Servicing is the wedge.
The renewal is the prize.
Large multi-national brokers are carrying small-commercial E&S books they cannot service profitably and cannot release. An MGA that stands up a licensed broker entity can take the servicing, then the placement, then the book itself — and there are four ways to do it, worth very different amounts.
The book loses $2.79m a year, and they cannot release it
Four ways to take the book — and four different businesses
Rung 2 — Servicing plus placement
Every rung, at every appetite fit rate
Three locks, then the prize
A broker who has released licensed staff, let their own filing calendar lapse and stopped answering the phone on this segment cannot recover that capability inside one renewal cycle. That is not a threat to make in the room — it is why the arrangement is an annuity rather than a contract, and why the servicing conversation is worth having even when placement is refused.
A licensed broker entity, not a service company
E&O does not rise gradually as you climb. It steps, once.
Answer the rung with evidence, in ten weeks
Name the ceding broker and secure permission to diagnose a real book
One broker, one ring-fenced segment by state, class or premium band. Instruct the licensing opinion in parallel — do not wait for the diagnostic.
Verify the fit rate as a distribution, never as an average
Report fit by class, state and premium band. A 78% average conceals classes at 40%. Calculate the break-even fit rate and show the case holds at 55%.
Settle the three-party liability allocation and the licensed-act register
A written allocation with matching insurance, or a pilot-period interim with a cap and a review date. The register enforced at the desktop first.
Enter at Rung 2 with a contractually pre-agreed escalator to Rung 3
The escalator fixes the triggers, the override, the consent protocol, eligibility, the liability handover and the reversal path — now, not later.
Build the segmentation gate even though the pilot is single-model
The segment rule, the authority flag and the account-level model marker cost little in phase one. Every realistic end state is hybrid.
Exercise the escalator on accounts that have performed
Transfer state by state as licences and appointments complete, sequenced by licensing readiness rather than book logic.
Four things to settle, and one recommendation
Blueprint all four rungs. Enter at Rung 2. Pre-agree the escalator to Rung 3. Build the segmentation gate anyway.
Rung 2 earns placement economics without asking the broker to surrender client ownership on day one. Rung 3 holds the value — but its year-one margin is lower than Rung 2's at almost any fit rate, so the case for it is the owned book, not the income statement.
- Which broker, and on what segment. Diagnostic permission on one ring-fenced book is the cheapest ask on the list, and it makes everything after it real.
- Whether the entity is being funded now. Producer licensing runs 9–14 months. If Rung 3 is a possible destination, the licensing programme starts before the pilot.
- Who carries the three-party liability, in writing. An agent employed by the operator, under the broker's brand, on the MGA's paper. Until this is allocated with matching insurance, every rung above 1 is unpriced.
- Whether the fit rate is measured or assumed. Both pivots land at roughly the same fit rate. One number determines the rung, and it is usually quoted from an MGA's own submission flow rather than measured on a broker's in-force book.
- Rung 4 shows the weakest year-one economics here because a fixed entity cost and a complexity premium sit on a narrow transferred segment. Raise the eligible segment in the assumptions and it converges. That is a finding, not a rounding error.