Record volume is the first number anyone quotes about a construction market and the least useful one they have. Two jurisdictions can differ by nearly two thirds in total filings and be worth the same to you in commercial work, then differ by more than fourteen times in a single trade category. Filed building activity answers questions like those well. There is another set it answers badly, and knowing which is which is most of the skill.
Volume is the number everyone quotes
A regional manager proposing a new territory almost always opens with a count. Permits up, starts up, square footage up. It is the easiest number to get and it feels like evidence.
It is also the number most likely to be wrong about your business specifically. Take two county building authorities in our coverage, across the same window — March 1 to August 19, 2026.
Marion County's building authority filed 13,867 records in that period. Wake County's filed 8,500. Marion is 63% larger by the only number most territory memos contain, and on that basis it is the obvious place to put a rep.
Now sort by what people actually sell. Commercial work: Marion 888 records, Wake 826. A 63% gap in total volume, and effectively the same amount of commercial activity. Roofing reverses the picture entirely — Marion filed 2,615 roofing records and Wake filed 181, a difference of more than fourteen times. And mechanical work inverts it again: Wake filed 1,919 to Marion's 1,541, so the smaller jurisdiction is the bigger mechanical market in absolute terms, not just proportionally.
Three categories, three different answers to "which one matters more," all from the same two files. Nothing about the total told you any of it.
Before any of that, one thing has to be settled, because it is the trap underneath every figure above. A jurisdiction label names an issuing authority, not a market. Both files here are county building authorities, which in Florida and North Carolina means chiefly the parts of each county outside incorporated city limits, plus any towns that contract their building department back to the county. Raleigh, Cary and Apex run their own and do not appear in the Wake file; Ocala runs its own and does not appear in Marion's. That is why these two are comparable to each other — same kind of authority, same scope rule — and why neither is "the Raleigh market" or "the Ocala market." Getting this wrong is not a rounding error. It is the difference between a market read and a boundary read, and it looks identical on a chart.
With that settled, the problem is solvable, because the composition is sitting in the same records as the count. Property type, record type, job value, location and the filing party are all on each row. Reading them in aggregate turns a number into a shape, and a shape is something you can make a territory decision against. It is the same underlying data used for reaching buyers on an individual project, read at a different altitude — many records at once rather than one project in depth.
What follows is what that read actually supports across the jurisdictions covered, and where it stops. Both matter, and the second is the part most vendors leave out. Different buyers weight these differently — a national supplier, a builder or contractor operating across regions, and a single-metro operator are not asking the same question of the same file.
Five readings commercial construction data supports
Each of these comes from fields that are on the record itself, so none of it requires inference about what a project is or who wants what. The fourth column is the one to read twice.
| Signal | What to pull | What it tells you | What it does not |
|---|---|---|---|
| Direction | Record count by month, as far back as that jurisdiction publishes | Whether activity is rising, flat or falling, and how sharply | Whether the growth is in anything you sell |
| Value | Job cost distribution, not the average | Whether the market is many small projects or a few large ones — a different sales motion entirely | Material spend, which is a fraction of job cost and varies by category |
| Mix | Property type and record type as a share of the total | Whether the growth is in your category — the reading that most often reverses a decision | Specification level, or whether your product is even in scope |
| Concentration | Record count grouped by contractor name | Whether the work runs through a handful of firms or many | Their existing supplier relationships or how loyal those are |
| Repetition | Record count grouped by owner name | Which owners build repeatedly rather than once, and in what categories | Their pipeline, financing or intent beyond what they have already filed |
Mix is highlighted because it is the reading that most often reverses a territory decision that direction and value had already justified.
Run them in that order. Direction tells you whether to keep looking. Value tells you what kind of sales organization the market needs. Mix tells you whether it is your market at all — the Marion and Wake comparison is a mix reading, and it reverses twice depending on which category you sell. Concentration and repetition then tell you how to enter it.
Mix sits third rather than first because direction and value are nearly free — they can be read off a summary in minutes, and a market that fails direction outright never needs a mix read. But it is the reading that decides, and it is the one most often skipped, because it is the only one of the three that requires actually grouping the file. A territory memo built on the first two numbers alone is a memo built on the two that were easiest to produce. The fields each record carries are what make the third reading possible, and an extract of your own markets will tell you within an hour whether a market survives it. If you would rather see the short version of how this works first, the questions we get asked most cover it.
Concentration cuts both ways
Of the five, concentration is the one most often read as a single-direction signal. It is not.
Worth saying what the two authorities above actually show on this measure, since it is not the dramatic contrast the argument might lead you to expect. Both are fragmented. Marion's ten busiest contractors account for about 16% of its attributed records across roughly 1,900 firms; Wake's account for about 14% across roughly 2,250. Neither is controlled by a handful of players, which means in both cases coverage is a headcount problem before it is a relationship problem. The sharper number is underneath those: Marion averages 6.6 records per firm and Wake 3.2, so the smaller jurisdiction spreads its work across more contractors doing less each. That is a real difference in how many doors a rep has to knock on, and it is the sort of finding that only appears once you group the file.
