Small-Bay Flex Industrial: Measuring Tenant Demand

Search for flex industrial demand analysis and the results split between two different subjects: demand-side flexibility in industrial electricity markets, and a thin layer of commercial real estate commentary quoting sales volumes from subscription databases. Neither answers the question a lender holding a term sheet on a 40,000-square-foot small-bay project actually has, which is who the tenants are, how many of them the catchment produces, and whether the demand that filled the last project is still being generated. This article answers that question from the public record. The tenant of a small-bay flex building, the contractor, the repair shop, the installer, the small distributor, the maker, is one of the most precisely counted actors in the American economy: County Business Patterns counts their establishments by employment size every year, the Business Formation Statistics count their births every month, the Quarterly Census of Employment and Wages tracks the adjacent big-box cycle they are so often confused with, and the Census Bureau's construction spending and e-commerce series date the capital wave and the freight demand around them. Every figure below carries a named source and a date.

Definitions first, because the asset class is defined by its tenant rather than its architecture. Small-bay flex means multi-tenant light-industrial buildings cut into bays, typically 1,000 to 5,000 square feet, each with a roll-up door, a small office component and clear heights modest by modern logistics standards. It is not bulk distribution, and the demand analysis that treats it as a smaller version of a fulfillment warehouse imports a cycle that does not belong to it. The distinction is measurable, and the public record measures it.

The tenant, counted: businesses of fewer than 20 employees

Start with who fits in a bay. A business with fewer than 20 employees fits in 1,000 to 5,000 square feet; a business with 100 does not. County Business Patterns for 2023, released on 26 June 2025, counts 8,361,342 establishments with paid employees in the United States, and 85.5% of them have fewer than 20 employees (U.S. Census Bureau, County Business Patterns 2023, 2025). In the sectors that actually rent small bays the share is higher still. Specialty trade contractors, code 238, the electricians, plumbers, HVAC installers and finish trades that are the archetypal flex tenants, count 521,315 establishments, 90.2% of them under 20 employees. Repair and maintenance, code 8111, counts 169,572 establishments at 95.7% under 20. Truck transportation, code 484, counts 170,286 at 90.1%. Against them stands general warehousing, code 493110, with 16,753 establishments of which only 60.1% are under 20 employees and whose average establishment employs 90.4 people; the all-industry average is 16.7 (U.S. Census Bureau, County Business Patterns 2023, 2025). The contrast is the asset class in two numbers: the flex tenant averages six to ten employees; the warehouse operator averages ninety.

The employer counts are the floor, not the ceiling, because the smallest flex tenants have no payroll at all. The Nonemployer Statistics, whose 2023 edition was released on 15 May 2025, count the sole-proprietor contractor working from a bay with a truck, a trailer and no employees, a tenant class every small-bay landlord knows and no employer series records (U.S. Census Bureau, Nonemployer Statistics 2023, 2025). A catchment's flex demand base is therefore read in two layers: the employer establishments under 20 employees in the tenant sectors from County Business Patterns, by ZIP code, and the nonemployer filings in the same sectors by county. The County Business Patterns article covers the mechanics, including the ZIP-level files that let the count be drawn for a real catchment rather than a metro.

The formation wave: a record year and a faster one after it

Demand for small bays is generated by business formation, and formation is the public series with the most remarkable recent history. The Census Bureau's Business Formation Statistics count applications for employer identification numbers: 3,498,990 business applications were filed in 2019; 4,356,498 in 2020; 5,390,816 in 2021; 5,062,563 in 2022; 5,469,302 in 2023; 5,224,176 in 2024; and 5,671,836 in 2025, the highest annual total in the series (U.S. Census Bureau, Business Formation Statistics, monthly data through July 2026, 2026). The first seven months of 2026 produced 3,867,036 applications, 13.8% ahead of the 3,397,482 filed in the same months of 2025. Six years after the pandemic reset, business formation has not reverted to its 2019 baseline; it has settled 50 to 60% above it and is still climbing.

