SBA Size Standards vs Lower Middle Market: A PE Guide

Executive Summary

SBA size standards are not a lower-middle-market screen. The U.S. Small Business Administration (SBA) generally uses employees or average annual receipts to set an industry's maximum small-business size, including relevant subsidiaries and affiliates. ^[1] A private equity (PE) mandate instead specifies the financial scale and company characteristics its buyer seeks. There is no defensible automatic conversion from a program flag to target revenue, earnings, ownership, or acquisition attractiveness.

The distinction is visible in the numbers. The National Center for the Middle Market defines the broad middle market as $10 million to $1 billion in annual revenue, which should not be relabeled as a universal lower-middle-market band. ^[2] Published buyer criteria differ: Great Point Partners targets $10 million to $100 million revenue and $2 million to $12 million EBITDA, while Shore Capital Partners publishes different platform and add-on profiles. ^{[3] [4]} EBITDA means earnings before interest, taxes, depreciation and amortization. A flag based on employees cannot supply either financial metric.

Version control is essential as of October 11, 2026. SBA's table landing page identifies a downloadable March 2023 version effective March 17, 2023. ^[1] The August 2026 proceeding proposed new standards for 338 industry groups and industries, alongside a revised methodology. ^{[5] [6]} Its comment deadline was extended to November 20, 2026. ^[7] Proposed thresholds should remain in a monitoring register until an applicable final rule establishes their status and effective date. Program records also need an observation date: SBA warns that SAM.gov can show status under old standards until registration is updated. ^[1]

The recommended workflow preserves the official classification in one layer and evaluates buyer-defined size in another. Retain the source, legal entity, North American Industry Classification System (NAICS) code and vintage, threshold unit, applicable rule, and representation date. Independently source the mandate's revenue, EBITDA, financial period, and company scope. Acquisition events need qualified review: FAR 19.301-2 specifies rerepresentation within 30 days after certain mergers or acquisitions. ^[8] This report's tables and fictional examples show how to retain useful evidence while leaving missing financial metrics and unresolved affiliation questions visible. It determines no real company's eligibility.

Introduction and Background

SBA size standards vs lower middle market compares two different decisions. The U.S. Small Business Administration (SBA) sets industry-specific upper limits for classification within its programs and federal contracting. Private equity (PE) buyers define acquisition size through their own investment criteria. The first is a program rule; the second is a commercial mandate. The Federal Acquisition Regulation (FAR) explicitly describes SBA standards as industry-by-industry, while published buyer criteria specify their own financial ranges. ^{[9] [3]}

The distinction matters when an origination team imports a government-services universe, supplier directory, or commercial dataset containing a small-business flag. That flag may be useful evidence about a particular representation or program. It does not supply the missing revenue or profitability observation. SBA explains that standards normally use employee counts or average annual receipts, which are different measurement axes. ^[1] A dataset containing the flag therefore needs a field definition before the team can decide how to use it.

External uses illustrate why context matters. California's State Trade Expansion Program (STEP) applies SBA standards in its eligibility requirements. ^[10] Walmart describes verification through SBA certification or its internal process. ^[11] SLAC National Accelerator Laboratory describes vendors self-certifying small-business status in the System for Award Management (SAM.gov). ^[12] These are different collection and application settings, even when a record ultimately refers to the same official threshold table.

Industry classification adds another layer. The North American Industry Classification System (NAICS) is the federal statistical standard for classifying business establishments. ^[13] A code identifies activity; it does not establish investment scale. Likewise, the Environmental Protection Agency (EPA) explains SBA's inclusion of parents and affiliates when describing small-business definitions. ^[14] A brand-level employee count and an affiliation-inclusive program size can describe different populations.

This report supplies a specific translation protocol: preserve the official observation and its provenance, then build a separately evidenced buyer size band. It does not determine a real company's program eligibility or substitute an investment attractiveness score for a size classification. Its practical deliverables are an industry-linked reference table, a two-layer worksheet, and a comparison of published buyer bands.

