Directions Brain

Directions — Tom Granit Ramu (ex-Lemonade) car-insurance operator interview

4 min read

Directions — Tom Granit Ramu (ex-Lemonade) car-insurance operator interview

Source: raw/meetings/legacy-transcripts/Nizan _ Tom (2026-08-03 15.32).txt

Summary

First expert call of the 2026-08-03 insurance offsite at Guy’s house. nizan-shifman and saar-arbel interviewed tom-granit-ramu, who ran car-insurance product at lemonade, for a practitioner’s view of how a US carrier actually operates. The call covered the regulatory weight of car insurance vs. light lines (renters/pet), the dual product org (UX product vs. “insurance product”/actuaries), why pricing changes take months and books must stay balanced, Lemonade’s telematics edge (CMT phone data, the Metromile acquisition, the Tesla partnership), and Lemonade’s late-but-total embrace of brokers and bundling.

Key takeaways

  • Lines differ radically in regulatory drag. Car insurance is the biggest TAM (mandatory for everyone) but the hardest regulated line — every price must be decomposed into justified risk factors, filed per state, and audited; renters/pet are “back door” lines where you can move fast. Lemonade skipped New York for car entirely (most customer-favorable state).
  • A price change is an event planned ~a year ahead. Filings for the year were essentially ready each December; regulator approval takes months. A/B tests are near-impossible in car insurance because you cannot treat populations differently without filing.
  • You carry two pricing models at once. The existing book can’t be migrated to a new model easily (contractual + legal limits, non-renewal windows), so old and new models run in parallel, both auditable. The book must stay balanced — good customers finance bad ones — which is also why incumbents (Geico, Allstate) can’t move their books to telematics while Lemonade started its whole car book on it.
  • Telematics was Lemonade’s alpha. Phone-based driving data via CMT, an acquired device company (Metromile, West Coast, OBD devices), and — the strongest move — the Tesla partnership: historical driving data available at signup, so underwriting quality is known instantly. Tom’s own feature: grant the ~5% telematics discount upfront from self-reported driving to lift conversion, then reprice on actual behavior; users under-claim their own driving quality.
  • Car insurance is a pure price market. “המחיר הוא מה שקובע” — price decides, nothing else.
  • The dual product org is a structural tension. A parallel “insurance product” team (ex-carrier veterans, actuaries, filing specialists, analysts) packages the product; they and the UX product team constantly enter each other’s lanes, mostly in car insurance. Tom’s view: the two roles should be one person.
  • Distribution reality: people buy insurance through brokers. Lemonade was militantly B2C for years, then went all-in on a broker-facing system after two years of car insurance. Bundling (“to bundle”) is the most familiar motion in US insurance — about half of churned users left because they bought a house and got a bundle offer. Cross-sell along life events is a must-have, but you can’t start multi-product.
  • AI-first hurts on hard claims. Car-accident customers want a human; renters/pet claims suit instant automation. Lemonade later opened proper help centers.
  • Where she’d build: broker price-comparison sites without real infrastructure are “bullshit”; Lemonade wins because it owns carrier “irons.” The interesting plays are genuinely tailored personal packages with lifecycle proactivity, or new/exotic lines (crypto — built at Lemonade but never shipped, clinical trials, ships, workplace insurance) rather than the old lines.

Decisions

  • None — discovery interview.

Action items

  • tom-granit-ramu — intro the team to more recently-departed Lemonade people and US “insurance folks” (will coordinate with nizan-shifman).

Open questions

  • How does the balanced-book / two-model constraint translate to commercial lines and to a broker (rather than carrier) position?
  • Does the “price is everything” rule of car insurance hold for SMB commercial, or does service/coverage-fit matter more there?
  • Which exotic/new lines (crypto, trials, workplace) have real demand vs. were Lemonade experiments?

Graph View

Backlinks