The honest caveat is the difference between an application and a business. Most applications never become employers, and the Bureau publishes a refinement, high-propensity applications, those whose characteristics (a planned wage date, a corporate form, an industry that typically hires) make an employer likely. High-propensity applications tell a slightly different story: 1,316,191 in 2019, a peak of 1,848,540 in 2023, then 1,715,458 in 2024 and 1,708,842 in 2025, still 30% above 2019 but two years past their own peak (U.S. Census Bureau, Business Formation Statistics, 2026). Both series belong in a flex memo. The total series measures the churn of very small operations, including the no-payroll tenants the Nonemployer Statistics later count; the high-propensity series measures the pipeline of future employers, the tenants who take a second bay in year three. A catchment's formation rate, applications per 1,000 residents from the state and county BFS files, is the single best public leading indicator of small-bay absorption, and it is published monthly with a two-week lag.

The other industrial: the big-box cycle that flex gets confused with

The industrial narrative of the last five years, the e-commerce boom, the million-square-foot fulfillment centers, the 2022 peak and the correction after it, belongs to a different tenant. The Quarterly Census of Employment and Wages puts numbers on that cycle. General warehousing and storage, NAICS 493110, employed 1,121,184 people on annual average in 2019; 1,344,544 in 2020; 1,569,065 in 2021; and 1,772,530 in 2022, a 58% expansion in three years. Then the cycle turned: 1,725,148 in 2023, and 1,755,061 in 2024, still 1% below the 2022 peak (U.S. Bureau of Labor Statistics, QCEW annual averages 2019 to 2024, NAICS 493110, 2020 to 2025). Establishments kept growing through the correction, from 13,795 in 2019 to 17,324 in 2024, and average annual pay rose from $42,306 to $55,067, but the employment plateau marks the end of the expansion phase that filled the bulk pipeline.

Set the two demand bases side by side and the divergence is the finding this article exists to document. The big-box tenant's cycle peaked in 2022 and has moved sideways since. The small-bay tenant base did the opposite: business applications set their record in 2025 and are running 13.8% ahead in 2026, and the employer universe of sub-20-employee establishments in the tenant trades has grown every year of the series. The two demand curves share the word industrial and almost nothing else. A flex project underwritten off the bulk-logistics narrative in 2022 bought the top of someone else's cycle; a flex project screened off the formation series in the same year was looking at a demand base still accelerating. The QCEW demand-driver article covers how the employment series are read at county level, where the same divergence can be tested catchment by catchment.

The capital wave, cresting on schedule

Supply for the industrial family arrives through construction spending, and the Census Bureau's Value of Construction Put in Place series dates the wave precisely. Manufacturing construction, the factory-building boom that reindustrialization policy set off, reached $235.7 billion in 2024 and fell to $220.4 billion in 2025, a decline of 6.5%; by December 2025 the seasonally adjusted annual rate stood at $202.4 billion, 11.4% below December 2024 (U.S. Census Bureau, Construction Spending, December 2025, release CB26-30 of 27 February 2026). Commercial construction, the category that carries warehouses and distribution buildings, fell 7.8% on the year, from $132.3 billion in 2024 to $122.0 billion in 2025, though its December rate had steadied at $121.9 billion, 0.7% above a year earlier. Total private nonresidential construction ran at a $730.9 billion seasonally adjusted annual rate in December 2025, 1.8% below December 2024.

For a small-bay project the series carries two messages. The first is context: the capital wave that priced land, bid up contractors and filled the big-box pipeline crested in 2024 and is receding, which relieves construction cost pressure on exactly the format that is hardest to justify building, since a multi-tenant building with twelve roll-up doors, twelve HVAC units and twelve demising walls costs more per square foot than a single box of the same area. The second is a caution about what the national series cannot say: construction spending has no category called small-bay flex, and the local pipeline is read from the county permit record, not from any federal table. The Census Building Permits Survey, covered in the supply pipeline article, tracks housing units only; industrial permits appear in local records with valuations and square footage, and a catchment where a hundred thousand square feet of small-bay space is permitted deserves a different memo from one where nothing has been permitted since 2019.

E-commerce: the freight tide that lifts the small tenant too

E-commerce is the demand story behind the big-box cycle, but it feeds the small-bay tenant as well: the returns processor, the kitting operation, the local courier, the seller who has outgrown a garage. The Census Bureau's quarterly series puts e-commerce at 17.1% of total retail sales in the second quarter of 2026 on an adjusted basis, with e-commerce sales growing 12.2% year over year while total retail sales grew 6.7% (U.S. Census Bureau, Quarterly Retail E-Commerce Sales, second quarter 2026, 2026). The share has climbed every quarter of the past year, from 16.3% in the second quarter of 2025 through 16.4%, 16.7% and 17.0% to the current reading; unadjusted, the quarter's e-commerce sales were $329.5 billion. Online retail is growing at nearly twice the pace of retail as a whole, and every percentage point of share migrates fulfillment, returns and last-mile activity into industrial space of every size.