Key Changes

Change the decision field before changing the filter

The first process change is to stop using SBA small as a synonym for lower middle market. SBA describes its threshold as the largest size compatible with small-business classification for SBA and federal contracting programs. ^[1] The National Center for the Middle Market (NCMM), by contrast, defines the broad middle market using annual revenue of $10 million to $1 billion. That research definition covers the broad middle market, rather than establishing a universal lower-middle-market boundary. ^[2]

A buyer's published mandate is the more relevant source for its acquisition screen. Great Point Partners publishes $10 million to $100 million of revenue and $2 million to $12 million of earnings before interest, taxes, depreciation and amortization (EBITDA). ^[3] HC Private Investments publishes a different revenue band and minimum EBITDA. ^[15] These observations establish variation among the sampled buyers. They do not justify selecting a universal boundary from the smallest or largest endpoints.

The operational consequence is simple: change a column named company_size into explicit fields for sba_status_observed and mandate_size_result. Preserve the first as evidence. Derive the second only from the mandate's required metrics. This is an analytical recommendation, rather than a new regulatory classification.

Change version control from page date to rule provenance

The SBA landing page identifies a March 2023 downloadable version effective March 17, 2023. ^[1] The eCFR page read for this report displayed Title 13 as current through October 7, 2026. That is a title-level currency statement, not the effective date of every industry row. ^[16] A collector should therefore store document vintage, regulatory effective date, retrieval date, and record-observation date separately.

The manufacturing final rule provides a directly stated effective date of March 17, 2023. ^[17] This is stronger evidence of when that rule took effect than a filename or a recently updated landing page. A recent access timestamp cannot repair an unidentified historical version.

A concrete provenance check found NAICS 311941 at 650 employees in the linked March 2023 SBA PDF, compared with 850 employees in the current eCFR. ^{[18] [16]} The manufacturing final rule also sets that row at 850 employees, with its stated March 17, 2023 effective date. ^[17] This is a document discrepancy, not evidence that the higher threshold first took effect after that date. Preserve the downloaded version and reconcile the row against the operative regulation and final rule before interpreting a status flag.

Change proposal monitoring without changing current thresholds

On August 20, 2026, SBA published proposed standards for 338 industry groups and industries, alongside a proposed revised methodology. ^{[5] [6]} A subsequent notice extended comments to November 20, 2026. ^[7] As of this report's publication date, those documents are proposal evidence, not instructions to replace operative thresholds.

Keep a separate change-monitoring register. It should record the proposal identifier, publication date, comment deadline, affected codes, proposed unit, and whether a final rule with an effective date has appeared. A proposed shift from receipts to employees would require revising data fields as well as values. The proposal discusses grouping standards at four-digit and five-digit NAICS levels, which also makes code granularity a monitoring concern. ^[5]

Industry Standards and Their Limits

Read the unit, code, and exception together

Section 121.201 of Title 13 of the Code of Federal Regulations (13 CFR) describes standards as employees or annual receipts in millions of dollars unless otherwise specified. The threshold applies to the concern and its affiliates. ^[1] Consequently, an unqualified number such as 750 is unusable without its unit, industry, rule version, and aggregation scope.

Table 1 provides a deliberately small reference asset. The codes and thresholds were read in the current eCFR table; the downloadable SBA table is used for its identified document vintage. Receipt-based entries should be applied using the relevant receipts-calculation provision, rather than treated as reported latest-year sales. SBA contracting guidance describes a completed-fiscal-year average. ^[19] This table supports source checking, not a company-specific eligibility verdict.

The interpretive point is the non-equivalence, rather than the relative magnitude of these numbers. Employee thresholds and receipts thresholds cannot be ranked on one acquisition-size axis. The date column distinguishes the identified table vintage from the manufacturing rule's stated effective date. For a live program matter, the reviewer must establish the applicable rule and transaction or procurement date before making a classification.

Read averages as averages

For the general receipts method, the regulation averages the latest five completed fiscal years, subject to its stated exceptions. ^[20] The employee calculation generally uses pay periods over the preceding completed 24 calendar months. ^[21] Part-time and temporary employees count the same as full-time employees for that calculation. ^[21] These definitions do not describe a single website snapshot, full-time-equivalent staffing, or a buyer's last-twelve-month EBITDA.

The data analyst should ask what each observation actually measures. A financial statement may concern one fiscal year and one entity. A payroll-derived average may concern a different period and affiliation scope. A program label may identify only the final representation. Store these as different observations, even when a spreadsheet would otherwise place them in one size column.

Read NAICS as an activity classification

Census explains that NAICS groups establishments according to similar production processes; its Economic Census guidance identifies the 2022 NAICS vintage used for that census. ^[22] Census also notes that an enterprise can contain establishments in different industries. ^[23] A multi-activity business therefore needs a scope record, rather than one automatically inherited code for every purpose.

