Business Support Services (U.S.) — NAICS 5614

A Histometrics rollup primer for a general investing audience — relevant to both public-market and private investors. NAICS (the North American Industry Classification System) is the U.S. government's standard scheme for grouping businesses; code 5614, Business Support Services, is a four-digit "industry group" that sits above six five-digit child industries. This page synthesizes the six already-written child primers plus our federal ground-truth statistics for this level. Figures are reported facts unless the wording marks them as projections or judgments.

1. Overview

Business Support Services is the federal filing cabinet for the outsourced back-office work that keeps commerce running but rarely carries a company's own name: answering its phones, chasing its unpaid bills, scoring its borrowers, preparing and shipping its documents, and running the mailbox-and-copy storefront on the corner. It is not one market. It is six loosely related service industries the Census Bureau grouped together because each is business-to-business support work that did not fit a larger, named category.

Because the group is a taxonomy convenience rather than a single competitive market, the honest and distinctive story here is the contrast across the six children — which is large, which is small, which is growing, which is shrinking, who owns them, and how differently each one actually makes money. That contrast is the whole point of looking at 5614 as a rollup rather than reading the children one at a time, and it is the subject of Section 2.

Three facts frame everything below. First, the level is enormous but lopsided: roughly $84 billion in 2022 receipts, but one child (telephone call centers) supplies a third of the revenue and well over half the jobs, while the smallest (document preparation) is barely 4% [1]. Second, it is extraordinarily fragmented in aggregate yet hides one concentrated oligopoly: the four largest firms across the whole group hold under 14% of revenue, but inside the credit-bureau child three firms dominate [1]. Third, the children run on genuinely different profit engines — labor arbitrage, data-network toll-collection, distressed-debt investing, retail service-stacking — so no single valuation lens fits the level, and an investor's route in depends entirely on which child they mean.

2. What's inside — the six children and how they differ

The single most useful thing this rollup adds is the comparison. Here is how the six stack up. (Shares are computed from our ground-truth federal file; "rev/employee" mixes 2022 receipts with 2023 employment and is a rough value-added proxy, not a margin.)

CR4 = share of revenue held by the four largest firms; HHI = Herfindahl-Hirschman Index, a standard concentration gauge that squares and sums market shares (below ~1,000 is "unconcentrated"). "Suppressed" means the federal file does not disclose that child's HHI, so none is stated.

And here is how they differ on the axes an investor actually cares about:

The headline contrasts:

• Size is dominated by one child. Telephone call centers (56142) alone are a third of the level's revenue and 57% of its jobs — because it is the most labor-intensive child, employing hundreds of thousands of agents at low pay. The next tier — collection, credit bureaus, and "other" — clusters at 16–18% of receipts each, and the storefront and document children are the small tail [1].

• Revenue-per-employee spans nearly 8×, and that is the real story. Call centers turn ~$73,000 of revenue per worker; credit bureaus turn ~$554,000 [1]. That gap is the difference between a people business (you sell hours) and a data business (you sell the same file over and over at almost no added cost). The level's average of ~$125,000 is a blend of these two opposite economic models — which is exactly why the aggregate is misleading.

• One child is concentrated; the rest are fragmented. The credit-bureau child has an HHI near 1,560 and a four-firm share of 75% — a genuine oligopoly on the nation's borrowing data [1]. Every other reported child is highly fragmented (HHIs of 82–207 where disclosed). That is why the public-market investor and the private investor gravitate to different corners of this level (Section 4).

• A big slice of "support services" is really a bet on consumer credit. Collection agencies (56144) and credit bureaus (56145) together are ~35% of the level's receipts, and both rise and fall with lending, delinquency, and the credit cycle — as does repossession inside the "other" child. So more than a third of this "business support" group is, economically, a credit-cycle play (Section 6).

• AI is the swing factor for the other half. Call centers, document preparation, and court reporting (inside 56149) are directly in the path of generative AI that can hold a conversation, draft a document, or produce a transcript. For those children AI is an existential threat to the billed hour; for credit bureaus it is mostly a tailwind (more analytics to sell). Same level, opposite exposures.

Where the code boundaries sit. Each child excludes its nearest neighbors, and those exclusions matter for sizing: credit scoring software and bond-rating agencies sit outside credit bureaus; a company's own in-house ("captive") call center is booked in the parent's industry, not in 56142; commercial printing and legal collection by law firms fall outside the storefront and collection children; and translation, data entry, and repossession each have their own codes [2]. The federal figures below count only what the code narrowly captures.

