SizeShift

1. One-liner

Shows which of your set-aside recompetes just gained bigger rivals, and drafts the SBA comment that argues your code down.

2. Trend signal — why now?

On 20 August 2026 the SBA published a proposed rule (RIN 3245-AI67, 91 FR Issue 160) that is the largest redefinition of “small business” in a generation. The numbers are not incremental:

• 978 six-digit size standards collapse into 338 four- and five-digit standards. All 18 subindustry exceptions — including ITVAR — are eliminated.
• 114,541 firms become newly “small.” Of those, 37,002 already hold federal contracts worth roughly $71 billion and would regain set-aside eligibility.
• Thresholds rise tenfold or more in professional services, IT, engineering, logistics and hospitality. NAICS 541511 (computer programming) goes $34M → $531M. Engineering (541330) $25.5M → $252M. Management consulting (5416) $19–29M → $295M. Hotels (721110) $40M → $503M. Nine NAICS codes see increases exceeding $450M.
• 64 industries convert from receipts-based to employee-based measurement; six more move from mixed to purely employee-based. Commercial construction (236220) goes from $45M receipts to a 600-employee headcount test — a completely different arithmetic.
• SBA proposes not to reduce any standard, even in the 45 industries where its own analysis supported a decrease.
• Comments are due 21 September 2026 — a 30-day window on a rule that rewrites the competitive field for a $117B+ set-aside market.

The asymmetry is the whole story. A $12M IT services firm that has spent six years building past performance inside 541511 set-asides will, on the day the final rule lands, be bidding against firms doing $400M. Nothing about that firm changed. The fence moved.

Federal contract spending was ~$793B in FY2025, with agencies awarding nearly 28% to small firms. FY2024 set-aside-restricted competitions alone exceeded $117B. This is a large, well-documented pool of money whose eligibility rules are being rewritten on a 30-day clock.

Counsel across the govcon bar is telling clients the same three things: remap your NAICS codes, model your status under both measures with affiliate-inclusive data, and do it now rather than late in the comment window. Every one of those is analytical work that no product currently does.

Provenance:

• Signal 1 (demand): SBA proposed rule consolidating 978 size standards into 338, making 114,541 firms newly small — including 37,002 holding ~$71B in current federal contracts. Comments due 21 September 2026. — https://www.federalregister.gov/documents/2026/08/20/2026-17042/small-business-size-standards and https://www.govinfo.gov/content/pkg/FR-2026-08-20/html/2026-17042.htm — 2026-08-25
• Signal 2 (feasibility): The full proposed size-standard table, the current 13 CFR 121.201 table, USAspending award history and SAM.gov registrant data are all public and structured. Mapping a firm’s codes across both tables and joining to its live recompete pipeline is a data-join problem, not a research problem. — https://www.pillsburylaw.com/en/news-and-insights/sba-proposal-small-business-size-standards.html — 2026-08-25
• Signal 3 (economic): Federal contract spending reached ~$793B in FY2025 with ~28% to small businesses; FY2024 set-aside competitions exceeded $117B. GovCon advisory labour is priced at $200–350/hr with GSA/procurement engagements running $20K–60K — an established willingness-to-pay far above a SaaS subscription. — https://cyberxgovsolutions.com/small-business-contracting-hit-a-goal-but-the-dollars-still-shrank/ and https://www.gsascheduleservices.com/blog/top-procurement-consulting-firms/ — 2026-08-25 Category: Regulatory arbitrage (a scope expansion that redraws a competitive field rather than creating a filing duty) + Underserved niche (govcon tooling sells opportunity discovery and proposal writing; nobody sells the defensive analysis of who you will be bidding against)

3. The opportunity

Every govcon software product on the market is built around finding opportunities. Sweetspot, Bidara ($299–599/mo), SamSearch, GovConToday, GovSpend ($12K–42K/yr) — they all answer “what should I bid on?” They index SAM.gov, filter by NAICS, track recompete windows, and increasingly draft proposals.

Not one of them answers the question that matters this quarter: “who else can bid on my work now, and what do I do about it?”

That gap exists because the incumbents’ data model treats the size standard as a static filter — a lookup value used to decide whether an opportunity is visible to you. It is a column, not a variable. When the column changes for 978 codes at once, and when 64 industries change measurement basis entirely, a lookup table is the wrong shape of software.

