Slices is the fractional-M&A primitive on the Milly Books marketplace. Instead of being forced into the historical all-or-nothing trade — sell the whole agency or keep operating — sellers can carve out a defined portion of the book and list it independently. This Tactical names exactly what ships, lists the four dimensions, and surfaces the safeguards (MVT + CRP) that determine whether a Slice can transact at all.
The four Slice dimensions, each as a card.
Each dimension has its own dbt macro and its own use case. A Slice is defined by exactly one dimension — you do not stack dimensions in a single Slice. If you have a Travelers commercial book in Florida that you want to carve, the cleanest path is usually to slice by carrier (Travelers) or LOB (commercial), not both — the matching engine performs better against single-dimension Slices in current production.
Carrier Slice.
Carve out the entire book of a single carrier. The most common use case is a carrier relationship that no longer fits — appetite shift, commission compression, geographic moratorium, or simply a non-strategic partner that you'd rather hand off cleanly. The Slice contains every policy with that carrier across all LOBs and states; CRP ensures multi-policy clients bundle correctly. Fringe-area (≤5% of book premium) carrier Slices are surfaced by Suggested Slices daily.
LOB Slice.
Carve out an entire line of business — commercial, personal, life-and-health, or a specialty practice. The use case is a non-core LOB that you no longer want to service, or a practice area that has been growing in a direction the agency does not want to specialize. Personal Lines is a frequent LOB-Slice candidate for commercially-focused agencies. CRP applies: a personal-lines auto policy and a commercial-lines workers' comp policy for the same client must stay together or split through CRP rules.
State / geography Slice.
Carve out the book for a single state. The use case is a geographic outlier — a few accounts in a state you no longer want to maintain licensing or carrier appointments in. State Slices simplify operationally; they often clear quickly because they map to a buyer's existing state footprint without integration complexity. Multi-state clients require CRP review.
Insurance-type Slice.
Carve out by insurance type — a specialty cut at the policy-type level (e.g., umbrella, professional liability, cyber). This is the most granular dimension and the one most often used when a specific specialty no longer fits the agency's strategic focus. Buyer demand is typically concentrated among specialists in that exact type.
A Slice must pass all three conditions to be listable.
- ≥ 5 policies
- ≥ $10,000 premium
- ≥ 0.01% of total book premium
Customer Relationship Protection — the CRP safeguard.
CRP is the platform's safeguard against client splintering. The Unique Customer ID bundles all policies for a given client into one indivisible unit. The logic is "all-in or all-out": a client cannot be split across two Slices to different buyers. If a client has an auto policy and a home policy, both move together — to the same buyer or not at all.
CRP exists as a concept in the data model; the Unique Customer ID is in place. The enforcement layer at the dbt/Trantor tier is shipping soon — until then, qualify it as "the model is in place; the enforcement layer is in progress." Practically, this is rarely the gating concern — the dimensions are defined such that CRP rarely conflicts with the Slice the seller wants to publish.
The four dimensions plus MVT plus CRP define what a Slice can be. They are the structural constraints that make fractional M&A work at all. Inside those constraints, you have the freedom of any all-or-nothing seller, plus the optionality of carving the book to match the buyer demand you can actually attract.
Phased Retirement as a seller strategy.
The most powerful use of Slices is sequential listing — selling portions of the book over multiple years. This is "Phased Retirement" in the operator literature, and it is a seller-side strategy, not a native platform listing type. The platform has a single publish/unpublish state per listing; sequential Slices accomplish the phasing.
The sequencing.
List a non-core Slice today (a carrier you no longer want, an LOB you do not specialize in). Close it; reduce operational load. Twelve to eighteen months later, list another Slice. Three to five years in, the principal book has been reduced to a focused core that the principal can either continue to operate at lower intensity, sell as the final transaction, or transition to internal successors. The combined sequence accomplishes Phased Retirement; no single platform feature does — Slices does.
"Phased Retirement" is what the multi-year Slice sequence accomplishes. The platform supplies the Slices and the marketplace; the strategy of phasing is the seller's.
What does not ship — the honest list.
Four claims that appear in older Milly Books marketing literature but do not match the production code today, surfaced here so the seller doesn't plan around features that don't exist:
- Producer Book as a Slice dimension. Not shipped. A retiring-producer Slice — sell the whole book of one producer as a unit — is in roadmap conversation but does not exist as a dbt macro or in the UI today.
- Hotspot / offensive Suggested Slices. Coming soon. The model exists (
int_book_appetite_score.sql); the wiring into the Suggested Slices pipeline is in progress. Today, Suggested Slices is fringe-area only. - Demand-sensitive EBITDA multiples. Not how the Book Valuation Engine works. It is a deterministic revenue-multiple formula; it does not vary multiples by Hotspot vs Fringe categorization.
- Phased Retirement as a native listing type. Not shipped as a feature; sequential Slices accomplish it as a strategy.
Honest mechanics work better than overstated ones. The four dimensions, the MVT gates, and the CRP safeguard are sufficient to do everything sellers actually want fractional M&A to do.
Slices is the structural answer to the all-or-nothing era.
Historical M&A on this asset class forced a binary — sell everything or keep operating. Slices removes the binary. A seller can carve out exactly the carrier, LOB, state, or insurance type they want to divest, list it anonymously, and transact against a buyer pool whose criteria already match. The marketplace creates the discovery; the four dimensions plus MVT plus CRP create the trust; the seller controls the timing and the sequence. The all-or-nothing era ends operationally, not just rhetorically.
Terminology on this shelf
- Slice
- A fractional book carve-out defined along one of four shipped dimensions: carrier, LOB, state, or insurance type.
- Four shipped dimensions
- Carrier, LOB, State/Geography, Insurance Type. Each is implemented as a dbt macro. Producer Book is not currently a shipped dimension.
- Minimum Viability Thresholds (MVT)
- The three gates that determine whether a Slice can be listed: ≥5 policies, ≥$10,000 premium, ≥0.01% of total book premium.
- Customer Relationship Protection (CRP)
- The safeguard preventing client splintering across multiple buyers. Unique Customer ID bundles all policies for a single client; "all-in or all-out." Enforcement layer shipping soon.
- Suggested Slices (fringe-area)
- The daily Dagster pipeline that surfaces book segments ≤5% of total premium across the four dimensions as ready-to-list Slice candidates.
- Phased Retirement
- The seller-side strategy of sequential Slice listings over multiple years to monetize a book incrementally. A seller strategy, not a native listing type.
- Grow by Subtraction
- The framing for fringe-area divestiture: streamline operations by carving out segments that cost more in management overhead than they generate in revenue.