3. How big it is (this level's rollup)

Our federal ground-truth statistics for the combined level, NAICS 5614. Receipts, firm counts, and concentration are 2022 Economic Census; establishment, employment, and payroll counts are 2023 County Business Patterns (CBP) — two programs from different years, so do not divide one by the other to compute a true margin [1].

The rollup reconciles almost exactly. The six children sum to this level to the unit: establishments (4,375 + 4,141 + 9,878 + 2,877 + 408 + 7,492 = 29,171) and employment (36,170 + 386,187 + 65,081 + 91,811 + 26,469 + 67,678 = 673,396) match the level file precisely, and receipts (~$84.10 billion) and payroll (~$32.18 billion) reconcile to within rounding [1][3][4][5][6][7][8]. That is a useful confidence check: the level file and the child files are drawn from the same underlying census.

What the numbers reveal:

• A low-wage, labor-heavy group on average — but the average lies. Payroll is about 38% of receipts at face value and average pay is ~$47,800 [1] — the signature of a wage-driven service group. But that blends call-center agents near $39,000 with credit-bureau staff whose employer earns half a million dollars of revenue apiece. Read the level as two economies stapled together: a large low-margin labor pool and a small high-margin data core.

• Fragmentation, oddly, is lower at the level than in any reported child. The level HHI of 82.2 sits below the call-center child (144.7), the "other" child (126.8), and far below the credit-bureau child (1,560) [1]. That is arithmetic, not a market fact: pooling six industries that do not compete with each other makes every firm's share of the bigger pie smaller, so the squared-share index falls. The genuine competitive concentration lives inside the children — especially credit bureaus. Read 82.2 as "this is a bundle of separate markets," not "this is one giant competitive free-for-all."

• Most firms are small businesses. Average receipts per firm are ~$3.8 million [1], and every child's typical firm sits below its Small Business Administration (SBA — the federal small-business agency) size standard, which range from $19 million to $41 million of average annual receipts across the children [9]. Those thresholds are federal-contracting eligibility rules, not valuations, but they confirm that the great majority of the 22,259 firms qualify as small.

Undercount caveat — the $84 billion is an employer-only floor. These are employer counts: CBP and the Economic Census largely omit nonemployer businesses — the self-employed and firms with no paid staff. That undercount bites hardest where solo and micro-operators dominate: freelance résumé writers and transcriptionists (document preparation), one-person mailbox and copy shops (business service centers), owner-operator repossession and 1099 court reporters (inside "other") [3][7][8]. Our federal file contains no nonemployer total for the level. Two further scope effects push the same way: captive in-house operations are excluded by definition (a bank's own collections or service desk is booked with the bank, not here), and much of a real storefront's or transcript shop's activity is classified under adjacent codes (commercial printing, revenue-cycle management, software). So treat $84 billion as the visible employer core, not the whole ecosystem. Private research firms publish larger, blended "market size" numbers for individual children; they use broader definitions and disagree with each other, so anchor on the federal figure and treat vendor sizing as directional.

4. Investable universe (where value concentrates across the children)

For a public-market investor, the level's defining feature is that clean, listed exposure exists in essentially one child. For a private investor, the opposite is true — nearly every child is a target-rich roll-up field. (Tickers and company figures appear here and in Section 10 only; treat listed names as thematic proxies, not pure-plays, since most report diversified segments rather than a clean NAICS line.)

The one clean public quality play — Credit Bureaus (56145). This is the only child that maps to a set of large, high-margin, listed pure-plays: Equifax (NYSE: EFX), TransUnion (NYSE: TRU), the scoring franchise Fair Isaac / FICO (NYSE: FICO), and the largest by revenue, Experian (London: EXPN; U.S. over-the-counter EXPGY) [6]. These are growth-and-quality "toll booth" businesses, not income plays, and they behave nothing like the rest of the level. Private-equity does own specialty bureaus (Dun & Bradstreet was taken private by Clearlake Capital for $7.7 billion including debt in 2025), but the nationwide bureaus are effectively closed to new private entrants [6].

Public exposure to the credit-cycle child — Collection Agencies (56144). Reached mainly through a handful of listed debt buyers — Encore Capital (Nasdaq: ECPG), PRA Group (Nasdaq: PRAA), and the recently public Jefferson Capital (Nasdaq: JCAP) — plus diversified business-process-outsourcing (BPO) firms with collections arms [5]. Do not compare their revenue mechanically with Census receipts: debt buyers recognize portfolio income differently, carry receivable assets and funding debt, and operate internationally. The bulk of the child is private (Transworld Systems, GC Services, IC System, PE-backed platforms) [5].