The alternative today is a lawyer. Govcon counsel bills $200–350/hr to do exactly this: map your codes, recompute your five-year average receipts under 13 CFR 121.104 with affiliate rollups, tell you whether you survive, and draft your comment letter. For a firm doing $8M in revenue, a $15K legal analysis to understand a proposed rule is a hard sell — so most of them will do nothing, find out when the final rule lands, and lose a recompete to a firm four times their size.

The wedge is that this analysis is genuinely mechanical. The proposed table is public. The current table is public. Your award history is public in USAspending. Your recompete windows are public in SAM.gov. The only private inputs are your revenue by year and your affiliate structure — two numbers and a short list that the customer types in once.

There is a second, better-paying half. The same engine that tells you that you are exposed also tells you exactly what to say about it. The rule is open for comment until 21 September, and counsel is publicly advising that effective comments cite “specific NAICS codes and dollar or employee figures” rather than arguing methodology. That is a templated, evidence-backed document — the firm’s own five-year receipts, its own award history, the specific threshold it objects to, and the specific alternate figure it proposes. A comment letter is the one action a $10M contractor can actually take between now and the final rule, and it is precisely the kind of document an LLM assembles well when handed structured facts.

4. Target market

Primary customer: Owner, VP of Business Development, or Capture Manager at a US federal contractor doing $3M–$40M in annual revenue, deriving the majority of revenue from small business set-asides, concentrated in NAICS 541xxx (IT, engineering, management consulting), 236xxx/237xxx (construction), or 4841xx/4931xx (logistics and warehousing). Typically 15–200 employees. Usually one or two people wear the entire capture-and-compliance hat.

Why they buy: Their whole business model is a fence. They price, staff and pursue work on the assumption that firms above $34M cannot bid against them. Industry counsel is now stating plainly that “businesses that qualified as small businesses under the current standard will be competing against much larger companies for small business set-asides,” and that “growing small businesses closer to the size standard are therefore likely to face the greatest competition from the newly eligible firms.” GovCon attorney Josh Duvall framed the tradeoff publicly as “more runway for companies to remain ‘small’ and compete longer” against “significantly stronger competition for set-aside opportunities.” The firm’s owner has read a version of this on LinkedIn four times in the last week and does not know what it means for their codes, their pipeline, or their next recompete.

Rough TAM reasoning: Roughly 28% of ~$793B in FY2025 federal contract dollars went to small businesses; set-aside-restricted competitions alone exceeded $117B in FY2024. The population of active small-business federal contractors with meaningful annual award volume is in the low tens of thousands. Target the band that is both exposed and able to pay: firms with ≥$1M in annual federal awards in an affected NAICS — a serviceable population realistically in the 15,000–35,000 range. At $299/mo, capturing 1.5% of a 25,000-firm serviceable base is ~$1.3M ARR.

Why now for them: The comment deadline is 21 September 2026 — under a month away as of writing. After that the question shifts from “can I influence this?” to “how do I survive it?”, and the product’s value shifts from advocacy to competitive intelligence. Both are real, but the first one has a hard date on it, and hard dates are what convert.

5. Product sketch (MVP)

• Code remap. Enter your NAICS codes; get the proposed four/five-digit grouping each one collapses into, the current threshold, the proposed threshold, and whether the measurement basis flips from receipts to employees.
• Status model. Enter five years of receipts (or headcount) plus any affiliates; get your computed size under 13 CFR 121.104 against both the current and proposed standard, per code, with the arithmetic shown.
• Exposure map. Pull your award history from USAspending and your tracked recompetes; for each one, show how many additional firms become eligible to bid under the proposed threshold — named, with their federal award volume.
• Head-to-head list. For your top codes, the specific firms that cross from large to small: name, revenue band, agencies they already serve, contracts they already hold.
• Comment drafter. Generates a submission-ready SBA comment letter citing your specific NAICS code, your own five-year figures, the objected-to threshold, and a proposed alternate figure with supporting reasoning.
• Threshold watch. Alerts when the final rule publishes, when your codes’ numbers change from the proposal, and when an effective date is set.
• Recompete risk score. For each upcoming recompete in your pipeline, a simple exposed / partly exposed / unaffected verdict with the driver named.
• Board-ready one-pager. Export the whole picture as a PDF the owner can hand to a partner, a lender, or an acquirer.