The honest use of the series in a flex memo is directional, not allocative: no public dataset says how much of a quarter's e-commerce growth lands in 2,000-square-foot bays rather than million-square-foot fulfillment centers. What the analyst can do is pair the national tide with the local counts that are public, the formation rate and the sub-20-employee establishment counts in transportation and warehousing codes for the catchment, and let the local series carry the weight. Where couriers and messengers and warehousing establishments under 20 employees are multiplying in the county files, the last-mile layer is arriving in small spaces; where they are not, the national e-commerce figure is someone else's demand.

Who owns the buildings: a landlord base smaller than its tenants

The supply side of small-bay flex has a structure the public record captures with an irony worth pausing on: the landlords are smaller businesses than the tenants. Lessors of nonresidential buildings except miniwarehouses, NAICS 531120, counted 34,559 establishments with paid employees in 2023, and 96.3% of them had fewer than 20 employees; 78.3% had fewer than five. The average lessor establishment employs 5.1 people (U.S. Census Bureau, County Business Patterns 2023, 2025). The Quarterly Census of Employment and Wages counted 27,222 private reporting units in the code on a 2024 annual average, up 9.7% from 24,810 in 2019 (BLS, QCEW, NAICS 531120, 2020 to 2025), and the difference between the two counts is itself informative: QCEW consolidates worksites into parent accounts, so the gap between 34,559 establishments and 27,222 reporting units is a rough measure of multi-property ownership, and it is small. Nonresidential landlording in this size class remains a cottage industry of local owners with a building or two, plus the no-payroll owners the employer series cannot see at all.

For demand analysis the landlord structure matters because it shapes the data a lender can get. There is no public occupancy series for flex space, no REIT disclosure covering the mom-and-pop inventory, and no reliable rent comp feed for buildings owned by 34,000 small operators; the paid databases thin out in exactly this segment. The supply inventory is therefore built the way the self-storage article builds one, from the assessor roll and the permit record: parcels with light-industrial use codes and building areas, unit counts read from aerial imagery and door counts, and the opening year from the permit. The reward for the effort is real: a catchment's small-bay inventory changes slowly, so a one-time address-level build stays current with modest maintenance, and the lender who holds it knows the market's supply better than any subscription can.

The edges of the class: expenses and the specialized neighbors

Two boundaries complete the definition of the asset class, and both have demand consequences a memo should state. The first boundary is operating cost. A twelve-bay building runs twelve HVAC units, twelve electrical panels, twelve entries and a parking field of shared circulation, and its expense ratio sits structurally above a single-tenant box of the same area, where one tenant pays the utilities and maintains the slab it occupies. Multi-tenant light industrial trades that cost against its diversification: twelve small tenants fail one at a time, a single tenant fails all at once. The evidence for where those expense ratios actually sit, drawn from the public CMBS filings that report line-item operating statements for the industrial collateral that reaches securitization, is worked through in the small-bay expense benchmarks article; the demand-side point is that the format's higher expense load is the price of the tenant diversity this article has been counting, and that a pro forma showing single-tenant expense ratios on a multi-tenant building has misread the class before any demand question is asked.

The second boundary is specialization. The industrial family runs from the perfectly generic to the perfectly specific, and small-bay flex sits at the generic pole: any bay serves any trade. At the opposite pole sits cold storage, where more than half the building is refrigerated envelope and machinery, where SBA guidance applies the special purpose classification, and where demand is read not from business formation but from the food system's own federal series, the USDA's biennial census of refrigerated warehouse capacity chief among them; the cold storage article in this series reads those. Between the poles sit the in-between formats, the shallow-bay distribution building, the contractor yard with outdoor storage, the service-heavy showroom hybrid, and the practical rule for a lender is that demand analysis inherits from the nearest pole: the more generic the space, the more the tenant-count method of this article governs; the more specialized the improvements, the more the analysis must be built on the specialized demand series and the exit must be underwritten like special purpose collateral even where no rule requires it; the SOP's reported examples make the point concrete, since a bay converted to an oil-and-lube service center with pits and in-ground lifts crosses onto the special purpose list itself. Misclassification runs in both directions, and each direction has a signature error: underwriting a true flex building on a single anchor tenant's covenant, or underwriting a specialized building on the strength of a generic industrial narrative.