Other markets do concentrate, and the reading flips there. A market where ten contractors hold most of the filings is a market with very few doors. That reads as bad news, and for a supplier trying to enter cold it usually is — those firms have incumbent relationships, and displacing an incumbent on price alone is the least durable win available. But it is also a market where a coverage strategy is cheap. Ten relationships, well maintained, put you in front of most of the work. One good hire can hold that.
A fragmented market is the mirror image. No incumbent to displace and plenty of room to win work on merit — but you need to reach a hundred firms rather than ten to see the same share of projects, and that is a headcount problem long before it is a relationship problem.
A concentrated market is expensive to enter and cheap to hold. A fragmented one is cheap to enter and expensive to hold. Neither is better. They need different plans.
Which is why concentration should be read alongside how you actually sell, not on its own. A supplier with a strong technical sales team and a differentiated product does better in a concentrated market than the raw number suggests, because the argument for switching is stronger than price. A supplier competing largely on availability and delivery does better in a fragmented one. The same figure supports opposite conclusions depending on which of those you are, which is why the supplier read of this data and the contractor read of the identical file come out differently. That is a feature of the data, not a defect in it — and it is visible across every market in the set once you group by contractor.
Want to run this on a market you are considering? We can put together an extract for specific jurisdictions, record types and value thresholds, so you can group it yourself rather than take our word for the shape.
Request an extract or call 888-888-1214What the records will not tell you
Building records describe filings. They do not describe commercial outcomes, and the gap between those two things is where most bad conclusions get made. Four limits worth stating plainly.
A record carries a status. It does not carry bid stage, award or buyout. You can see that a filing exists and where it stands with the issuing authority — you cannot see whether the contractor has priced it, won it, or bought it out. Anyone telling you a records feed shows bid stage is describing something other than public records. What the status field actually contains is worth checking before you build a process on it.
Job cost is not material spend, and the ratio between them varies enormously by category. A warehouse and a hospital at the same declared value are not the same opportunity for a roofing supplier. Job cost is a useful proxy for project scale and a poor one for your addressable share of it — which is why the filtering happens against your own thresholds rather than a generic one.
Depth of past activity varies by jurisdiction, because authorities publish different amounts of it. A year-over-year comparison needs two years of history in that specific county, not two years somewhere in the coverage set. Where the history is shallow, direction is a weaker reading and mix is still a strong one — composition does not need a long baseline. Worth confirming the depth for your markets before a trend claim goes into a plan.
A filing is evidence that a project reached the point of being filed. It is not evidence of what an owner intends next, who they will hire, or how they finance. Repetition tells you an owner has built repeatedly, which is a reasonable basis for cultivating them — it is not a pipeline, and we would rather you treat it as the former. Looking at real rows makes the distinction obvious faster than any description does.
Why Alliance Data Solutions for market-level work
The five readings work off any complete records set. Here is the specific case for this one.
Grouping needs the whole market, not a sample of it. Concentration and mix are share calculations, so a partial file gives a confidently wrong answer rather than an incomplete one. Every figure in this article came out of two county files and about twenty minutes of grouping. Coverage runs across 39+ states and 300+ jurisdictions, defined at the level of the issuing authority rather than the state, so you can pull a market as the market actually is.
The fields survive grouping. Contractor name, owner name, property type, record type and job cost are on every row, which is what makes four of the five readings arithmetic rather than interpretation. The full record specification is published rather than described on a call.
Records are deduplicated before you get them. This matters more for market work than for project work. A duplicate is a nuisance in a prospecting list and a distortion in a share calculation — it inflates whichever contractor or category it lands in. Records are also scored on how many of the up to twenty fields are populated, then filtered to the thresholds you set, so a thin row does not quietly skew a percentage.
It arrives in a form you can actually group. CSV, Excel or SFTP, cut to your jurisdictions, record types and value thresholds. You run the analysis in your own tools rather than receiving our interpretation of it. API access is on the 2026 roadmap and is not live today. Published plans exist for single-market use; multi-market work is custom-quoted, with territory, volume and frequency set in the agreement.
And the limit, stated the same way it is stated everywhere else: this is public building activity, assembled, deduplicated, scored and filtered at national scale. It is not a proprietary intelligence network and we do not claim one. Everything in section two is arithmetic on public filings. What makes it useful is that the arithmetic is impractical to do yourself across three hundred jurisdictions, not that the underlying facts are secret.
Common questions
What can commercial construction data tell you about a market?
Filed building activity supports five readings of a market: how record volume is moving over time, how job value is distributed, which property and record types are growing, how concentrated the contractor base is, and which owners file repeatedly. Together those describe the shape of a market rather than only its size.
How far back does construction record history go?
Alliance Data Solutions holds the record history each jurisdiction makes available, which varies by jurisdiction because authorities publish different amounts of past activity. A year-over-year comparison needs two years of history in that specific county or city, not two years somewhere else in the coverage set.
What can construction records not tell you?
Records describe filings, not commercial outcomes. They do not carry bid stage, award, material specification, budget split, supplier relationships or financing. A record confirms that a project was filed at a stated value in a stated category — the rest is inference, and worth treating as inference.
How is market-level construction data delivered?
As CSV or Excel files, or over SFTP, filtered to the jurisdictions, record types and value thresholds agreed with the customer. Analysis is run by the customer in their own tools. API access is on the 2026 roadmap and is not live today. A sample extract shows the layout, and the field list shows what each row carries.