Directions Brain

Directions — Shai Slobodov (Lemonade) carrier-economics interview

4 min read

Directions — Shai Slobodov (Lemonade) carrier-economics interview

Source: raw/meetings/notetaker/2026-08-03-meeting-with-shai-14-45.json

Summary

Second expert call of the 2026-08-03 insurance offsite. shai-slobodov (5.5 years at lemonade, currently leading three platform teams focused on claims) gave the team a primer on how US insurance actually works: carrier vs. ecosystem structure, per-state political regulation, carrier unit economics (~8–10% margin vs. ~90% SaaS), pricing filings, broker-channel dynamics, and retention. His core message: founding a full-stack carrier as first-time founders is “borderline a suicide mission”; realistic entries are a light single product (pet/travel), the carrier ecosystem (full of 1990s-era duopoly vendors), or a genuinely AI-native broker play. The team agreed to a longer follow-up session.

Key takeaways

  • Market structure is carriers + a huge ecosystem of entrenched vendors. The US market is 150+ years old; very specific functions are controlled by duopolies/monopolies built in the 1990s (e.g., two companies own all auto-insurer↔repair-shop communication; similar systems for hospital communication). Hard to displace, but ripe targets — Shai’s preferred direction for the team: big prize without carrier regulation.
  • Carrier vs. front (the Robinhood debate). Against Saar’s “Lemonade missed being the Robinhood of insurance” thesis, Shai argued: “If you want to be a real insurance company in the US, you cannot not be a carrier” — a front-only company has a high ceiling but not an insurance-company ceiling. End-to-end carriers control both acquisition spend (top) and claims-automation savings (bottom). Lemonade is end-to-end carrier in >90% of its business.
  • Regulation is per-state, political, and ongoing. US insurance regulators are elected and skew pro-consumer. Lemonade’s first product (NY renters) took ~1.5 years to approve with near-zero revenue in the meantime; fully-built state launches still sit awaiting sign-off. California is “terrible” for home (wildfires + pro-consumer regulator; carriers exit); Texas among the best. Europe has unified EU regulation — a legitimately easier arena. “You need an American figure — this is not for Israelis” (not doable purely from Israel).
  • Carrier economics are structurally thin. Regulators enforce a loss-ratio band (pay out too little → forced price cuts; too much → financial collapse). GEICO model: ~8–10% margin on premium, invested Buffett-style — vs. ~90% margins at real SaaS (monday, Slack). Being a carrier requires enormous capital; Lemonade raised most of its money as a public company. Insurers pay money out at scale — a non-trivial capability (even Stripe is weak at payouts), flagged as an ecosystem opportunity.
  • Pricing filings are public; the moat is elsewhere. Telematics/AI pricing parameters must be explained to regulator and customer; people tweet Lemonade’s state launches from public filings before announcements. New carriers do “me-too filings” copying an incumbent’s model (usually Progressive). The real secret sauce is telemetry collection, data accessibility, and automation — none of it in the filing.
  • You die with bad customers. Non-renewal sits under underwriting and needs a state-approved reason (a filed claim isn’t one); rates are filed formulas; you can’t refuse claims. The only real lever is exiting a whole state and running off the book. Root Insurance ran the opposite strategy — deliberately courting high-risk drivers (“we want the garbage, we’ll price it”) — nearly died, then recovered.
  • Retention mechanics. Renters churn is high but predictable (life events); bundled car+home customers retain significantly better; pet has inherently high engagement, most lines want silence + fast claims. US policy terms are semi-annual/annual only — the team’s continuous “one policy” idea would be a hard, novel regulatory pitch.
  • Broker channel is too big to ignore. Lemonade was born militantly B2C (“Bituach Yashir” style) and later embraced agents/brokers/affiliates. Broker compensation is a commission on premium for the entire life of the policy. A broker owns the customer experience — reputational risk cuts both ways. “If you can crack something around AI and brokers — taking less money and keeping the relationship — there’s definitely a market.”

Decisions

  • None formally — but the call materially reframed the entry question: full-stack carrier is effectively off the table for a first venture; ecosystem, light-line carrier, or AI-broker are the live options.

Action items

  • saar-arbel — schedule a longer follow-up session with shai-slobodov (more than two days’ notice; Shai may join in person).

Open questions

  • Are both rating parameters and weights public in filings? (Shai unsure.)
  • Typical broker commission percentages by channel? (Findable in public disclosures.)
  • How is specialty/surplus-lines (yachts, art) regulation structured, and does it hold opportunities?
  • Is a multi-year / perpetual “one policy” actually fileable and financially viable?
  • Do the team’s differentiation points survive these regulatory constraints?

Graph View

Backlinks