Stated as arithmetic, the boundaries change the same ratio in opposite directions. Higher operating expenses lower the net income a bay's rent produces, so the tenant-per-bay cushion the catchment must supply rises with the expense ratio; specialization narrows the set of tenants a bay can accept, so the effective tenant base shrinks even when the counted one is large. The five-number method absorbs both: the expense ratio enters through the reasonableness check on the pro forma, and specialization enters through which NAICS codes are allowed into the tenant count. A flex memo that counts every industrial code in the county for a building whose bays have 480-volt three-phase service and floor drains is counting tenants the building cannot serve; the honest count starts from what the improvements permit.

Counting supply where no federal series does

The supply method deserves its own statement, because flex is the asset class where the federal record's silence is loudest. County Business Patterns counts lessors, not buildings; the construction series counts spending, not bays; the permits survey counts homes. What the public record does provide is the frame for a local census: the assessor's parcel file identifies every light-industrial parcel with its building area, year built and owner; the permit record dates every addition and conversion; and the utility of the address-level employer counts is that they place the tenants inside the buildings, so a reconciliation of County Business Patterns ZIP-code counts against the parcel inventory shows which buildings are full of countable tenants and which are warehousing owner-occupiers or storage. The output that belongs in a memo is square feet of multi-tenant light-industrial space in the drive-time catchment, dated, with the pipeline from permits added, and the tenant base beside it: employer establishments under 20 employees in codes 238, 484, 4931 and 8111, plus the county's nonemployer filings in the same codes, plus the formation rate. Demand per square foot, stated with both numerators and the denominator sourced, replaces the borrowed vacancy figure the memo would otherwise quote.

What the lender reads

Small-bay flex reaches lenders mostly as small-balance loans: a $2 million refinance of a six-bay building, a $4 million construction loan for twelve bays, an SBA 504 project for an owner-user taking half the building and renting the rest. In SBA lending guidance the special purpose property examples reported from SOP 50 10 8 name car washes, hotels and motels, marinas, heavily refrigerated cold storage and service centers with pits and in-ground lifts, categories whose improvements have no alternative user; the ordinary light-industrial bay is not among the reported examples, and the list, which is longer than any summary, is governed by the SOP itself (504 Capital Corporation, guidance on SOP 50 10 8, 2025; NAGGL, SBA Procedural Notice 5000-872764, 2025). The economics behind that distinction are the asset class's quiet strength: a bay that loses its electrician can take a plumber, a courier or a cabinet shop without a build-out, which is precisely what the single-purpose formats covered elsewhere in this series cannot say. The public SBA 7(a) and 504 datasets, which MMCG Analytics' SBA layer is built on, carry the lending history of the tenant trades themselves by NAICS code, subject to the rule that no performance rate is shown for a cohort of fewer than ten loans, and the small-balance performance article reads that record by property type. MMCG Analytics supplies the data and the analysis; the credit decision rests with the lender.

What the lender's memo needs from the demand side is proportion. A twelve-bay building needs twelve tenants from a catchment that typically holds several thousand sub-20-employee establishments in the tenant trades; the demand question is never whether small businesses exist but whether the catchment's stock of them is growing or shrinking, what the formation rate has done since the last delivery, and how much competing small-bay space the permit record shows arriving. Those are the three numbers that move a flex vacancy, and all three are public.

Method: the five numbers a flex memo should carry

First, the tenant base: employer establishments with fewer than 20 employees in codes 238, 484, 4931 and 8111 for the catchment's ZIP codes from County Business Patterns, with the county's nonemployer filings in the same codes beside them, both dated. Second, the formation rate: business applications per 1,000 residents for the county from the Business Formation Statistics, with the high-propensity subset shown separately and the trend since 2019 stated. Third, the inventory: multi-tenant light-industrial square feet in the drive-time catchment from the assessor roll reconciled with the permit record, with the pipeline of permitted, unbuilt space added and dated. Fourth, the cycle context: the QCEW warehousing employment series and the construction spending categories, national and, for QCEW, county where disclosure allows, to show which phase of the big-box cycle the catchment is importing through land prices and construction costs. Fifth, the reasonableness check: tenants per bay implied by the pro forma against the catchment's tenant base per existing bay, a ratio that exposes an overbuilt submarket faster than any borrowed vacancy figure.

