Directions Brain

Geography Alternatives — Europe and LatAm

12 min read

Geography Alternatives — Europe and LatAm

Definition

The AI-brokerage direction has been US-only since inception, and that choice was never argued — it was inherited from alon-huri’s framing and from the fact that harper and coverwatch are US companies. This page is the first deliberate test of it (web-researched 2026-08-10): does the smb-insurance-portfolio-brokerage play work better in Europe or Latin America?

Verdict: no. Keep the US. If a second market is ever needed, it is the UK — not “Europe,” and not LatAm.

Key points

  • The gap this page closes. Before 2026-08-10 the entire repo contained zero references to Latin America as a market, and exactly one strategic remark about Europe: in 2026-08-03-directions-shai-slobodov-lemonade-carrier-interview, shai-slobodov said “Europe has unified EU regulation — a legitimately easier arena.” Nobody followed up. Every axiom, the TAM model, the segment sprint and the competitive map are US-only and were never justified against an alternative.
  • Correction to that in-repo claim. Shai’s “easier arena” reading does not survive research. US 50-state licensing is faster and cheaper than a single EU authorisation: ~$11–14K in fees and 1–7 days per state through NIPR with 49 reciprocal states (see us-producer-licensing-mechanics), versus a 105-day average in Ireland (2025 Central Bank data), or a German-language IHK exam in Germany, plus €1,564,610 per-claim mandatory PI cover (indexed 2024), plus a home-state substance test — and then 27 separate passporting notifications. The meeting page is left as an accurate record of what he said; this is the correction.
  • Passporting is largely a fiction in practice — the single most decisive European finding. Of 685,846 registered EU intermediaries, only 7,269 hold any passport (~1%), 6,944 freedom-of-services and 322 establishment, and the count fell year-on-year. EIOPA states outright that it cannot measure whether any business actually flows, because intermediaries notify only an intention. Cross-border notifications cluster on neighbouring states. MarshBerry: “Europe is not one single MGA market. It is a set of local markets with different value propositions and routes to scale.” One licence unlocking 27 markets is an investor narrative, not an operating plan.
  • Europe socialised the two lines that make a US small-commercial account worth brokering. Workers’ compensation is a public monopoly in Germany, France, Italy, Spain, Austria and Luxembourg; health is largely statutory. US workers’ comp alone is 10% of total commercial-lines premium. A comparable European SMB therefore carries materially less privately-brokered premium than its US equivalent — which directly compresses revenue per account for a commission-based BoR model. (This is why Alan, at €785M ARR, is a health insurer rather than a broker.)
  • European SMB premium per account is thin at the bottom. ABI/Public First (Jan 2026, n=1,002 UK SME decision-makers) median annual insurance spend: sole traders £250–499; 1–9 employees £1,000–1,999; 10–49 employees £3,000–4,999; 50–249 employees £10,000–19,999. At 10–20% commission a sole trader is worth £25–100/yr. The economic band is 10–249 employees, which is far smaller than “32.3 million EU enterprises” implies.
  • The same UK survey is the best quantified evidence for our continuous-optimisation thesis found in any market. 50% of UK SMEs had not reviewed which insurance they hold in 12+ months and 47% had not reviewed adequacy, while 52% had experienced a significant business change since last reviewing (19% revenue up >20%, 17% new product lines, 16% hired 5+, 15% bought major equipment). 25–32% of accepted claims were paid only partially because limits were exceeded. 64% trust brokers most for advice; 87–94% would renew even with a 10% price increase. Usable in the pitch as the measured version of “the trigger isn’t a date, it’s the business changing” — flagged as UK data, not US.
  • If Europe, then the UK specifically. Brokers arrange 94% of all UK commercial insurance (BIBA) — the highest broker share in the developed world — one regulator, English language, and the Appointed Representative route goes live in 2–3 months with no authorisation fee (versus 6–12 months and £30–75K all-in for direct FCA authorisation). Price: commission share to the principal, and operating inside their permissions.
  • Germany is an acquisition market, not a passporting target. 47,027 registered Versicherungsmakler (Apr 2026), ~90% owner-operated, top-three commercial share under one third versus 53% in the US, ~€30B commercial non-life GWP, 20–25% commissions on private liability lines, and a 10–15-year principal retirement wave. Someone has noticed: Modern Insurance Technology launched June 2026, PE-backed, explicitly to buy German brokers and optimise them with AI. That is a capital play, not a product play — and it is the European mirror of the roll-up route AX-INS-3 forbids.