6. AI angle — what’s load-bearing

Two places, both real.

The remap is genuinely ambiguous work. 978 codes collapsing into 338 is not a clean parent-child rollup — subindustry exceptions are deleted, and 64 industries switch measurement basis. Determining which proposed grouping actually governs a given firm’s work requires reading the firm’s own contract descriptions and PSC codes against the proposed groupings’ scope language. That is document interpretation over unstructured text at a scale nobody wants to do by hand, and it is the thing counsel bills $300/hr for. A deterministic lookup gets you maybe 70% of the way; the remaining 30% — the ambiguous, exception-deleted, basis-flipped codes — is exactly where the customer is at risk and exactly where language models earn their keep.

The comment letter is a generation task with a factual spine. The inputs are structured (your receipts, your codes, the proposed threshold, your award history). The output is a persuasive regulatory document in a specific register, citing specific figures, making a specific alternate proposal. Strip the AI out and you have a fill-in-the-blank template that reads like a fill-in-the-blank template — and SBA staff reading 800 comments will treat it as one.

If you removed the AI you would be left with a threshold lookup table, which is a blog post, and several people have already published it for free. The defensible product is the interpretation layer plus the generated advocacy.

7. Localization angle (if any)

N/A — this is a US-only play. The entire product is defined by one US federal regulatory instrument (13 CFR 121) and one US procurement data estate (SAM.gov, USAspending, FPDS). There is no meaningful localization axis; the analogous EU and India procurement set-aside regimes are structured differently enough that they are separate products, not translations.

8. Business model — path to $1M–$5M ARR

• Pricing: Three tiers.
  • Comment ($149 one-time) — remap, status model, and a drafted comment letter. Deliberately priced as an impulse purchase before 21 September. This is the wedge, not the business.
  • Watch ($299/mo) — the exposure map, head-to-head competitor lists, recompete risk scoring, and final-rule alerts. The core product.
  • Capture ($699/mo) — multi-entity/affiliate modelling, unlimited codes, quarterly refreshed competitive briefs, board one-pager exports. For firms with a real capture function or a private-equity owner.
• ACV: Blended ~$4,200/yr assuming a mix skewed toward Watch.
• Rough math to $1M ARR: 240 customers × $299/mo × 12 = $861K, plus ~25 Capture seats × $699/mo × 12 = $210K → ~$1.07M. Roughly 265 paying firms out of a serviceable base of 15,000–35,000. That is a ~1% penetration requirement.
• Rough math to $5M ARR: ~1,000 Watch customers plus ~150 Capture customers. Requires the product to survive the transition from “comment on the proposed rule” to “live competitive intelligence for set-aside bidding” — i.e. the exposure map must be valuable in 2028 when the rule is old news. Plausible, because the underlying question (“who else can bid on this?”) is permanent and the size standards get revisited on a rolling five-year cycle regardless.
• Expansion path: More codes, more entities, more tracked recompetes. Natural upsells into size-protest support (when you believe a competitor is misrepresenting size, the evidence assembly is the same engine pointed the other way) and into recertification monitoring at option exercise under FAR 52.219-28.

9. Go-to-market wedge — first 100 customers

• Mine USAspending for the exposed, by name. Every firm holding set-aside awards in the top-20 affected NAICS codes is a public record, with award dollars and a registered point of contact. Filter to $1M–$40M in annual awards in codes where the threshold rises ≥5×. That is a target list of several thousand firms, each of which can be sent a pre-computed, firm-specific exposure summary: “Under the proposed rule, 1,343 additional firms become eligible to bid on 541511 set-asides. Here are the 40 of them that already hold contracts at your top agency.” Personalised, factual, public-data, and useless to fabricate. Expect a far better reply rate than generic cold email because the email is the product demo.
• Ride the comment deadline. SBA held virtual forums on the proposal and the docket is public. Publish a free “what your code becomes” lookup — the deterministic 70% — as an ungated tool, and convert on the drafted comment letter. Every law firm in the govcon bar has published a client alert in the last week driving the same audience to the same question with no tool at the end of it. Be the tool at the end of it.
• The govcon commentariat is small and reachable. Federal News Network, GovConWire, OrangeSlices AI, GovCon Giants, and a handful of high-follower LinkedIn attorneys (Duvall, Crusius and peers) are actively publishing on this rule right now. A genuinely useful free calculator with real numbers is the kind of artifact this specific community shares aggressively, because their audience is asking them for exactly it.
• PTAC / APEX Accelerators. Federally funded procurement assistance centers advise small contractors for free in every state and are being asked this question by their clients this month. They cannot build software. Offer them a co-branded lookup for their clients; they become a distribution channel with institutional trust.
• Size-protest and recompete triggers. Firms that just lost a set-aside recompete are the most motivated buyers alive. Award notices are public; the losers are inferable. Time outreach to the two weeks after award.