Each of the five is a measurement with a source and a date; none is a forecast. The forecast that remains, absorption pace and rent, sits on top as a stated range, and the memo says which measured number would have to be wrong for the range to fail. That is the provenance standard applied to the one industrial format the commercial data industry covers worst, and it is the reason the method leans harder on federal series here than anywhere else in the asset-class pillar: for small-bay flex, the public record is not the budget alternative to a subscription; it is the only census of the tenant base that exists.

A worked sequence for one building

Run for a proposed twelve-bay building on a service road, the sequence goes as follows. The analyst draws a fifteen-minute drive-time polygon, pulls County Business Patterns ZIP-level counts for codes 238, 484, 4931 and 8111, and sums establishments by employment size: say 2,400 employer establishments under 20 employees, plus 5,100 nonemployer filings in the same codes for the county, scaled to the polygon by employment share. The county's Business Formation Statistics show applications running 40% above 2019 and high-propensity applications 25% above. The assessor roll and permit record yield 610,000 square feet of existing multi-tenant light-industrial space in the polygon, 34,000 square feet of it permitted in the last 18 months and unbuilt. The pro forma's twelve bays at 2,500 square feet add 30,000 square feet, taking the pipeline to 3.4% of stock in a catchment whose tenant base has grown 5 to 6% a year since 2021. Tenants per existing bay run roughly ten to one. On those numbers the memo can say something precise: the project asks the catchment to absorb a 3.4% stock addition against a tenant base compounding faster than that, and the number that would have to be wrong for the deal to fail is the formation rate, which is published monthly and can be watched.

The same sequence with the signs reversed is the protection. A catchment where the formation rate has rolled over, where sub-20-employee establishments in the tenant trades have been flat since 2022, and where the permit record shows 12% of stock arriving is a market where the national small-bay narrative will not save a local project, and the analyst who has built the five numbers can show it before the term sheet goes out rather than after the certificate of occupancy.

What 2026 looks like from the series

Read together in mid-2026, the public series describe an unusual alignment for this asset class. The tenant base is expanding at a record pace: 5,671,836 business applications in 2025 and a 2026 running 13.8% ahead of it. The competing capital wave is receding: manufacturing construction down 6.5% in 2025, commercial construction down 7.8%, the big-box employment cycle flat since 2022. The freight tide keeps rising: e-commerce at 17.1% of retail and growing at nearly twice retail's pace. And the landlord base remains fragmented, 96.3% of lessor establishments under 20 employees, which means the supply response will stay slow and local. None of these facts prices a specific building, and this article deliberately stops short of the market-by-market absorption and rent evidence that decides a deal. What the alignment does say is that the demand base for the smallest industrial format is the one industrial demand base still accelerating in the public record, and that the analyst who tracks it needs four bookmarks, not a subscription: the BFS monthly file, the CBP annual release, the QCEW quarterly county files and the county permit portal.

The cadence of the four bookmarks is worth writing into the memo's maintenance plan, because each series answers on a different clock. The Business Formation Statistics arrive monthly, about two weeks after the reference month, and are the early-warning line. The QCEW county files arrive quarterly with a five-month lag and confirm what formation predicted. County Business Patterns arrives annually with an eighteen-month lag and resets the establishment census. The permit portal updates continuously and is the only forward-looking series of the four. A flex position reviewed on that calendar is watched by the same instruments that justified it, which is more than most small-balance industrial credit can say.

Frequently asked questions

What is small-bay flex industrial space?

Multi-tenant light-industrial buildings divided into bays of roughly 1,000 to 5,000 square feet, each with a roll-up door and a small office component, leased to contractors, repair shops, couriers, installers and small distributors. It is defined by its tenant size rather than its architecture, and it is not bulk distribution.

What public data measures flex industrial tenant demand?

County Business Patterns counts establishments by employment size in the tenant trades (specialty trade contractors, repair and maintenance, truck transportation, warehousing) by ZIP code; the Nonemployer Statistics count the no-payroll tenants; and the Business Formation Statistics count new business applications monthly by county. Together they census the tenant base; no federal series measures flex occupancy or rents.

