Your committed capital buys a fixed number of months, and the shared databases sell the same export to every searcher in your cohort. We screen the entire active web against your thesis in the first weeks — so the rest of the search is spent calling, not compiling.
A search is a fixed budget divided by a monthly burn. Every sourcing month that produces recycled names is a month subtracted from the only outcome that matters — one closed acquisition.
You raise search capital on a thesis and a sourcing story. Investors back the story partly because you promised proprietary deal flow — companies they will not see in a banker's process.
The licensed database yields a few hundred filtered names. Sequences go out, interns enrich rows, response rates look survivable. It feels like a machine. It is a queue.
The queue empties. The same polished mid-size companies every profile database indexes have now heard from you, your cohort, and three consolidators. Reply quality drops before reply volume does.
Intermediated deal flow fills the calendar because it is available, not because it is right. Whatever arrives through a broker is, by definition, not proprietary — and rarely priced as if it were.
Now the funnel math is public: owners contacted, conversations held, IOIs signed. If the top of the funnel was never a census, no amount of cadence discipline fixes the denominator.
We deliver the full universe for your thesis — every qualifying company on the active web, scored and evidenced — inside the first month. The grind still happens. It just starts from everyone, not from an export.
Every searcher in your cohort licensed the same tools, applied the same industry filters, and exported overlapping lists. The companies inside are real — and thoroughly contacted.
Profile databases index what their crawlers and analysts found: the visible, the funded, the polished. Your thesis lives mostly outside that index, in businesses whose websites say "trusted since 1987", not "platform".
We start from 100M+ classified domains — 24.7M in business and finance alone, organized across 700+ categories — and read every candidate site against your thesis as you wrote it.
Not code tables standing in for the thesis. The thesis itself, applied by LLM analysis with a verbatim quote and source URL behind every inclusion and every exclusion.
If your list started where everyone else's list started, it was never proprietary — however good your outreach. A searcher's real edge is a denominator: knowing, with evidence, that you are working the entire population of qualifying companies, most of which no shared export has ever surfaced.
Pass one is triage: separating live operating companies from directories, parked domains, and resellers of record. In one industrial category, a 25,000-domain US-focused triage left roughly 17,300 genuine operators.
Pass two reads the survivors in depth — fifteen structured signals per company, from ownership language to certifications to service-model classification, each carrying the exact sentence that supports it. The method page documents every signal.
Scoring is opinionated on purpose: Mandate Fit carries 70%, Outreach Suitability 20%, Transition Context 10% — and group ownership zeroes a company out entirely, because a subsidiary is not a lower-priority target. It is not a target.
The screen is only as good as the question. These are the six things we push searchers to pin down in the intake session — most take one working call.
Define the subsector by what companies do — "installs and services industrial water systems" — not by an industry code. Websites describe activities; codes describe filings.
A radius, a set of states, or a density argument. We extract stated branch locations and service areas, so the universe can be sorted against where you will actually live and operate.
Nearly every searcher wants an independent company; few write it down. We make it a hard gate: any group, platform, or brand-family language zeroes the score, with the quote attached.
Service contracts, maintenance programs, consumables — the visible language of repeat business. If recurring character matters to your model, it becomes a scored signal, not a diligence surprise.
Residential mix, pure distribution, franchise systems — whatever your model cannot digest. Disqualifiers written into the thesis save hundreds of research hours downstream.
Every criterion must be answerable from what companies publish. We refuse thesis terms that require guessing — that discipline is why the positives are trustworthy. See our standards.
One industrial category, screened end to end. This is what a counted universe looks like before anyone writes a single email.
Ten subverticals from the same category, each with its count of eligible independent US companies — the kind of denominators searchers put in investor updates.
All fifteen ship with every row. For a searcher planning two years of owner conversations, these six do most of the ranking work.
Explicit language only — "second generation", "founder-led", "family-owned" — never inferred. Roughly half of confirmed industrial fits carried it. For a searcher, this shapes the first letter you write, because these owners answer stewardship, not slideware.