  • The German “data rail” is verified and weak — this resolves an open test on policy-ingestion-requires-a-data-rail. A comprehensive Maklervollmacht does create an enforceable insurer duty, but only as a derived secondary obligation under §241(2) BGB — not a statutory disclosure right. German trade press documents that insurers routinely refuse to supply contract data despite valid mandates, invoke §203 StGB on portfolio transfers, and that recourse is soft (written justification, the Versicherungsombudsmann, reputational pressure — no API, no deadline, no fine). The technical standard exists but BiPRO Norm 430.4 — the sub-norm for automated contract-data updates — is “selten im Markt vertreten” (rarely present). clark’s own arc confirms it: an automatic-retrieval service degraded into customers uploading their own certificates, because the insurers were the bottleneck. Germany is a middle case between Israel’s mandated rail (cover) and the US’s absence of one — better than nothing legally, useless operationally.
  • FIDA is the real European prize, and it is too far out to underwrite. The EU Financial Data Access Regulation would compel financial institutions to expose customer data via API to authorised providers, and its scope explicitly includes non-life insurance — precisely our ingestion motion, and something the US will never legislate. But it is still unadopted (Council general approach Dec 2024, trilogue from Apr 2025, listed as a pending proposal as of Feb 2026), phased over 24–48 months post-entry-into-force, with Germany among five member states pushing for delay. Realistic application for non-life: 2029–2031. Option value, not a wedge.
  • The European digital-broker graveyard is recent and specific. clark marked down from €1B to €613M on declining revenue and expanding losses; wefox rescued from insolvency (€151M, closed its Liechtenstein carrier, exited Italy, founders off the board); Luko sold for parts to Getsafe. The post-mortem consensus — “insurance distribution is a category that punishes growth-at-all-costs” — applies to us verbatim. European insurtech H1 2026: >€900M raised but only 24 deals (−29% YoY), with Alan taking over half the money. The capital market for this story in Europe is thin and specifically burnt.
  • Cycle is a headwind in both regions, slightly worse in Europe. Marsh Global Insurance Market Index Q2 2026: global composite −6%, the eighth consecutive quarterly decline; Europe −6%, UK −8%. A commission-on-premium model carries a negative organic tailwind before selling anything — and the UK, the best market, is the softest. Consistent with H-INS-9.
  • LatAm fails on substrate, not on difficulty — this is decisive. Roughly 17–18% of Mexican businesses and 26.7% of Brazilian business owners carry any business insurance. The modal LatAm SMB has zero policies, not five. The thesis therefore inverts from an optimisation/switching play into a first-sale/penetration play — a different company, with demand creation instead of visible arbitrage, higher CAC per closed account, and a smaller first-year commission. Bottom-up corroboration: GNP, Mexico’s largest domestic insurer with 15,000 agents, holds ~108,000 MiPyme policies against 5.45M economic units, of which 95.5% are micro (0–10 people). Brazil’s entire patrimonial pool (residential + condo + commercial) is R$35.7B ≈ USD 7B.
  • LatAm regulation is genuinely easier than the US — and it does not help. One national licence in all six major markets; no state patchwork, no non-resident licences, no surplus-lines stamping. Brazil’s SUSEP registration is free, with no capital requirement and no compulsory E&O. But every market legally compels a locally-resident licensed principal: Brazil requires a permanent-resident diretor técnico who signs every proposal (Circular SUSEP 127/2000); Mexico requires apoderados holding a CNSF cédula (which requires CURP, i.e. residency) and takes ~8 months for corporate authorisation; Colombia’s corredor route needs an S.A. with five shareholders, a three-member board, a revisor fiscal and indexed minimum capital (~USD 115–180K). Capital is not the constraint; the licensed local human is.
  • Two LatAm mechanics break the model outright. Brazil: commission attaches to the broker who signed the proposal (Lei 4.594/64 art. 13), so a carta de nomeação does not reliably move economics mid-term — you wait for renewal, stacking deferred monetisation on top of demand creation. Mexico: coverage ceases automatically 30 days after non-payment (Ley sobre el Contrato de Seguro art. 40), with regional commission clawback practice — attacking the ~90%-retention leg for micro-SMBs on fractioned premiums.
  • LatAm exit math is fatal even on success. Wiz Co (listed Brazilian brokerage): R1.36B market cap — 1.0x revenue, P/E 7.0, down 4.3% CAGR since 2015. Alper delisted Sept 2024 at 2.8x sales on a −7.3% net margin. Total LatAm insurtech funding was USD 199M in 2025 and USD 90M in H1 2026 — less than harper’s single round — with capital rotating out of distribution (48% of the ecosystem and falling) into enablement.
  • The informed money in LatAm chose the adjacent thesis four independent times: a16z/Kaszek → Segura (arm the brokers, R2M in seven years and was absorbed. And Betterfly, the region’s best-capitalised insurtech, exited Argentina, Brazil, Colombia, Ecuador and Peru in March 2025 to chase the US, stating the US healthcare market alone is 30x those five combined.
  • What this changes about the US choice: nothing — but it converts a default into a decision. The honest pitch line is that we pressure-tested both alternatives and the US won on substrate (policies exist to optimise), premium density per account, capital availability, and exit comparables — while conceding that US regulation is the messiest of the three and that the UK holds the single best evidence base for the continuous-optimisation claim.

