Buy-and-build lives in the long tail that profile databases never indexed. We screen the entire active web against each platform's bolt-on thesis and hand your BD team a radar it can defend in front of the IC.
The strategy is a spread trade: the platform's multiple against the add-on's. What the spread pays out depends on volume — and volume depends on a number most funds have never actually counted.
Small independents trade well below the platform. Every add-on closed at the lower multiple re-rates on consolidation. This number your partners know by heart.
The model assumes a closing rhythm — add-ons per platform per year. Miss the cadence and the hold stretches, the IRR sags, and the thesis quietly becomes a hope.
How many independent companies actually fit the bolt-on thesis? Most BD teams work from a licensed export and honestly do not know. That unknown is what we remove.
The licensed database returns a few hundred names for the bolt-on thesis. The team works them hard — sequences, calls, conference passes — and by month ten the list is exhausted. What arrives after that is broker flow: by definition not proprietary, rarely priced like it is, and pipeline reviews start recycling the same names with fresher last-contacted dates.
We start from 100M+ classified domains — the entire active web — and screen the relevant categories against the thesis as your deal team wrote it, not SIC codes standing in for it. The output is a counted population: every independent fit, every keyword-missed fit, every disqualified look-alike, each with quoted evidence and a source URL.
A one-off buyer needs one good target. A platform needs a population — enough qualified independents to sustain a multi-year cadence. That makes the denominator a portfolio-level asset, and guessing it a portfolio-level risk.
A platform thesis in precision machining or water treatment starts from a category pool that can run past 300,000 domains globally. Nobody should read that by hand, and no keyword filter should be trusted to.
So the screen runs twice. Pass one triages the pool down to live operating companies. Pass two reads the survivors deeply against your thesis and extracts fifteen structured signals per company.
Eligible independent US companies per subvertical, from one full-universe run. Each count is a defensible denominator — the number an IC memo can cite. Method detail on the method page.
All fifteen signals of the framework ship with every row. For platform work, a handful do most of the ranking — and each answers a question your IC will ask anyway.
The instrument is your thesis text, not a code table. Subsector, services performed versus resold, customer types and contract language are matched sentence by sentence against what each company publishes about itself.
"design, installation and 24-hour service of industrial water systems" — scored against the thesis, quoted in the row
A company inside another consolidator is not a lower-priority target; it is not a target. Roughly one in ten keyword-perfect candidates in our specimen runs carried group ownership visible on their own sites — this is the signal that stops a research week, or a first call, being spent on a competitor's portfolio company.
Service contracts, maintenance agreements, scheduled programs, consumables streams — the published language of repeat revenue. Your IC asks about contract mix at the first screen; the radar means BD walks in already knowing what the company says about it.
Add-on logic is usually density logic: tuck-ins inside the platform's footprint or beachheads one territory out. Extracted locations let the radar sort against your actual density map rather than an HQ pin.
In quality-gated verticals the certifications are the thesis: UL 508A panel shops for automation, AS9100 and ITAR for aerospace machining, ISO/IEC 17025 for calibration, ASME stamps for boiler work.
"ISO certified" and "ISO/IEC 17025:2017 accredited" are different companies — we capture the exact claim text
Roughly half of confirmed industrial fits carried explicit founder or family language on their own sites. We flag founder-associated, long-established, independently positioned businesses with an identifiable decision-maker and limited visible leadership bench — website-visible statements only, never inferred circumstances.
None of these channels is useless. They answer different questions — and only one of them answers the question the platform model depends on.
| What the platform needs | Profile databases | Broker / banker flow | Full-web screen |
|---|---|---|---|
| Long-tail coverage (15–60-person independents) | Skewed to size, polish, funding events | Only what's already for sale | Starts from the entire active web |
| Screens your exact thesis, not a code proxy | Filter fields only | Their mandate, not yours | Thesis text is the filter |
| Current ownership status per candidate | Ownership fields go stale | Known for their own deals | Read from the company's own site, re-checked on cadence |
| A defensible denominator for the IC | Counts its index, not the market | No population view | Counted universe with exclusion log |
| Proprietary first contact | Same export your competitors license | Shown to every buyer on the list | Fits nobody else's tools surfaced |
| Cost shape for multi-platform funds | Per-seat licenses, every year | Success fees give back the spread | Per-thesis project or annual radar |
For polished mid-caps, contact enrichment and comps work, profile tools serve you well. We are not a replacement for them — we are the census they never claimed to be.