Depressed public proxies plus a software layer — Call Centers (56142). The listed operators — Concentrix (Nasdaq: CNXC), TTEC (Nasdaq: TTEC), TaskUs (Nasdaq: TASK), ibex (Nasdaq: IBEX), and French-listed Teleperformance (Euronext: TEP) — trade at depressed valuations after an AI-driven sell-off [4]. The single largest is not U.S.-listed, and the biggest operators (Foundever, Alorica, Sutherland) are private/PE-owned. A separate lens is the adjacent contact-center software vendors (Five9, NICE, RingCentral, Twilio), which give exposure to AI as the disruptor rather than the disrupted [4].

Indirect and immaterial public exposure — Business Service Centers (56143) and Document Preparation (56141). No pure-play exists for either. Storefront exposure sits inside far larger parents — UPS (NYSE: UPS) and FedEx (NYSE: FDX), whose retail networks are a rounding error against ~$90 billion logistics businesses, plus adjacent online-print Cimpress (Nasdaq: CMPR) [7]. Document-preparation exposure is diluted inside transcription-technology owners (Microsoft, which owns Nuance; Solventum), the consumer-legal platform LegalZoom (Nasdaq: LZ), and physical-document proxies [3].

Thin and indirect — Other Business Support Services (56149). One narrow public segment (Pitney Bowes' mail-presort unit), plus infrastructure proxies for repossession (Motorola Solutions' license-plate-data network; auction houses Copart and RB Global) and credit exposure to the court-reporting roll-ups via publicly traded business development companies (BDCs — listed lenders to private firms, e.g., Trinity Capital) [8].

Bottom line for public investors: value concentrates in the credit-data child (56145) and, more speculatively, the collection and call-center children. Everything else is a diversified parent or a private-market proposition. For private investors, the level is the main event — a fragmented, cash-generative, contract-based services universe (Section 10).

5. How the money works

Strip away the labels and the level runs on two opposite economic models, plus a hybrid.

Model one — sell hours (the labor children: 56142, 56141, most of 56143 and 56149). These are labor-arbitrage service businesses. Profit is the spread between what you bill for a unit of work — an agent-hour, a transcript page, a packed parcel, a résumé — and the fully-loaded cost of the person doing it, multiplied by how busy you keep them. The metrics that decide whether an owner makes money are operational and shared: utilization/occupancy (billable hours ÷ paid hours), attrition and ramp cost (turnover is chronically high, and every departure means re-hiring and re-training), revenue per productive hour and mix (complex, regulated work bills more than commodity work), and quality gates that trigger bonuses or penalties. Margins are thin — typically mid-single-digit to low-double-digit earnings before interest, taxes, depreciation, and amortization (EBITDA) — because labor dominates and much of the work is commoditized. The classic levers are offshoring (call centers run most agents in the Philippines, India, and Latin America) and pooling many small clients onto a shared roster to kill idle time [4][3].

Model two — sell the same data twice (the data child: 56145). Credit bureaus are the opposite business. Lenders furnish borrower data for free under a reciprocity system, and the bureau sells it back per report, per score, and per batch screened. Because the data is already collected, each additional report costs almost nothing, so incremental margins are very high — the listed bureaus run adjusted-EBITDA margins in the mid-30s to low-40s percent, and FICO's per-score royalty runs operating margins near 90% [6]. This is a data network with a near-insurmountable moat (a new entrant cannot rebuild decades of contributed history), and it is why one child earns ~$554,000 of revenue per employee while the rest earn a fraction of that.

The hybrid — buy distressed cash flows (part of 56144). Debt buyers are neither a labor shop nor a data toll: they purchase charged-off receivables and collect for their own account, so they are watched like a specialty-finance investor — portfolio purchase multiples, estimated remaining collections (ERC — the backlog of cash expected from portfolios already owned), cash collections versus cost to collect, and leverage, since purchases are debt-funded and interest rates hit returns directly [5]. Contingency (fee) agencies inside the same child are simpler: revenue ≈ recovery rate × fee percentage × placement volume, minus labor and compliance cost.

The unifying financial trait of the whole level is that returns come from scale, efficiency, utilization, contract retention, and — for owners — buy-and-build consolidation, not from organic pricing power. The exceptions that do have pricing power (credit bureaus, the license-plate-data network inside 56149, a virtual-mailbox subscription platform inside 56143) are exactly the assets where durable value concentrates. Standard factory "capacity utilization," retail "same-store sales," regulated-utility "rate base," and real-estate "funds from operations" lenses do not fit this level; revenue-per-productive-hour and contribution-margin-per-seat fit the labor children, and EBITDA/free-cash-flow yield and (for debt buyers) collection multiples fit the rest.