For procurement, the contracting officer selects the NAICS code that best describes the acquisition's principal purpose. ^[9] This helps explain why a firm's vendor-assigned primary code, establishment classification, and contract-specific code can differ. Keep all supported codes with their roles. Do not replace a contract code solely because another code better describes the PE thesis.

Implementation Considerations and Process Changes

Build a two-layer screening worksheet

The recommended design separates program evidence from mandate evidence while keeping them connected to the same resolved entity. GSA's registration guidance asks businesses to confirm size-standard information using SBA guidance and keep ownership details consistent across relevant records. ^[24] Those instructions support checking identity and provenance before interpreting a flag.

Table 2 compares the fields needed for each decision. The research-delivery row places Docket's documented offering on the sourcing axis. It does not place a research provider inside the official threshold system.

The worksheet prevents an official field from silently becoming a buyer field. Docket describes researched answers as containing a source, supporting excerpt, and collection date. Those are useful provenance components, but a supported answer still needs the correct metric and entity scope. Research delivery cannot make a legally scoped employee average interchangeable with acquisition revenue.

Preserve the raw observation before deriving fit

Use the following recommended collection sequence:

• Resolve identity: attach the observation to the legal concern and record the trading name separately.
• Preserve raw status: retain the original flag, label, program, and source wording.
• Record observation date: distinguish the representation date from the date the analyst downloaded it.
• Attach code provenance: retain NAICS vintage, code role, and any relevant exception.
• Attach rule provenance: identify the official row, threshold unit, and effective rule.
• Collect mandate metrics: source revenue and EBITDA independently, with period and scope.
• Apply the buyer rubric: evaluate the documented platform or add-on criteria.
• Retain the gap: return pending when required evidence is unavailable.

This is a proposed operating procedure. It makes the non-equivalence auditable without requiring the collector to decide program eligibility. California's STEP guidance illustrates the scope issue by factoring the parent's and subsidiaries' size into its stated application requirements. ^[10] That program example should not be generalized into every contracting or lending rule.

Keep the company boundary visible

Census defines an establishment as a single physical location and an enterprise as an organization with establishments under common ownership or control. ^[23] These definitions help identify the measurement boundary, but the SBA affiliation boundary must still be established under the applicable program rules.

Recommended scope controls are:

• Parent observation: identify whether evidence describes a holding group.
• Subsidiary observation: avoid inheriting parent metrics without a scope decision.
• Carve-out observation: identify the assets and entities the mandate considers.
• Location observation: label site-level staffing or sales explicitly.
• Aggregate observation: record the consolidation or affiliation method.
• Identifier observation: treat a Unique Entity ID as identity evidence, not a size metric; SAM.gov says an ID alone does not enable direct applications for federal awards. ^[25]

Keep conflicting observations in the record. Resolution should state which entity, period, metric, and decision the chosen value supports. A collector should not select whichever observation produces a favorable acquisition-fit result.

Affiliation and Acquisition Review

Review control, not just the operating-company label

General affiliation principles consider the power to control even when it is not exercised, and examine ownership, management, and relationships. ^[26] They also describe consideration of the totality of circumstances and domestic and foreign affiliates. ^[26] These provisions make affiliation a qualified review issue for sponsor-controlled groups and other multi-entity structures.

An analyst should therefore flag a private-equity ownership observation for review without turning it into a universal eligibility conclusion. Ownership percentage alone may not document the relevant rights, exceptions, or program basis. The useful output is a scope question with supporting evidence, rather than an automated statement that every portfolio company must have the same size status.

Program context matters. Section 121.301 supplies affiliation principles for Business Loan, Disaster Loan, and Surety Bond Programs. ^[27] The receipts provision also includes stated alternative averaging treatment for covered financial-assistance programs. ^[20] An origination database should therefore distinguish contracting evidence from lending evidence before it applies an aggregation or averaging rule.

Treat acquisitions as review events

The receipts rule explains that acquisition-related aggregation can apply over the entire measurement period, with specific treatment for acquired divisions. ^[20] Contracting rules also address timing. FAR 19.301-2 specifies rerepresentation within 30 days after a merger or acquisition that does not require novation, under its stated conditions. ^[8] A team should send an identified transaction to a qualified government-contracting adviser rather than infer the legal consequence from the prior SAM flag.