How is flex demand different from warehouse demand?

The tenants differ by an order of magnitude: the average general warehousing establishment employs 90.4 people, while the flex tenant trades average six to ten. Their cycles diverge too: warehousing employment peaked in 2022 and has been flat since, while business applications set a record in 2025 and are running 13.8% ahead in 2026.

Is there a public vacancy rate for flex industrial space?

No. No federal series measures flex vacancy or rents, and the fragmented ownership base (96.3% of nonresidential lessor establishments have fewer than 20 employees) thins commercial coverage as well. The defensible substitute is a tenant-per-bay ratio: the catchment's counted tenant base against its counted inventory from assessor and permit records.

What do Business Formation Statistics say about small business demand?

Applications for employer identification numbers reached 5,671,836 in 2025, the highest annual total on record and about 62% above 2019; the first seven months of 2026 ran 13.8% ahead of the same months of 2025. High-propensity applications, those most likely to become employers, peaked in 2023 and remain about 30% above 2019.

Is flex industrial a special purpose property for SBA lending?

The special purpose examples reported from SOP 50 10 8 name car washes, hotels, marinas, heavily refrigerated cold storage and service centers with pits and in-ground lifts; the ordinary light-industrial bay is not among them, and the SOP itself governs the full list. The boundary runs through the building: a generic bay changes tenant trades without a build-out, while a bay converted around fixed specialized improvements crosses toward the special purpose treatment.

Sources

1. U.S. Census Bureau, County Business Patterns 2023, national file (cbp23us), NAICS 2017 codes 238, 484, 4931, 493110, 531120, 8111 and all-industry totals by employment size class, released 26 June 2025. https://www2.census.gov/programs-surveys/cbp/datasets/2023/cbp23us.zip
2. U.S. Census Bureau, Business Formation Statistics, monthly national series (business applications and high-propensity applications), data through July 2026. https://www.census.gov/econ/bfs/csv/bfs_monthly.csv
3. U.S. Bureau of Labor Statistics, Quarterly Census of Employment and Wages, annual averages 2019 to 2024, NAICS 493110 General Warehousing and Storage, private, United States, 2020 to 2025. https://data.bls.gov/cew/data/api/2024/a/industry/493110.csv
4. U.S. Bureau of Labor Statistics, Quarterly Census of Employment and Wages, annual averages 2019 and 2024, NAICS 531120 Lessors of Nonresidential Buildings, private, United States, 2020 and 2025. https://data.bls.gov/cew/data/api/2024/a/industry/531120.csv
5. U.S. Census Bureau, Construction Spending (Value of Construction Put in Place), December 2025, release CB26-30, 27 February 2026. https://www.census.gov/construction/c30/pdf/pr202512.pdf
6. U.S. Census Bureau, Quarterly Retail E-Commerce Sales, 2nd Quarter 2026, 2026. https://www.census.gov/retail/mrts/www/data/pdf/ec_current.pdf
7. U.S. Census Bureau, Nonemployer Statistics 2023, released 15 May 2025. https://www.census.gov/newsroom/press-releases/2025/2023-nonemployer-statistics.html
8. 504 Capital Corporation, SBA 504 Loans for Special Purpose Properties and Real Estate (guidance on SOP 50 10 8 classifications), 2025. https://504capital.com/blog/financing-special-purpose-properties-sba-504-loans/
9. National Association of Government Guaranteed Lenders, SBA Notice Revising SOP 50 10 8 (Procedural Notice 5000-872764), 2025. https://www.naggl.org/sba-notice-revising-sop-50-10-8/
10. MMCG Research, SBA 7(a) Performance Series: MMCG analysis of the public SBA 7(a) loan register, 2026. https://mmcganalytics.com/sba-default-rates/
11. U.S. Bureau of Labor Statistics, Quarterly Census of Employment and Wages, 2024 annual averages, NAICS 238 Specialty Trade Contractors, private, United States, 2025. https://data.bls.gov/cew/data/api/2024/a/industry/238.csv
12. U.S. Census Bureau, County Business Patterns 2023 press release (18,564-establishment self-storage and related size-class tabulations cited across this series), 26 June 2025. https://www.census.gov/newsroom/press-releases/2025/2023-county-business-patterns.html