6. Demand drivers

Because the children answer to different forces, the level has no single demand driver — but three cross-cutting themes tie most of it together.

• The credit cycle — roughly a third of the level. Collection agencies (56144), credit bureaus (56145), and repossession inside 56149 all track consumer and business lending. U.S. household debt reached about $18.8 trillion by early 2026, with credit-card balances near $1.25 trillion; the end of the student-loan payment pause pushed 90-day-plus student-loan delinquency toward ~10% on roughly $1.66 trillion of balances, and auto repossessions hit an estimated 1.73 million in 2024, the most since the Great Recession [5][8]. Note the split personality: bureaus are pulled by lending volume (mortgage most of all), while collectors and repossessors are pulled countercyclically by delinquency — so within the credit-linked children, demand peaks at different points of the cycle.

• The make-versus-buy (outsourcing) decision. Call centers, medical coding, mail presort, and business-service storefronts all exist because companies choose to buy support functions rather than staff them. Corporate cost pressure and recessions tend to push more volume out the door to outsourcers, making parts of the level counter-cyclical to their clients' budgets [4][8].

• AI, cutting both ways. Generative AI is the dominant swing factor for the labor children: it destroys demand where it deflects simple calls, drafts documents, or transcribes hearings, and it creates demand where cheap automation pulls in buyers who never bought the service (a micro-business that could never afford a human receptionist) or where enterprises need help deploying and supervising the AI itself. For the data child it is mostly additive — more analytics and fraud/identity products to sell [4][3][6].

Layered on top are child-specific drivers: e-commerce and returns traffic (business service centers), litigation volume (court reporting), small-business formation and remote-work address demand (mailbox and document services), and accessibility and compliance mandates. Demand across the level is economically sensitive but less tied to physical production than manufacturing — it tracks clients' activity and, for a large slice, the health of consumer credit.

7. Regulation

There is no single regulator or license for the level; regulatory intensity is sharply bifurcated, and that split is itself a diligence signal.

• The consumer-finance children are among the most regulated businesses in the country. Collection agencies live under the Fair Debt Collection Practices Act (FDCPA, 1977) and the Consumer Financial Protection Bureau's (CFPB) Regulation F (validation notices, the "7-in-7" call cap), plus state licensing and bonding [5]. Credit bureaus live under the Fair Credit Reporting Act (FCRA, 1970) — accuracy, disputes, permissible purpose, adverse-action notices — plus the Gramm-Leach-Bliley Act and Federal Trade Commission (FTC) data-security rules; the 2017 Equifax breach settlement (at least $575 million) is the cautionary reference case [6]. A cross-cutting development: federal oversight pulled back sharply in 2025–2026 (CFPB staff and budget cuts, a vacated medical-debt rule), pushing compliance risk toward state attorneys general rather than eliminating it — a theme that touches both credit-linked children at once [5][6].

• The phone children carry calling and data rules. Call centers face the Telephone Consumer Protection Act (TCPA) enforced by the Federal Communications Commission (FCC) — restricting autodialed and prerecorded "robocalls," with AI voices increasingly treated as covered "artificial" calls — and the FTC's Telemarketing Sales Rule (TSR) and Do-Not-Call Registry on the outbound side [4]. Answering and service operations that touch health or payment data pick up the Health Insurance Portability and Accountability Act (HIPAA) and card-data rules.

• The storefront and "other" children carry activity-specific rules. Mailbox stores are Commercial Mail Receiving Agencies (CMRAs) under U.S. Postal Service know-your-customer rules; franchisors owe a Franchise Disclosure Document under the FTC Franchise Rule; repossession runs on the Uniform Commercial Code's "breach of the peace" limit and a lender's non-delegable duty; court reporting is gated by state stenographer-licensing rules; and charitable fundraising requires registration in most states [7][8].

Common thread: in every child, compliance quality is simultaneously a competitive asset and a material liability — a CMRA delivery suspension, a TCPA class action, an FCRA accuracy failure, a wrongful repossession, or a data breach can be existential for a small operator and is a core diligence item for a buyer.

8. Consolidation

The level's structural signature is a barbell: extreme fragmentation at the base (thousands of tiny operators, low measured concentration) with durable value consolidating at the top — but the mechanics differ by child.

