When Do Contractors Actually Need a CFO? The Data Says 0.7% Hold a Third of Permit Value

The market is smaller than you think, and more concentrated

The retained public-permit corpus covers contributing jurisdictions from 2022 through 2024, not every U.S. permit. The roughly 409 million source records are the broad corpus, not the eligible count for this table. The analysis filters to positive declared value, valid 2022-2024 effective dates and a nonempty contractor business name, then groups by business name within a state partition. Those 577,373 business-name and state groups are associated with $7.478 trillion in declared permit value. One firm can appear in several states or under several names; this is not a deduplicated national company census.

Here's the shape of that market:

Method: public county and state building-permit records from contributing jurisdictions, 2022-2024, organized into 51 U.S. state partitions (50 states plus DC); coverage within each partition depends on which jurisdictions publish records. Contractor = distinct business name within a state; job value > 0. Value share = tier total ÷ $7.478T national total. n = 577,373 business-name/state groups. Business names were used only as a grouping key and are not published.

0.7% of business-name and state groups account for 32% of declared permit value. The top ~4% (Mid + Large combined) account for 61%, $4.6 trillion of the $7.5 trillion total. The median contractor pulled just 5 permits in three years. The top 1% pulled 387 or more.

How this compares with Census receipts

The number I'd normally be skeptical of is the concentration, 0.7% → 32% sounds too clean. For separately defined context, set it beside a government source: the U.S. Census Statistics of U.S. Businesses (SUSB) 2022, which reports receipts by firm size for specialty-trade contractors (NAICS 238).

Census, measuring revenue instead of permit value, finds:

• Firms with $50M+ in receipts = 0.64% of all employer firms → 32.7% of industry receipts.
• Firms with $10M+ in receipts = 4.3% of firms → 60.3% of receipts.

Our permit data said 0.7% → 32% and top 4% → 61%. Census said 0.64% → 32.7% and 4.3% → 60.3%. The rounded shares happen to be similar, but the tiers are cut on different bases: permit count versus receipts. They measure different quantities in different populations. The similarity is not independent validation of either distribution. Jurisdiction coverage and entity matching also affect the permit distribution; this is not a shared complete market census.

The counterintuitive part: bigger contractors don't do bigger jobs

Here's what surprised us. Break the tiers into finer bands and watch the average value per permit:

Method: same dataset and filters as above; business-name and state groups bucketed into selected finer permit bands (10-19, 50-99 and 200-499 omitted, so the rows do not sum to 577,373); avg value/permit = mean of each contractor's (total value ÷ permits). Every band n ≥ 25.

The displayed band averages for value per permit run roughly $392K to $548K, from the smallest contractor to the largest. By this measure, a 1,000-permit contractor is not pulling permits with larger declared values than a 30-permit contractor. It is pulling similar-value permits, many more times, across many more places. One caution: a specialty contractor's permit can carry the declared valuation of a whole project, so this is declared value per permit, not the contractor's own contract size. Distinct cities served roughly doubles at every tier: 1.2 → 2.4 → 5.1 → 9.8.

That single fact reframes the whole "when do I need a CFO" question.

~100 permits as a complexity marker: an operations question, not a size threshold

Below 20 permits over three years, the average contractor's permits carry $2.3M of total declared permit value, in about one city. At 100-499 permits, it's $106M of declared permit value across about five cities. The data does not establish a statistical cliff. We use roughly 100 permits across several cities as a judgment-based complexity marker, not a measured break point. Declared permit value is not the contractor's revenue, backlog or book of work, because a specialty permit can carry a whole-project valuation. The useful signal is the jump in permit count and jurisdictions. The Small and Mid groups differ by roughly 5× in average summed declared value; this comparison does not follow any company crossing a line. That is not because the average declared value per permit rose. It comes mechanically from many more permit records at a similar value per permit; the higher-volume groups also have records in more cities. City count is a correlated feature, not proof of what caused the value jump. The data does not directly measure crews, simultaneous workloads or entity count.

That is a point to test whether spreadsheet tracking can still handle the actual workflow. Multi-crew job costing, work-in-progress, over/under-billing, cash timing across projects that invoice on 30-60 day terms, none of that is a bookkeeping task. A bookkeeper records what already happened. The financial questions to test at that footprint are about what's going to happen: which jobs are bleeding, which crews are underwater, whether next month's payroll clears. That's a CFO's job.

Tier-to-tier multiples:

• Micro → Small (~20 permits): average summed declared value differs 9.5× ($2,271,316 vs $21,524,786)
• Small → Mid (~100 permits): 4.9× ($21.5M → $106M)
• Mid → Large (~500 permits): 5.3× ($106,360,386 vs $567,806,780)

What this means for you

If you're a contractor doing 100+ permits over three years (roughly 33 per year) across more than a couple of cities, you're already in roughly the top 4% of business-name and state groups in this permit dataset by permit count (Mid plus Large tiers: 3.6% + 0.7%), and financial blind spots can become costly if controls lag the work. Check whether job costing, WIP and cash timing are keeping up with the actual crews, contract values and billing dates.

The risk at this point tends to be quiet rather than loud. Illustrative examples, not measured findings from this dataset: a job that ran 15 points under bid that nobody caught until it closed, a city that has been unprofitable for two years, a cash crunch every January. Operator checks: compare estimated versus actual margin on every closed job by city and crew each month, keep a 13-week cash forecast, and review over/under-billing on open jobs before payroll weeks. This permit dataset does not measure whether contractors with stronger financial operations grow faster.

That's exactly the gap Level was built to close, CFO-level financial operations for contractors who've outgrown their bookkeeper but aren't ready for a full-time CFO hire.

Related reading

• How Many Competitors Do You Really Have?
• The October Problem: Why Q4 Revenue Funds a Q1 Cash Crunch
• Bookkeeper vs. Accountant vs. CFO

FAQ

Where does this data come from? Licensed aggregation of county and state building-permit records from contributing jurisdictions covering 2022-2024 (over 400M source permit rows), aggregated to business-name and state groups. Coverage is not a complete national census. The U.S. Census Statistics of U.S. Businesses (SUSB) 2022 receipts-by-firm-size tables for NAICS 238 provide separately defined context. Similar concentration shares do not validate this licensed grouping or reproduce its declared values. Full method and a machine-readable version are published at levelcfo.com/data/benchmarks/permit-market.json.

Is 100 permits a hard rule? No. The table compares permit counts and jurisdictions over three years; its dollars are declared permit value, not the contractor's revenue or backlog. Use multi-crew costing, commitment and cash-timing complexity as the decision trigger rather than translating that permit value into a universal CFO revenue threshold. Your own threshold depends on job mix, geography, and how many crews you run. The signal isn't the permit count itself; it's operating in multiple jurisdictions with multiple crews at once.

Why permit value instead of revenue? Declared permit value is a proxy for activity and scale. Published aggregates and their definitions are inspectable; the underlying licensed records are not an unrestricted public raw download. It can run higher than booked revenue when it records whole-project valuation rather than the contractor's share; understated valuations and unpermitted work can move the relationship the other way, so we don't equate it with revenue, we use it to measure concentration and volume, while Census receipts describe a separately defined firm population. Agreement in rounded shares does not reconcile the two sources.

How is confidentiality handled? Every figure is a tier-level or percentile aggregate. No individual contractor, company, or property is named or identifiable, and every published cell is backed by at least 25 contractors.