The matching engine — how buyers are connected to slices.

Most buyers experience deal sourcing as a search problem, and that's exactly why it's so frustrating. The things that actually determine whether a book fits you — its line mix, its carrier composition, its retention, its concentration, the capital it takes to absorb — none of them fit neatly into a search box, and none of them appear in a listing title. A matching engine is built for the real shape of the problem: it works on structured data, not text, and it does the looking so you don't have to.

§ 01 · Why keyword search failsMatching ≠ search.

A search engine returns keyword matches; a matching engine pairs structured criteria against structured asset attributes to evaluate compatibility. The distinction is decisive for agency M&A, and keyword search fails on three counts. First, the relevant attributes aren't in listing titles — line mix, carrier composition, retention, geography, concentration, and capital structure don't reduce to keywords, so a search for "commercial P&C Ohio" might surface a personal-lines book that merely mentions a commercial tail. Second, matching criteria are buyer-specific — two buyers looking at the same slice get different fits depending on their carrier appointments, line targets, and capital, so there's no single "relevance" a search could rank. Third, inventory is continuous — new books appear constantly, and passive browsing only captures what happens to be visible at the moment you log in. Saved searches help a little, but they still match on keywords, still wait passively, and still can't reason about a compatibility trade-off. The buyer profile that feeds all of this is built in the buyer profile.

§ 02 · The six dimensionsHow fit is scored.

The engine scores fit on a transparent, published weighting — state at 50%, carrier at 33%, and line of business at 17% — aggregated into a single match score, with a roughly 10% overlap marking a listing pre-qualified. That published weighting is the antidote to a black box: you can see exactly why a book scored the way it did. Around that core score, the engine evaluates six dimensions in all — line-of-business fit, carrier compatibility, geographic fit (gating on whether you're licensed, then preferring priority markets), size and capital match (mismatches are filtered out automatically rather than shown as near-misses), retention and quality against your profile minimums, and strategic-intent alignment with your declared play. Crucially, it ranks near-misses with the failing dimension flagged — a book that matches five of six dimensions with a negotiable miss is exactly the kind of deal worth evaluating, and a saved search would never reason its way to surfacing it. The deal-sourcing view of this same engine is covered in the matching engine, buyer-side.

§ 03 · Hidden inventoryBooks that aren't listings.

Because the engine works on structured attributes rather than posted listings, it can surface books that aren't open listings at all — the hidden-inventory advantage. The shipped source of this today is the fringe area: accounts at the geographic or line periphery of a larger agency's book, typically under 5% of its premium, that the agency isn't investing in and is a natural candidate to divest. The engine flags those and attaches a matched buyer, so they arrive as suggested slices rather than something you had to find. (A demand-driven counterpart — where buyer demand itself would point the engine at agencies to approach — is a market concept that frames how supply behaves, not a feature you act on today; the live suggested-slice flow is fringe-driven.) The full picture of how supply forms is in hotspots and fringe areas.

§ 04 · Working a matched feedFour disciplines.

A matched feed rewards a different posture than a search bar, and four disciplines get the most out of it. Invest time in the profile — it's the lever, because the engine's output is a direct function of profile quality: specific criteria produce specific matches, a vague profile produces vague ones. Expect matches, not listings — this is a curated feed, not a search-and-scroll, so a shorter, sharper set of results is the system working, not failing. Treat near-misses as strategic signals — a five-of-six match with a negotiable failing dimension is often the most interesting deal in the feed. And update the profile as your strategy evolves — match quality moves with it, so a profile that still reflects last year's thesis surfaces last year's deals. Worked that way, the matching engine stops being a place you visit and becomes a system that works the inventory for you — which is the whole point of putting the burden on the system instead of the buyer.

◆