Most agency transactions still run through a broker, a spreadsheet, and a stack of phone calls. The Milly Books marketplace app collapses that process into a single product that both sides of the table use directly — a seller lists a book, a buyer finds it, and the deal moves from first interest to funded close without ever leaving the platform. This Pillar is the end-to-end walkthrough of how that works.
The product is organized around a simple premise: the friction in agency M&A is in the process, not the book. A seller loses value to missing data and a thin local buyer pool; a buyer loses value to a competitive, expensive, visible market. The marketplace app re-engineers the process so neither side pays that tax — confidential reach, objective valuation, and intelligent matching are built into the surface, not bolted on as advice. The six capability areas below are the product expression of that premise.
§ 01 · The two-sided modelBoth sides, one product.
A marketplace only works when both sides are present and the matching between them is good. The Milly Books app is built as a genuinely two-sided product: the seller's listing flow and the buyer's discovery flow are two faces of the same data model, not two separate apps stitched together. A book a seller publishes becomes, the same day, a candidate the matching engine scores against every active buyer profile.
That single-model design is what makes the rest of the product coherent. The valuation a seller runs is the same number a buyer sees framed against the market. The line-of-business mix a seller enters is the same signal the matching engine uses to find fitted buyers. The offer a buyer submits threads into the same record the seller manages. One book, one data model, two views — that is the architectural choice that the rest of this walkthrough rests on, and the tech stack overview covers the design decisions behind it.
The marketplace is not a listings board with a buyer directory bolted on. It is one data model with two views — every seller action is, the same day, a signal in the buyer's matched feed.
§ 02 · Anonymous listingsConfidential by default.
The single largest reason owners do not test the market is fear of exposure — that staff, carriers, or competitors will learn the agency is for sale before the owner is ready. The marketplace answers that fear structurally: a listing is anonymous by default. The market sees the shape of the book — premium volume band, line-of-business mix, retention, carrier count, geography at a coarse grain — but not the agency's name, address, or identity.
Anonymity is not a setting the seller has to remember to enable; it is the default state of every listing. Identity is revealed only when the seller chooses to advance a specific, vetted buyer — and even then, on the seller's timing. This inverts the broker model, where going to market means going public to a curated list. Here, the seller reaches a broad, competitive buyer pool while staying invisible until the moment they decide otherwise. The seller theme's anonymous listing strategy covers how owners use that confidentiality as leverage.
§ 03 · SlicesSell what you don't want to keep.
Not every seller wants to sell the whole agency. A principal may want to shed a personal-lines book while keeping commercial; offload a geography that no longer fits; or release a retiring producer's book without unwinding the firm. Slices is the product primitive for that partial exit — a structured listing of a defined portion of the book rather than the entire agency.
A Slice carries the same machinery as a full listing — anonymity, valuation, matching — applied to a carved-out segment. The economics are often better than owners expect: because a Slice can be matched precisely to a buyer with overlapping carrier appointments and adjacent geography, the per-dollar multiple frequently holds within a quarter of the whole-agency comparable rather than collapsing the way a fire-sale carve-out would. Slices turn what used to be an all-or-nothing decision into a continuum, and the buyer-side fractional acquisitions cluster covers how acquirers use them to enter at lower capital intensity.
§ 04 · The Book Valuation EngineAn objective anchor.
The Book Valuation Engine is the seller's first interaction with the platform and the anchor for everything downstream. It takes the agency's own performance data — revenue, line-of-business mix, retention, growth, profitability — normalizes it the way the M&A market does, and computes a defensible indicative value. The seller gets a number grounded in the same benchmark discipline a buyer's analyst would apply, before any conversation with a buyer begins.
The value is "objective" in a specific sense: it is computed from the book's data against published market benchmarks, not negotiated from an opening ask. That matters because the single most expensive mistake a seller makes is accepting an offer they have no independent basis to evaluate. With a defensible valuation in hand, both sides negotiate against the same anchor — the seller is not guessing, and the buyer cannot anchor the conversation on a lowball. The market theme's agency benchmarks reference covers the benchmark datasets the valuation rests on, and the engine itself is built on the platform's data pipeline.
The most expensive mistake a seller makes is accepting an offer they have no independent way to judge. An objective valuation, computed before the first conversation, is the structural fix — both sides negotiate against the same anchor.
§ 05 · Intelligent matchingMatched, not searched.
A buyer browsing thousands of listings by hand is doing the platform's job for it. Intelligent matching does that work once, on every listing, against every buyer profile. Three signals carry most of the weight: appointment overlap (does the buyer carry the same carriers, so the book transfers cleanly), line-of-business mix (does the book's commercial-versus-personal balance fit the buyer's appetite), and geographic fit (is the book in or adjacent to the buyer's footprint).
The result is that a buyer does not search a catalog; they receive a fitted shortlist. A tuck-in acquirer who would otherwise scroll past thousands of listings is shown the handful of books — sometimes Slices — that actually match their thesis. That same scoring runs in both directions: a seller's listing surfaces to the buyers it genuinely fits, which is what produces competitive tension without the seller having to shop the book around. The full scoring model — including the binary checks that drive most of the signal — lives in the matching and appetite-scoring engine page.
§ 06 · Offers to escrowFrom interest to funded close.
Once a buyer and seller engage, the deal mechanics stay inside the product. A buyer submits an offer; the seller responds or counters; the offer chain threads with a clean audit trail; and when terms are agreed, the deal advances through the escrow flow to a funded close. The product handles the state transitions — interest, offer, counter, agreement, escrow, close — so neither side is reconciling a deal across email, spreadsheets, and a separate escrow agent.
The discipline here is atomicity and auditability: an offer chain that holds up in due diligence, a document trail that does not depend on anyone's inbox, and an escrow integration that sequences funding and document hand-off without ambiguity. The deal-workflow mechanics — the state machine, the offer chain, the escrow integration — are the subject of the deal-workflow backend Pillar.
- Anonymous listing published — the market sees the shape of the book, not the identity.
- Objective valuation computed — both sides anchored on the same defensible number.
- Intelligent matching scores the listing against every active buyer profile.
- Fitted buyers engage; offers and counter-offers thread with a clean audit trail.
- Agreed deal advances through escrow to a funded close — inside the product.
§ 07 · One app, two viewsBuyer flow, seller flow.
The same application serves two audiences with two distinct journeys. The seller's journey runs from valuation to listing to managing inbound interest; the buyer's journey runs from profile to matched feed to offer. Both live in one authenticated product, which is why the data stays coherent — there is no export, re-import, or reconciliation between a "seller tool" and a "buyer tool."
For the seller, the app is a confidential, data-anchored path to a competitive outcome without the broker tax or the local-bubble discount. For the buyer, it is a fitted deal-flow engine that surfaces off-market-grade opportunities the visible market does not. The two journeys meet at the listing — the one object both sides act on — and the rest of the platform exists to make that meeting fair, fast, and confidential. The platform-success thesis covers what that produces for each side that the legacy broker process does not, and the seller persona architecture covers how the product speaks to different kinds of sellers.
The product does not ask sellers and buyers to trust a process. It removes the friction structurally — confidentiality, objective valuation, and fitted matching are defaults in the surface, not advice the user has to act on.
The six capabilities this Pillar walks — the two-sided model, anonymous listings, Slices, the Book Valuation Engine, intelligent matching, and the offer-to-escrow flow — are the user-facing expression of the whole platform. The tech stack overview covers how the services compose beneath them; the data pipeline covers the intelligence layer that powers valuation and matching; and the transactional backend covers the deal mechanics that carry a listing to a funded close.