• Already consolidated: credit bureaus (56145) are a stable oligopoly whose roster has barely changed in decades; growth comes from buying specialty data and analytics firms, and the headline 2025 move was PE's take-private of Dun & Bradstreet [6].
• Consolidating at the top, fragmented at home: call centers (56142), where labor-arbitrage BPO is a scale game — Concentrix's ~$4.8 billion acquisition of Webhelp (2023) and the Sitel–SYKES merger into Foundever created a global oligopoly that can fund the AI reinvention clients now demand [4].
• Classic PE roll-ups of cottage industries: collection agencies, court reporting (Veritext, Lexitas, U.S. Legal Support), business-service-center franchisors (Annex Brands, Fortidia), and the virtual-mailbox software layer (iPostal1) — all aggregating small units into scaled, recurring-revenue platforms [5][7][8].
• Consolidating around the field, not within it: repossession, where local agencies resist roll-up (state licensing, local relationships) so value migrates upward to asset-light "forwarders" and a license-plate-data near-monopoly [8].

The recurring lesson: consolidation concentrates margin at data, scale, and infrastructure chokepoints, but it does not eliminate local competition, and a broad roll-up across unrelated children mixes incompatible businesses — capital works child-by-child, not code-wide.

9. Risks

• AI substitution and price deflation — the defining risk for the labor half. Live deployments already replace agents at scale in call centers (public markets have repriced the sector severely), commodity document and transcription work is drifting toward near-free AI economics, and word-processor and medical-transcriptionist occupations are among the fastest-declining in the country [4][3]. The data child (56145) is largely insulated and may benefit.
• Regulatory whiplash — the defining risk for the credit half. A stricter future administration, a revived national medical-debt rule, or aggressive state attorneys general could re-tighten collection and credit-reporting rules quickly, on top of constant litigation and complaint exposure [5][6].
• Cyclicality and end-market concentration. More than a third of the level rides consumer credit; storefronts ride e-commerce and small-business formation; call centers ride client budgets. Losing or repricing one large contract can strand facilities and trained staff in any labor child [4][5].
• Thin margins with little cushion. Across the labor children, low profitability leaves scant room for wage inflation, chronic turnover, or a lost program.
• Data security and privacy liability. Operators across the level hold sensitive personal, medical, financial, and business data; a serious breach is existential for a bureau and costly for anyone else [6][4].
• Roll-up and leverage risk. The PE platforms that own the best assets carry meaningful, often undisclosed debt; integration missteps or rising rates strain the most levered operators.
• Classification and data risk. Employer-only federal data understate the contractor- and nonemployer-heavy reality, and third-party "market size" figures routinely blur these children with adjacent codes — read labels carefully before sizing a market or a target.

10. How to invest and the outlook

Public routes — narrow and uneven. The cleanest, highest-quality listed exposure is the credit-bureau child: Equifax (EFX), TransUnion (TRU), FICO, and Experian (EXPGY) — durable, high-margin data toll booths, priced accordingly, exposed to the mortgage cycle and, for the first time in years, to real competition over the score at the center of the system [6]. The collection child offers the listed debt buyers (ECPG, PRAA, JCAP), valued on cash generation, ERC growth, purchase multiples, and leverage rather than a simple price-to-earnings multiple [5]. The call-center child offers contrarian, higher-beta bets on depressed operators (CNXC, TTEC, IBEX, TASK, TEP) that AI augments rather than eliminates, or the opposite tilt via contact-center software (Five9, NICE, RingCentral, Twilio) [4]. The remaining children (storefronts, documents, "other") are reachable publicly only through diversified parents (UPS, FDX, Cimpress, Pitney Bowes, Motorola Solutions) or BDC credit — treat those as parcel, logistics, print, or lending investments, not clean plays [7][8]. There is no listed pure-play, exchange-traded fund (ETF), or real estate investment trust (REIT) for the level as a whole.

Private routes — where most of the level actually trades, chosen by child:

• Buy or operate. SBA-friendly targets abound — a call-answering service, a copy-and-ship franchise, a contingency collection agency, a court-reporting or medical-coding shop. Prize recurring revenue, low churn, vertical specialization, documented compliance, low owner-dependence, and AI wrapped around the workflow rather than competing with it on price [3][4][7][8].
• Back a roll-up or PE platform. Every fragmented child draws sponsors — a scale play in call centers, a fragmentation play in collection and court reporting, a franchisor play in storefronts [4][5][7][8].
• Venture and specialty credit. Fund the AI-native disruptors (receptionists, ambient scribes, autonomous coding) or the data/infrastructure layer, or extend private credit secured by fee streams.

