We build data pipelines for firms that do M&A work, not slide decks, so the conversation about industry consolidation trends usually starts in a place that's wrong for our clients. The trend report is a chart in a PwC or McKinsey deck showing global M&A volume year over year. Those reports do their job well. They are useful for the quarterly LP letter and useless on a Tuesday call, when somebody on the team has to pick which thirty operators in which two metros to dial before another buyer does.
The version of the question that gets a sourcing team paid is smaller and earlier. Where is consolidation about to happen, in which county, in which trade, with which kind of owner. That trail lives in public web data, and the work of pulling it is what we do for a living.
Where the macro reports stop
The annual outlook reports aggregate up to the sector. "Industrials M&A was up 18%." "Consumer multiples compressed." That framing works for the LP letter but doesn't help the analyst building Tuesday's call list.
That aggregation drops three things a sourcing team actually needs.
The first is geography below the metropolitan statistical area. A metro can read flat overall while one of its counties is two quarters into an obvious operator shake-out. The second is trade-level mix inside a broad sector. "Home services" covers HVAC and roofing and pool service and pest control, and the consolidation curves on those trades don't run in lockstep. The third is owner-side timing. A sector whose owner cohort skews near retirement age behaves nothing like the same sector with a younger ownership distribution, even if revenue per operator and growth rates look identical.
A macro report can't see any of that. It's aimed at the wrong unit of analysis.
The signals worth reading
For any fragmented, locally-licensed trade, four public-data signals triangulate consolidation pressure earlier than the trade press picks it up. None of them are sold to you. They sit in public portals written for compliance and consumer protection, not sourcing.
Establishment counts at the county and NAICS level, quarter over quarter. The Bureau of Labor Statistics Quarterly Census of Employment and Wages publishes establishment counts by county and six-digit NAICS code every quarter. The Census Bureau's Business Dynamics Statistics breaks the same population into establishment births and deaths annually. Cross those with state contractor-license rolls for the same trade and you get an operator-count time series at the unit of analysis a sourcing team actually works at. A multi-quarter establishment drop in a county that isn't losing population is the kind of pattern worth looking at. Operators don't vanish on their own. They get absorbed or quietly shut.
Permit-issuance velocity in the same county. Most municipal permit portals publish in a machine-readable form, and the ones that don't are scrape-friendly. Permit velocity per active operator is a proxy for how much work is chasing how few hands. When the ratio climbs sharply, the remaining operators get squeezed on capacity, the bigger ones acquire to grow headcount faster than they can hire, and the smaller ones pick up the phone when somebody calls. This is a county-by-county data problem because permit systems are run by municipalities, not states.
Owner-age curves from Secretary of State filings. Officer and registered-agent records, joined against voter rolls and property records where available, give a usable distribution of operator-owner ages by county. The right side of that curve is the supply of motivated sellers eighteen to thirty-six months out. A trade where a large share of operators are within a decade of typical retirement is a different sourcing problem than one where the owner cohort skews younger.
License-renewal lapses from state boards. State licensing boards publish renewal cadence and active-license rolls. The cadence varies by state, usually annual or biennial, and the lapse rate is publicly recoverable from those boards. Lapses are a noisy signal alone, but a cluster of lapses in a county where permit velocity is still rising tends to be the back half of a roll-up that's already moving.
These four sit in different portals, written for different readers, with no common schema. Reading them at scale is a tooling problem, not a data-access problem.
How they combine into something you can call on
Any one of these signals on its own is too noisy to act on. Together they constrain each other. Establishment-count decline in QCEW plus rising permit velocity plus an aging owner curve, in the same county and the same trade, is the configuration that argues a roll-up is forming under the surface. A press release can lag the underlying movement by quarters; the public-data trail does not.
The output a sourcing team wants from this is not a chart. It's a ranked list of counties and trade combinations, refreshed on whatever cadence the slowest underlying portal supports, with the underlying operators attached. Some share of what ranks high will already be inside someone's roll-up by the time you call. The rest is the part worth working.
Macro reports vs. county data
This is the part the trend reports can't fix. PwC, Bain, McKinsey and DealRoom write to a CFO and corp-dev audience that wants sector-level framing for board decks. The work of running a permit scraper across nine hundred county portals and reconciling it weekly against state contractor rolls is not the product they sell, because their buyers haven't asked.
So the report you read every January is real. It's just answering a different question than the one you're being asked to act on. The sector is up 18% means nothing for the operator in Lexington whose son doesn't want the business and whose competitor just bought two yards over in Fayette. That deal is going to close whether or not consolidation volume nationally is flat next year.
A trend deck and county-level data answer different questions. Pay for the trend deck and skip the county work, and the pipeline ends up holding the same operators every other fund is also chasing. The chart is calibrated to the wrong audience. The signal you need is in a different file, on a different portal, written for a different reader, and nobody is going to package it for you. We build the thing that does.