10. Build complexity — justification

Medium. The data is all public and mostly structured: the current 13 CFR 121.201 table, the proposed table in the Federal Register, USAspending award history, SAM.gov registrant data. Ingesting and joining these is standard ETL. The genuinely custom work is the remap interpretation layer (ambiguous code collapses, deleted exceptions, receipts-to-employee basis flips) and the affiliate-inclusive receipts calculation under 13 CFR 121.104, which has real edge cases and must be right because customers will check it against their accountant. Comment generation is an off-the-shelf LLM task over structured inputs. Realistically 8–12 weeks to a paid v1 for a pair, with the free lookup tool shippable in 2–3 weeks — which matters enormously given the 21 September deadline.

11. Gating checklist

Gate	Pass?	Note
Legal in target market	✅	Analysis of public regulatory data and assistance drafting public comments. Not legal advice; disclaim clearly and avoid rendering size-status certifications.
Ethical — no harm / dark patterns	✅	Helps small firms understand and participate in a rulemaking that affects them. Comment letters are the intended democratic mechanism.
Market exists (evidence above)	✅	$117B+ in annual set-aside competitions; established $200–350/hr advisory spend on exactly this analysis.
1–5 person team can build this	✅	Two people, 8–12 weeks. Public data, off-the-shelf LLM.
Launchable with <$50K / ₹40L	✅	Data is free. Main costs are inference and two months of runway.

12. Feasibility score

Axis	Weight	Score	Notes
Problem intensity	20	16/20	Real money, real fence-moving, and a hard 21 Sep deadline. Docked because the proposed rule is not yet law — the pain is anticipated rather than currently bleeding, and a firm can rationally choose to wait.
Demand evidence	15	12/15	Strong indirect evidence: heavy legal-alert volume, named attorney commentary on competitive squeeze, established $200–350/hr spend on this analysis. Docked because I found no verbatim contractor complaints — the rule is five days old and the community reaction is still mediated through law firm alerts.
Build feasibility	15	12/15	Public structured data, standard ETL, off-the-shelf LLM. Docked for the affiliate-receipts calculation edge cases and remap ambiguity, which must be accurate to be trusted.
Distribution clarity	15	13/15	Named, public, filterable target list from USAspending with firm-specific pre-computed outreach. The email is the demo. Docked slightly because govcon cold email is a crowded inbox.
Revenue mechanics	15	11/15	Pricing sits far below the $200–350/hr alternative and ~1% penetration reaches $1M. Docked because the $149 wedge is one-time and the $299/mo retention thesis depends on the exposure map staying useful after the rule finalises.
Time to first revenue	10	8/10	The $149 comment tier can convert within weeks of the free lookup shipping, well before 21 September. Docked because recurring revenue takes longer to establish than the one-time spike.
Defensibility	10	4/10	Execution-only moat. The remap logic is copyable once published and the data is public. Real defence is speed, the accumulating recompete/competitor graph, and becoming the referenced tool in the govcon commentariat.
Total	100	76/100

13. Qualitative modifiers

Founder-fit tags

technical-heavy · domain-expertise-required

You need someone who can do the ETL and someone who genuinely understands 13 CFR 121 affiliation rules. The second is not optional — get the receipts calculation wrong once, publicly, and the product is dead in a community this small and this chatty.