Evidence

  • 2026-08-10-directions-smb-survey-verticals-and-fida — the team consumed this research in-session the same morning: LatAm struck off on the 17%-insured substrate (“למחוק לט”ם”), the UK kept as the only real second-market option (~2-month license), the 52% business-change stat adopted into the pitch narrative, and Guy proposing a contrarian FIDA pre-positioning play (two local brokers per EU country from day one so the 2029 API arrives onto ready infrastructure).
  • 2026-08-03-directions-shai-slobodov-lemonade-carrier-interview — the only prior in-repo remark on Europe (“unified EU regulation — a legitimately easier arena”), corrected above.
  • us-producer-licensing-mechanics — the US baseline this compares against.
  • policy-ingestion-requires-a-data-rail — the hypothesis whose named Clark test this research resolves.
  • Web-verified 2026-08-10, Europe: EIOPA (IDD Arts. 3/4/6/10 rulebook; 2nd IDD application report incl. the 7,269-of-685,846 passporting figures and country-by-country annex; 3rd IDD report press release, Mar 2026); Commission Delegated Regulation (EU) 2024/896 (PI indexation); Central Bank of Ireland authorisation statistics 2025; IHK §34d GewO guidance; AFM Wft process pages; Estonian Finantsinspektsioon; FCA/RQC and AR-network sources (authorisation timelines); BIBA evidence to Parliament (94% of UK commercial); ABI/Public First “Small Business, Big Risk” (Jan 2026, n=1,002); DIHK Versicherungsvermittler register (Apr 2026); MarshBerry (European MGA market; broker-share-by-country); EY (continental broker consolidation); Strategy& (German commercial brokerage); OSHwiki (workers-comp public monopolies); Insurance Journal (US workers-comp share of commercial premium); iww.de (Maklervollmacht §241(2) BGB duty and insurer refusal practice; portfolio-transfer obstruction); ConVista (BiPRO 430.4 adoption); Clark mandate documentation and independent German app review; Startuprad/Coverager (Clark €613M markdown); Insurance Journal (wefox recapitalisation); TechCrunch (Getsafe/Luko; Alan); Freshfields and financial-data-access.com (FIDA status and phasing); Marsh Global Insurance Market Index Q2 2026.
  • Web-verified 2026-08-10, LatAm: MAPFRE Economics (LatAm 2024 premiums, penetration, protection gap; insurtech ecosystem 2025–H1 2026); Swiss Re Institute (2026 outlook); AMIS (Mexico >MXN 1trn, 2.94% penetration; SMB insurance share; fire-claims severity); INEGI (5,451,113 economic units; 95.5% micro); GNP (108K MiPyme policies); CNseg/FenSeg and IRB (Brazil R144.5B; patrimonial R$35.7B; empresarial +14.0%); CNseg + Maena PME study (26.7% — small sample, n=86 entrepreneurs, flagged); Sincor-SP (~25% buy fire cover); Sebrae/Receita Federal (~24M active small-business CNPJs); SUSEP Painel de Corretores via CQCS (165,225 registrations — conflicting count of ~149K reported elsewhere, unreconciled); Lei 4.594/64 and Circular SUSEP 127/2000; Ley sobre el Contrato de Seguro art. 40; CNSF/Ormuz/ELAAA (Mexican authorisation process, ~8 months); Fasecolda and UNDP Colombia (premiums, penetration, 4.9% broker access-point share); StockAnalysis (Wiz Co) and SimplyWall.St (Alper); LatamList (Betterfly exits); NeoFeed (Segura/a16z); The Shift and SUSEP (Open Insurance impact slipped to 2028, brokers excluded from SPOC).
  • Flagged soft: the Mexican 17–18% figure is syndicated PR sourced to AMIS with no retrievable primary publication; the Brazilian 26.7% comes from an n=86 survey; no LatAm regulator publishes premium by insured company size, so SMB commercial pools are genuinely unavailable rather than estimated; Colombia has no published SMB insurance-penetration statistic at all.

Open questions

  • Does the UK deserve a proper scored assessment as a second market (AR route, 94% broker share, the ABI evidence base), or is any second-market work a distraction before US kill-gates are run?
  • Can the ABI review-gap statistics be used in a US pitch, or does quoting UK data invite the “you have no US evidence” counter? Is there a US equivalent survey nobody has found yet?
  • Does FIDA’s non-life scope survive trilogue — and if it does, does it change the long-run structural attractiveness of Europe enough to matter to a company founded in 2026?
  • Germany’s 47,027 owner-operated brokers and retirement wave are the strongest fragmentation story anywhere. Does that make Germany interesting despite AX-INS-3, if the acquisition is a pure book purchase?
  • Is there a third geography nobody has looked at (Canada, Australia, Nordics) where broker share is high, workers’ comp is private, and no AI-brokerage wave has started?

Graph View

Backlinks