Stated founding year, decades in operation, independence language. Long-established, founder-associated businesses with an identifiable decision-maker and a limited visible leadership bench are the classic search profile — documented strictly from what each company publishes about itself.
The signal that saves your calendar. About one in ten keyword-perfect candidates in our runs was already inside a group — visible on its own site, stale in profile databases. Each is excluded with the disqualifying sentence quoted.
Maintenance agreements, scheduled programs, consumables streams. Your investors will ask about revenue quality at the first update; this signal means you walk into every owner call already knowing what the company says about its own repeat business.
How many principals are actually named, and whether visible management extends beyond one or two people. It calibrates the transition you would be stepping into — a real consideration when the plan is you, in the seat, on day one.
Most recent dated content, news cadence, posting rhythm. A dormant site is not a verdict, but it changes outreach order — and in a two-year search, sequencing the population well is half the return on the data.
Every engagement starts with a specimen you can audit: 8 top fits, 5 keyword-missed fits, 5 documented exclusions, 2 insufficient-evidence flags. The machining specimen is public.
Confirmed matches to the written thesis, ranked by score, each with certifications and ownership language captured as exact claim text.
"a fourth-generation, family-owned precision CNC machining company" — AS9100D, ITAR, president named on the site.
Real qualifiers whose homepages lack the category's obvious vocabulary. A fifth or more of confirmed fits carry this profile — the closest thing to genuinely uncontacted deal flow a searcher can buy.
"family-built, American-owned since 1965" — the machining story starts two clicks deep.
Keyword-perfect candidates we removed, with reasons quoted: subsidiaries, brand families, one company announcing its own acquirer on its news page.
An exclusion log is what makes the inclusions credible — to you, and to the investors reading your updates.
Companies whose sites do not answer the thesis either way. We flag them instead of guessing — a small honesty that tells you exactly how much manual follow-up the tail requires.
Vendors who never say "we don't know" are telling you something about their positives.
The number searchers remember: roughly one in ten keyword-perfect candidates was already group-owned — companies that would have absorbed a research week and a first call each, working from an export alone.
Profile databases are good at what they index. The question is whether what they index is where your search will be won.
| Dimension | Shared profile databases | Full-web thesis screening |
|---|---|---|
| Starting pool | Companies their crawlers found | 100M+ domains — the active web itself |
| Thesis fidelity | Industry codes and keyword filters | Your thesis text, read against every site |
| Long-tail coverage | Thins out below the polished mid-market | Built for 10–60-person operators |
| Evidence per claim | Profile fields of uncertain vintage | Verbatim quote + source URL, every row |
| Exclusion log | None — you discover dead ends yourself | Every removal documented with its reason |
| Who else works the list | Every subscriber with similar filters | Built once, for your thesis alone |
| Cost model | Annual seat licenses | Project pricing from €4,900 |
Contact enrichment, firmographic lookups on companies you already know, and coverage of large, well-documented targets in banked processes.
The census itself: who exists, who is independent, who fits, who is already owned — with evidence your investors can audit. Then run a gap analysis against your current pipeline.
A search fund does not need another annual license. It needs the population, once, done properly — then a way to re-cut it as the thesis sharpens after the first fifty conversations.
Custom ICP re-runs are included: tighten geography, add a certification gate, exclude a service mix, and the scoring re-runs across the whole universe at no additional cost. The lens adjusts; the asset stays.
Compare it to the alternative you were about to choose: months of intern hours assembling a list that still starts from the same shared index. Full detail on the pricing page, and the search fund use case shows the deliverable anatomy.
One subvertical, screened end to end, specimen format. The cheapest way to test whether your thesis survives contact with the full web — most searchers start here.
The core searcher buy: the complete evidenced universe for the thesis, scored and ranked, with the shortlist your outreach starts from. Re-runs included.
Scheduled re-screens and structured deltas per thesis. Most single-thesis searches will not need it — it earns its keep for accelerators and funds running several searchers on adjacent theses.
A vendor that cannot describe its bad fits is asking you to find them for it. Ours are below, in writing.
Send one email with a sentence about your thesis. You get the specimen report the same day — every company scored and ranked with full signal transcripts, including the ones we refused to include. Then decide.