Key assumptions to validate (3–5)

1. Assumption: Small contractors will pay for analysis of a proposed rule rather than waiting for the final one. How to test: Ship the free lookup, gate the drafted comment letter at $149, and measure conversion before 21 September. If under 2% of lookup users pay, the advocacy wedge is dead and only the post-final-rule competitive intelligence play survives.
2. Assumption: The exposure map (who can now bid against me) retains value after the rule finalises, sustaining $299/mo. How to test: Interview 20 capture managers on whether they currently pay for, or would pay for, named competitor-eligibility intelligence independent of this rule.
3. Assumption: The NAICS remap can be made accurate enough to trust. How to test: Compute remaps for 50 firms across the messiest codes (deleted ITVAR exception, receipts-to-employee flips) and have a govcon attorney grade them. Target ≥90% agreement.
4. Assumption: The USAspending-driven personalised cold email materially outperforms generic outreach. How to test: Send 300 personalised (with pre-computed firm-specific numbers) against 300 generic. Compare reply rates.

Risk flags

1. Regulatory risk — the rule may not finalise as proposed. This is the big one. It is a proposed rule with a live comment period, significant industry opposition, and no announced effective date. It could be substantially softened, delayed, or withdrawn. The comment-letter wedge is somewhat insulated (the advocacy is valuable precisely because the outcome is undetermined) but the $299/mo exposure map loses much of its urgency if the rule stalls. Build the recurring tier so it works on the current size standards too.
2. Timing risk — a very short window. Comments close 21 September 2026. If the free tool is not live by mid-September, the highest-conversion moment passes and the product launches into a lull between the comment close and the final rule.
3. Platform dependency — public data sources. SAM.gov and USAspending API availability, schema changes, and rate limits are outside your control, and federal data infrastructure has been unreliable at points.
4. Incumbent response. Sweetspot, SamSearch and Bidara already hold the audience and the SAM.gov pipeline data. Any of them could add a size-standard module in a quarter. The defence is being first and being the one the attorneys link to.
5. Advice-boundary risk. Telling a firm it is or is not small edges toward legal advice and toward size self-certification, which carries False Claims Act exposure for the contractor. Frame outputs as analysis, never as certification, and say so loudly.

14. Structured verdict

Score:                  76/100
Verdict:                GO
Confidence:             Medium
Best-fit builder:       Technical founder who can do public-data ETL, paired with
                        a govcon domain advisor fluent in 13 CFR 121 affiliation rules
Time to revenue:        3–6 weeks (free lookup → $149 comment tier before 21 Sep)
Capital to launch:      $8–15K
Top 3 assumptions to validate first:
  1. Contractors pay for proposed-rule analysis — gate comment letter at $149,
     measure conversion off the free lookup before 21 September
  2. Exposure map sustains $299/mo after finalisation — 20 capture-manager interviews
  3. Remap accuracy ≥90% vs attorney grading on 50 hard-case firms
Kill criteria:
  - Abandon if <2% of free-lookup users convert to the $149 comment tier before 21 Sep
  - Abandon if attorney grading of the remap comes in below 80% agreement
  - Abandon if the proposed rule is formally withdrawn AND fewer than 30 firms have
    paid for the exposure map on current-standard data alone
  - Abandon if an incumbent govcon platform ships an equivalent size-shift module
    before your paid v1

15. Next step — 1-week validation sprint

• Day 1–2: Build the deterministic remap for the 25 highest-volume set-aside NAICS codes — current threshold, proposed grouping, proposed threshold, basis flip. Ship it as an ungated public lookup with an email capture on the “draft my comment letter” button. No backend, no accounts.
• Day 3–4: Pull USAspending for firms with $1M–$40M in FY2025 set-aside awards in those 25 codes. Send 300 personalised emails containing that firm’s own pre-computed exposure line — the number of newly-eligible competitors in their primary code and three named firms that already work their top agency. Link to the lookup.
• Day 5: Decide on two hard numbers. Go if ≥8% of emails produce a click into the lookup and ≥25 people hit the gated “draft my comment letter” button. No-go if the click rate is under 3% or fewer than 10 people request the letter — that means the pain is real in the trade press but not yet real in the owner’s head, and the 21 September window is too short to educate the market into existence.

The falsifiable result is the gated-button count, not enthusiasm. People forward alarming LinkedIn posts about regulatory change constantly and buy nothing. Clicking “draft my comment letter” is the smallest action that proves someone intends to actually do something about it.