Recommended review triggers are:

• New ownership: a documented sponsor, parent, or controlling shareholder change.
• Control rights: new governance evidence requiring program-specific analysis.
• Acquired entity: a business combination that may change aggregation scope.
• Divisional transaction: a carve-out or asset purchase needing its own scope analysis.
• Contract event: an award, renewal, or rerepresentation with a different relevant date.
• Conflicting program labels: contracting and financial-assistance observations that use different bases.
• Incomplete structure: an unresolved parent or affiliate relationship.

The adviser handoff should include the relevant program, legal entities, dated ownership evidence, rights documents if available, code, threshold vintage, measurement records, and contract dates. It should request a scoped answer for the specific matter. No qualified adviser review of an actual company was performed for this report; these are generic research and escalation procedures.

Separate investment scope from affiliation scope

A buyer might evaluate the earnings of an operating subsidiary while the applicable program calculation considers affiliates. FAR describes size status as including affiliates at the relevant contracting representation. ^[9] The acquisition team should preserve the commercial scope it intends to buy and the separately documented program scope, even if their totals differ.

The same principle applies to company type. A small-business label does not establish founder ownership, recurring revenue, margin quality, customer concentration, or willingness to sell. Millpond Equity Partners, for example, separately identifies customer retention and recurring revenue among its sought characteristics. ^[28] These belong in independently sourced mandate fields.

Data Analysis and Evidence

Published buyer bands are heterogeneous

Table 3 compares selected first-party investment criteria retrieved for this report. It is a purposive sample, selected to show different published size axes. It is not a survey of the PE market, and its endpoints do not estimate the distribution of acquisition targets. The columns preserve revenue, EBITDA, and transaction value as separate quantities.

These rows support a bounded conclusion: the sampled mandates differ in their financial bands and structure. Shore's separate add-on criteria also demonstrate that the investment role matters within a single buyer's published approach. ^[4] The team's own mandate must specify whether the screen concerns a platform, add-on, minority investment, or another transaction type.

Additional published criteria reinforce the point without extending the table. Millpond specifies $3 million to $15 million EBITDA. ^[28] Mereo Capital Partners specifies $15 million to $200 million revenue and $3 million to $15 million normalized EBITDA. ^[31] Normalized EBITDA should be preserved as the buyer's stated metric, rather than silently equated with an unadjusted earnings figure. These are observations of published criteria as retrieved, not independently measured portfolio characteristics.

What the official numbers establish

The August 2026 proposal describes the existing system as 102 threshold levels, covering 978 NAICS industries and 18 subindustries or exceptions. ^[5] Those figures explain why a generic national small-business flag needs industry and rule provenance. They count elements of the regulatory framework; they do not count buyable companies, estimate revenue, or provide a lower-middle-market market size.

NCMM's broad revenue span also requires segmentation in its research. ^[2] That statistical context should remain separate from the sampled PE criteria. A research institution's population definition and a fund's investment mandate answer different questions, even when their revenue ranges overlap.

Measure the screen's evidence, not a proxy conversion

Recommended internal quality measures are:

• Status provenance coverage: records with a status source and observation date divided by records containing a status flag.
• Version coverage: records with a documented applicable code vintage and rule version divided by records requiring program interpretation.
• Mandate-size coverage: records with all required financial metrics on the correct scope divided by screened entities.
• Unresolved-scope share: records awaiting entity or affiliation review divided by records assessed for scope.
• Conflict rate: records with incompatible supported observations divided by records with multiple observations.
• Ready-for-review share: records with documented mandate fit and completed required checks divided by the researched universe.

These are proposed management calculations. Define each denominator before reporting a result, and display unknowns separately. No conversion factor from employees to revenue, and no empirical acquisition-fit accuracy, is established by this report's source set.

Case Studies and Real-World Examples

Employee-based flag (Hypothetical Example)

A fictional food manufacturer appears in a supplier dataset as small under NAICS 311911. The relevant current table lists 750 employees as the standard. ^[16] The analyst has the flag and its source date, but no independently supported company revenue or EBITDA. The record cannot be placed inside a revenue-and-EBITDA acquisition mandate.

The recommended output is program status observed; mandate size unknown. If the dataset also contains a current website headcount, preserve it as a separate observation. It is not automatically the payroll-period average required by the employee rule. ^[21] Additional research should seek the mandate's financial metrics and establish their entity scope. The example supplies no estimate of this fictional firm's revenue from its employees.

