Tags, without cited text | Verbatim snippet + source URL per claim |
| Custom ICP re-screens | Limited to available filters | Plain-language thesis, re-run included |
For buyers whose thesis centers on generational transitions, we flag founder-associated, long-established, independently positioned businesses with an identifiable decision-maker and limited visible leadership bench.
Every element of that sentence is website-visible. We do not profile private individuals, and Transition Context carries only 10% of the composite score — see our standards for what we refuse to infer.
The same five steps run whether the mandate is a first universe map, an add-on radar for a platform, or a search-fund list.
Your acquisition thesis, in plain language, becomes the screen. “Independent US machining businesses with regulated end markets and repeat production work” needs no translation into database filters.
We start from 100M+ classified domains, isolate the 367,478-domain industrial category, and focus a 25,000-domain US triage that confirms roughly 17,300 live operating companies.
Each surviving site is read in full against the 15-signal framework: certifications as exact claim text, founder language verbatim, footprint, recurring-work indicators, and the ownership check.
Composite scoring weighs Mandate Fit at 70%, Outreach Suitability at 20%, Transition Context at 10% — and zeroes any company with group-ownership evidence, however good the rest looks.
A CRM-ready file with every claim cited, plus documented exclusions. Annual monitoring (from €18,000 per thesis) re-runs the universe and reports deltas; custom ICP re-screens are included.
A vendor that documents its limits is one whose positive claims you can weigh. Three apply squarely to machining.
Machine-shop websites do not publish revenue, and we do not guess it from page text. Facility size, headcount hints, and customer mix are reported as stated — nothing more.
We never claim a company wants to transact. No website signal supports that claim, and any vendor selling it is selling a guess dressed as data.
A sparse or dormant site limits what can be verified. Two of twenty specimen rows are flagged insufficient-evidence for exactly this reason — flagged, not silently dropped.
It is the eligible-independent count from our most recent full US machining sweep, and it moves. Shops get acquired, sites go dormant, new ones qualify.
Engagements deliver a fresh run, not a cached list; annual monitoring then re-screens the universe and reports what changed, with evidence for each delta.
Twenty real, anonymized rows: 8 top fits, 5 fits whose homepages lack category keywords, 5 documented exclusions, and 2 insufficient-evidence flags.
Each carries classification, scores, and verbatim quotes with page references. The machining specimen is published; client versions add names and URLs.
The deep read looks for ownership language everywhere it appears — about pages, footers, press releases, investor boilerplate. Any group-ownership evidence zeroes the composite score.
In machining, roughly 1 in 10 keyword-perfect candidates failed this check. Those exclusions are delivered too, so your team never re-discovers them by phone.
Yes. Your ICP is a paragraph of plain language, not a filter set, so “AS9100-certified shops citing 5-axis work and defense customers, no group ownership” is a valid screen.
Custom re-runs on any subset are included in every engagement — see pricing for what each tier covers.
A proof project starts from €4,900 — enough to test the method on your thesis before committing. The full universe with a deep shortlist starts from €9,900.
Annual monitoring runs from €18,000 per thesis and keeps the universe current, including custom ICP re-screens along the way.
Keep the database — it serves polished mid-caps well. Our edge is the long tail it structurally misses: engineer-written sites, keyword-free homepages, and unstated ownership changes.
Point us at a thesis you believe is covered, and a gap analysis shows what was missed or misclassified, with evidence per company.
Most machining theses widen into neighbouring trades — the same buyers, the same regional clusters, often the same certifications. Each of these verticals has its own screened universe with eligible-independent counts from the same industrial sweep.
One email starts it. We send the 20-row machining specimen the same day — including the five companies keyword screens would have missed and the five we refused to include.
Request the specimen report Browse it online