Every article on the first page of "HVAC roll-up strategy" tells you the same three things: the market is fragmented, the operators are aging out, PE is consolidating. None of them tell you which operators, and none rank the universe. The playbook is everywhere; the scoring is nowhere. That's a strange equilibrium given that ranking the universe is the actual job nobody on page one has done.
The Census County Business Patterns data for NAICS 238220 ("Plumbing, Heating, and Air-Conditioning Contractors") is the public starting line. It combines plumbing and HVAC into one code, which is annoying for an HVAC-only sizing claim but is the canonical public count. From there, the industry trade press (ACCA and its member surveys) gives you the qualitative shape of the operator base: heavily small-business, heavily owner-operator, with a long tail of multi-truck shops that aren't national brands.
That public sizing is good enough to argue the thesis. It isn't good enough to hand a buyer a hitlist. The gap between the two is where the scoring work lives.
What the consolidators are actually buying
The Grata playbook and the CT Acquisitions tracker both lean on the same implicit profile: a multi-truck shop, around twenty to fifty employees, owner over sixty, in a metro with population growth. That's the modal target, and also the most-contested one. Owners that match that profile in Sun Belt metros tend to report receiving several buyer letters a year, and Grata's own write-ups and HVAC trade-press coverage have documented this for the last two cycles.
The interesting slice is the next ring out. Operators just under the size the platforms care about (call it eight to fifteen trucks) are too small for the platform's initial scan and too big to leave out of a regional roll-up that's run with any discipline. The Census CBP universe of HVAC and plumbing contractors skews heavily to establishments in this band and smaller, and that's where the disciplined consolidator can move before the platform's BD team finishes the next quarter's outbound. The exact share depends on how you draw the line; CBP's establishment-size breakdowns give you a defensible bracket to work in.
Three signals worth scoring
A scoring function is only useful if the signals are reproducible and the weights survive the next campaign. Three input categories are well-supported by the public-data and trade-press literature, and they're the ones a buyer-side team can actually compute against the CBP universe.
Permit-issuance velocity in the operator's primary metro, rolling 90 days. New residential and light-commercial permits are a leading indicator of installation demand, which feeds the operator's hiring problem, which is when the owner starts taking buyer calls. Municipal building-department permit feeds and Census Building Permits Survey data both expose this. The mechanism is well-documented in HVAC trade publications and is the cleanest of the three to compute.
Hiring-cycle shape from the operator's job postings. Not how many postings, but the shape. An operator reposting the same senior tech role every six to eight weeks for a year is signaling something specific: the owner is hands-on, the bench is thin, and the firm has hit the ceiling of what one person can run. BLS job-openings and labor turnover data describes the industry-level dynamics; the per-operator signal sits in public job-board scrapes.
Owner-tenure proxies from state filings and LinkedIn. Not age, but tenure. Years since the LLC was formed, years since the owner's LinkedIn role last changed, the gap between those two. Operators where the LLC is twenty-plus years old and the owner's role hasn't changed since the filing are the population the succession-cliff thesis is actually about. State Secretary of State filings are the canonical public source; LinkedIn's public profile data is the second leg.
What's notably absent from a defensible scoring function: revenue estimates, employee-count proxies, and technology-adoption scores. Public-data hygiene on those is poor and the signal-to-noise rarely justifies the engineering.
Where the under-served slice probably is
Cross the size band with the three signals and the map narrows. Trade-press coverage and Census population-growth data both point to the same regions where the platform consolidators have been slower to push: secondary Texas markets outside Dallas and Houston, the Carolinas inland from the coast, the Phoenix exurbs, and the Ohio-Indiana corridor. The platforms favor the largest metros because that's where the modal target lives; the band just below the modal target, in metros just below the obvious ones, is structurally where there's room.
The point of building a scoring function is that the resulting list belongs to the firm running the thesis, not to the SERP. But the shape of the slice is the answer to the question the existing articles dance around: yes, the window is open, and no, it isn't open in the metros the playbooks point at.
What this is and what it isn't
This is a thesis post about how to score the HVAC universe, not a deal-flow promise. A scoring function tells you where to look. The conversation with the owner is still the conversation with the owner, and the half of roll-up math that lives on the integration side is unchanged. What scoring does is collapse a sourcing universe of tens of thousands of HVAC and plumbing establishments to the much smaller subset where size band, geography, and the three signals agree.
The playbook for HVAC roll-ups is everywhere. The scored list that lets you act on it is the work. Anyone with a Substack can write the playbook. Building the index, keeping it fresh, and pointing it at the right names is what a sourcing infrastructure has to do.