Point us at one live bolt-on thesis. The gap analysis shows, with evidence, what the current stack missed — before you commit to anything.
Each platform is its own universe with its own scoring configuration. Here is how the first ninety days typically run.
Deal team plus the platform's operators. Bolt-on criteria in plain language: subsectors, service model, geography and density logic, certifications, hard disqualifiers.
Output: written thesis, agreed scoring configCategory sweep, triage, deep extraction, analyst verification. You see a same-day specimen before committing; the full run lands as a ranked, evidenced file.
Output: universe CSV — fits, hidden fits, exclusions with reasonsCRM gets the keyed, deduplicated file. The weekly BD meeting gets working views: top-scored uncontacted, newly changed, newly disqualified. The IC memo gets the denominators.
Output: BD calling from the top decile, not the top of the alphabetRe-screens surface new entrants, ownership changes, offering pivots and momentum shifts as structured diffs. A universe delivered once is a snapshot; platform BD needs a feed.
Output: the add-on radar, kept currentSpecimen format: eight top fits, five keyword-missed fits, five documented exclusions, two insufficient-evidence flags. Anonymized here, as all public examples are; the full anatomy is on the machining specimen page.
"a fourth-generation, family-owned precision CNC machining company"
AS9100D, ITAR, CMMC Level 2. Defense, aerospace and medical end markets documented in case studies. President named on the site; visible bench beyond the family: one operations lead.
"family-built, American-owned since 1965"
A stamping-and-assembly shop whose machining depth only appears two pages deep. Keyword-driven indexes file it under the wrong category. Across verticals, a fifth or more of confirmed fits carried this profile.
"a wholly owned subsidiary of a global industrial group"
Keyword-perfect, capability-perfect — and already consolidated. Scored zero with the disqualifying sentence quoted. The exclusion log is why your IC trusts the list instead of auditing it.
site live, capabilities stated, ownership language absent
We do not guess. Where the website cannot support a classification, the row says so explicitly — a flag for a human follow-up, not a silent inclusion that pollutes the ranking.
The commercial point: the hidden fits are the closest thing to proprietary that sourcing data can honestly offer — companies your competitors' tools cannot see, reachable before any process exists.
Funds running multiple platforms consolidate the engagement; each bolt-on thesis stays its own universe with its own scoring lens. Refine a thesis mid-hold — tighter geography, a certification gate, a service-mix exclusion — and we re-run the scoring with the custom ICP at no cost.
The universe is the asset. The scoring is a lens you keep adjusting. And the IC gets counted claims — N in the universe, N confirmed independent, N already consolidated — instead of "we believe this vertical is fragmented."
One intermediated add-on gives back part of the multiple spread in fees. Against that arithmetic, a counted universe per thesis is the cheapest line item in the sourcing stack. Full detail on the pricing page.
One live bolt-on thesis, screened end to end. Use it as a coverage audit against your current stack — we show what it missed, with evidence, before you commit further.
The complete counted population for one thesis: ranked fits, hidden fits, documented exclusions, insufficient-evidence flags — CRM-ready and IC-citable. Custom ICP re-runs included.
Scheduled re-screens with structured diffs: new entrants, ownership changes first, offering pivots, momentum shifts. The radar stays current while your team works it.
A vendor that cannot describe its own limits is asking you to find them at your expense. Ours are specific.
One email, one live bolt-on thesis. The specimen report lands the same day — every company scored and ranked with every inclusion and every exclusion justified. If the denominator disappoints, you have lost nothing but a guess.