Diligence across the level centers on verified cash flow and revenue-by-service-line; contract and customer concentration; utilization, attrition, and fill rate; carrier, licensing, and compliance records; leverage; and — everywhere — a credible answer to whether AI is the target's tailwind or its obituary.

Outlook (forward-looking judgment; the federal data contain no growth forecast). The level splits into two stories running in opposite directions. The data and credit core (credit bureaus, and the supply side of collection) looks durable and even favored near-term — record household debt, normalizing delinquencies, and a lighter federal enforcement posture expand the collectible and scorable pool, while the bureaus' moats hold. The labor core (call centers, document preparation, court reporting, commodity storefront work) faces a structural, AI-driven squeeze, where the winners reposition around complex, regulated, high-value, human-in-the-loop work and the AI-and-infrastructure layer, and the losers are undifferentiated sellers of the plain billed hour. For the level as a whole, value is created less by the sector expanding than by operators improving mix and consolidators rolling up small units and data networks. Public investors get concentrated quality in one child and contrarian optionality in two others; private investors get a rich, heterogeneous roll-up field — provided they underwrite each child and niche on its own economics and treat the four-digit code as a filing convenience, not a market.

---

Sources

Drawn from the six child primers (56141, 56142, 56143, 56144, 56145, 56149) and our ground-truth federal statistics for this level.

1. U.S. Census Bureau. 2022 Economic Census (Concentration of Largest Firms) and County Business Patterns 2023 — NAICS 5614, Business Support Services (receipts $84,095,887K; 22,259 firms; 29,171 establishments; 673,396 employees; annual payroll $32,176,563K; Q1 payroll $8,531,706K; CR4 13.9% / CR8 22.4% / CR20 33.3% / CR50 44.4%; HHI 82.2). 2022–2025. https://data.census.gov/
2. U.S. Census Bureau. 2022 NAICS Manual — codes 5614 and children 56141, 56142, 56143, 56144, 56145, 56149; definitions and exclusions (captive centers; scoring software; commercial printing; translation; legal collection; repossession). 2022. https://www.census.gov/naics/
3. U.S. Census Bureau et al. NAICS 56141 / 561410 — Document Preparation Services (child primer: receipts ~$3.51B; 36,170 employees; AI displacement of transcription/typing; Microsoft-Nuance, Solventum, LegalZoom proxies). 2022–2026. https://data.census.gov/
4. U.S. Census Bureau; SEC filings; Bloomberg. NAICS 56142 — Telephone Call Centers (child primer: receipts ~$28.17B; 386,187 employees; HHI 144.7; Concentrix/Webhelp; Teleperformance, TTEC, TaskUs, ibex; TCPA/TSR; contact-center software). 2022–2026. https://data.census.gov/
5. U.S. Census Bureau; SEC filings; Federal Reserve Bank of New York; CFPB. NAICS 56144 / 561440 — Collection Agencies (child primer: receipts $15.16B; 91,811 employees; debt buyers ECPG/PRAA/JCAP; FDCPA/Regulation F; household debt ~$18.8T). 2022–2026. https://data.census.gov/
6. U.S. Census Bureau; SEC and LSE filings. NAICS 56145 / 561450 — Credit Bureaus (child primer: receipts $14.67B; 26,469 employees; CR4 75.1%, HHI 1,559.8; Equifax/TransUnion/Experian/FICO; Dun & Bradstreet–Clearlake; FCRA; Equifax breach settlement). 2022–2026. https://data.census.gov/
7. U.S. Census Bureau; SEC filings; franchise disclosures. NAICS 56143 — Business Service Centers (child primer: receipts $8.72B; 65,081 employees; CR4 36.6%; The UPS Store, FedEx Office, Cimpress; CMRA rules; iPostal1 virtual mailbox). 2022–2026. https://data.census.gov/
8. U.S. Census Bureau; SEC filings; Cox Automotive; industry sources. NAICS 56149 — Other Business Support Services (Repossession, Court Reporting, All Other) (child primer: receipts $13.87B; 67,678 employees; HHI 126.8; Pitney Bowes presort, Motorola LPR, Copart, court-reporting PE roll-ups, BDC credit). 2022–2026. https://data.census.gov/
9. U.S. Small Business Administration. Table of Small Business Size Standards — NAICS 5614 children ($19.0M–$41.0M average annual receipts). 2023. https://www.sba.gov/document/support-table-size-standards