Skip to main content
milly logo
Tactical · prose B01 For Buyers · Deal Sourcing

The matching engine — three dimensions, scored and shown.

A matching engine that surfaces acquisition targets is only useful if a buyer can trust why a listing scored. This one isn't a black box: it scores fit on three observable dimensions with a published weighting, shows the work per dimension, and — critically — operates on data tuples alone, so a seller's identity is never in the match computation.

An outbound matching engine is only as useful as a buyer's confidence in it — a black box that surfaces "matches" without explaining why teaches a buyer to ignore it. This engine is built the other way. It scores a buyer's stated appetite against live listings on three observable dimensions, publishes the weighting, shows the per-dimension overlap, and computes on data alone rather than identities. The result is a match score a buyer can interrogate, which is what makes the outbound vector worth using rather than second-guessing.

§ 01 · The three dimensionsThe published weighting.

DimensionWeight
State (geography)50%
Carrier33% — carrier-appointment overlap
Line of business17% — the lines you write

The weighting is published, not hidden: state at 50%, carrier at a third, and line of business at the remaining sixth, aggregated into a single match score. The logic behind the order is practical — geography is the hardest constraint to work around in an agency acquisition, so it carries the most weight, with carrier overlap next (because shared appointments drive post-close synergy) and line of business completing the set. A buyer reading a match score can see how each dimension contributed, which is the difference between a number to trust and a number to ignore.

§ 02 · How a match scoresClean fit vs. miss.

The scoring is deliberately legible. A listing where all three dimensions overlap the buyer's appetite scores the full premium weight; a listing that misses on a dimension the buyer specified is steeply discounted. Where a buyer's profile leaves a dimension open — no stated preference — that dimension is treated as "no preference" rather than a miss, so a buyer who's flexible on geography isn't penalized for fitting listings in any state. A 10% overlap threshold is the line that marks a listing pre-qualified — enough fit to be worth a buyer's attention. The transparency is the point: the per-dimension overlap is shown, so a buyer can see exactly why a listing cleared the threshold or fell short, rather than taking the engine's word for it.

§ 03 · Confidential by designTuples, not identities.

Journal axiom · 1 of 2

The engine is confidential by design: it operates on data tuples — carrier, line of business, state, and premium — and the seller's identity is never part of the match computation. The score is produced from what the book is, not who owns it. That's what lets a confidential seller participate in matching without exposing their identity until they choose to — the match runs on the numbers, not the name.

The confidential-by-design property is what reconciles two things that would otherwise conflict: a seller's need to keep a sale confidential, and a buyer's need to find fitting targets. Because the engine computes on the book's characteristics rather than the seller's identity, a seller can be matched to a buyer without ever being named in the process — the identity is revealed only when the seller decides to engage. For the buyer, that means the match score reflects genuine fit (the actual carrier, line, state, and premium profile) even when the seller behind it is still anonymous, so the outbound vector works without forcing sellers to choose between participation and confidentiality.

§ 04 · Today and coming soonWhat's shipped, honestly.

The three-dimension scoring with the published 50/33/17 weighting is shipped and running — it's what powers the match data a buyer sees on every listing today, plus a daily curated outreach to fitting buyers. What's coming soon, and worth naming honestly rather than over-claiming: fuzzy matching (treating adjacent carriers or related lines as partial fits rather than binary), passive matching that fills portfolio gaps, and real-time alerts the moment a fitting listing publishes. Those are on the roadmap, not in the product, and a buyer should plan around what ships today: a transparent, three-dimension match score, a 10% pre-qualified threshold, and a daily curated set of fits. The engine's honest pitch is its transparency and its confidentiality — it shows the work and protects the seller — not a claim to intelligence it doesn't yet have.

Terminology on this shelf

Three matching dimensions
State (50%), carrier (33%), and line of business (17%) — the published weighting behind the match score.
Match score
The aggregated, per-dimension-transparent fit score a buyer can interrogate.
Pre-qualified threshold
The 10% overlap line that marks a listing worth a buyer's attention.
No-preference handling
An open profile dimension treated as flexible, not a miss — so a flexible buyer isn't penalized.
Confidential by design
The engine computing on carrier/line/state/premium tuples, never the seller's identity.
Coming soon
Fuzzy matching, passive gap-fill, and real-time alerts — on the roadmap, not yet shipped.

From the buyer theme

One piece every other Tuesday.

The next long-form piece in your inbox the morning it goes live. No marketing. Unsubscribe in one click.

Anonymous by default · One click to unsubscribe