Receipt-based flag (Hypothetical Example)

A fictional programming-services company is marked small under NAICS 541511, whose current receipts threshold is $34 million. ^[16] The acquisition rubric asks for a specified revenue band and positive EBITDA, but neither required financial observation has been sourced.

Even a correctly documented classification would not reveal the company's exact revenue, identify its position within the threshold, or establish profitability. SBA contracting guidance describes a completed-fiscal-year receipts average. ^[19] The recommended output remains buyer size pending, with separate research questions for revenue period, EBITDA basis, and the contemplated transaction scope. A program threshold is an upper limit, not an observed point estimate or a buyer's minimum size.

Old representation and new ownership (Hypothetical Example)

A fictional government-services record contains an active SAM registration and a small-business flag collected before an ownership change. The analyst has no evidence that the size representation was updated afterward. SBA warns that a SAM profile can display status under old standards until registration is updated. ^[1] SAM.gov permits updates at any time or during renewal. ^[25]

The recommended output is historical observation retained; current interpretation pending. Preserve the original date and collect the changed ownership evidence. Check the applicable program and contract events, then route the matter for qualified review. FAR's merger-and-acquisition rerepresentation provision is a reason to review the relevant event and timing, not a basis for this fictional analyst to declare a new eligibility result. ^[8] The record's investment size also remains unknown until separately sourced.

Implications and Future Directions

Refresh on the decision's relevant date

SAM.gov requires registration renewal every 365 days, but an active registration should not be treated as a guarantee that every underlying research field was recently confirmed. ^[25] Its documentation also permits updates between renewals. ^[25] Recommended event-driven refreshes should therefore accompany any routine calendar review.

Use a field-level precedence policy:

• Official threshold: use the applicable regulation and final rule for the relevant date; retain the table as a versioned reference.
• Procurement code: use the documented contract or solicitation code for that procurement question.
• Program representation: preserve the dated representation and identify who made it.
• Buyer criteria: use the approved mandate version and transaction role.
• Financial observation: prefer evidence with an explicit metric, period, and entity scope.
• Ownership event: retain the event date and supporting source, then reassess affected fields.
• Unresolved conflict: keep both supported observations and assign the open question.

FAR instructs the contracting officer to apply the standard effective when the solicitation is issued. ^[9] Rerepresentation uses the standard effective at that later event under the specified contract code. ^[8] A follow-on or renewal contract is a new contracting action for size purposes. ^[32] This is why a current threshold lookup cannot, by itself, settle a historical procurement question.

Monitor changes as separate evidence

The 2026 threshold and methodology proceedings are material monitoring items because their comment deadline extends beyond this report's publication date. ^[7] Census also lists proposed 2027 NAICS updates. ^[13] Neither a proposal nor an anticipated classification revision should overwrite a currently supported record.

If a final rule changes a threshold unit, averaging basis, exception, or code grouping, identify the affected observations and re-run the appropriate program checks. Maintain the buyer mandate independently. A regulatory change can alter the official interpretation while leaving the PE revenue-and-EBITDA criteria untouched. Conversely, a buyer can change its mandate without any change to an SBA rule. Keeping those histories separate makes both kinds of update reviewable.

Conclusion

An SBA small-business flag belongs in a PE research record when it answers a relevant program or procurement question and carries usable provenance. It should not fill a missing revenue field or stand in for a lower-middle-market acquisition band. Official standards and buyer criteria operate on different axes, with different scopes and dates.

The recommended operating model preserves the raw classification, NAICS role and vintage, applicable threshold unit, source, observation date, and review status. It then evaluates mandate size from separately supported financial evidence. Where the revenue, EBITDA, entity boundary, or program context remains unresolved, the result stays pending. Research completeness should remain visible alongside apparent fit.

The reference table shows why the threshold unit cannot be discarded. The buyer comparison shows why the mandate cannot be assumed. The worksheet connects the records without merging their meanings. The hypothetical examples demonstrate the resulting decisions: retain useful program evidence, collect the missing acquisition metrics, and escalate affiliation or transaction questions to the appropriate reviewer.

For origination leaders, the immediate implementation is a data-policy change: define what each size field means, specify which evidence can populate it, and preserve its history. For analysts, the discipline is to attach every observation to its entity, period, unit, and decision. That produces a screen the deal team can inspect and update without treating an official program classification as an investment conclusion.