Directions Brain

Carrier appetite knowledge is the SMB commercial bottleneck

6 min read

Carrier appetite knowledge is the SMB commercial bottleneck

Claim

In SMB commercial insurance, the binding constraint on quoting throughput is the broker’s tacit knowledge of which carrier will accept which risk — not data entry, not document parsing, not form-filling. If true, an AI-native broker’s first defensible capability is an appetite-matching engine, and automating submission mechanics alone produces little advantage.

Raised by

Supporting evidence

  • 2026-08-17-directions-jim-coronado-multiline-broker-interview — carrier portals accept a complete submission and reject on appetite only at the final screen, so a broker who doesn’t know appetites pays the full data-entry cost per carrier and gets nothing. Jim’s response is to stop quoting and start phoning: “I don’t want to waste my time quoting five or six or seven different commercial carriers to find out that nobody wants it.”
  • 2026-08-17-directions-jim-coronado-multiline-broker-interview — the explicit junior-broker answer: a junior “will waste time going to every carrier and putting in the information and then finding out at the end that they don’t take that particular client.” The experience curve is appetite knowledge.
  • 2026-08-17-directions-jim-coronado-multiline-broker-interview — worked examples held in memory, not in software: hibachi restaurants → non-admitted (griddle fire/burn); buildings 30+ years old → several carriers decline; 1920s condo association → nobody; a 2–3-year-old condo association with a one-year coverage lapse → Travelers will take it, no surcharge.
  • 2026-08-17-directions-greg-ehly-independent-broker-interview — the same logic as a binary gate: inside direct-carrier appetite means same-day quote and bind; outside it means wholesaler, ACORD forms, supplementals, 3–4 days to quote and 1–2 weeks to bind. Appetite fit determines a ~20x difference in cycle time.
  • 2026-08-17-directions-jim-coronado-multiline-broker-interview — the decisive appetite call is made by a human marketing rep, who then hands off to an underwriter. The knowledge is relational, not just informational.

Counter-evidence

  • Neither broker was asked to rank appetite knowledge against other time costs, and greg-ehly separately named service and billing — not quoting — as his dominant daily load. Appetite may gate quoting speed while quoting itself is a minority of broker time. That would make this a real bottleneck in an unimportant place.
  • Carriers publish appetite guides on their websites; Jim uses them. To the extent appetite is documented, it is acquirable rather than tacit, and the bottleneck is retrieval rather than knowledge.
  • jim-coronado’s agency has no commercial rater for cost reasons, not capability reasons (ezlynx sells one). Some of what looks like an appetite bottleneck may just be an unbought-software bottleneck.
  • Unknown: whether appetite data is stable enough to model. Jim noted appetites change, and carriers swing between retention and growth mode — an appetite engine trained on stale data could be worse than a phone call.
  • 2026-08-20-directions-jasmyne-mcdonald-farmers-captive-interview — the first evidence that the job is already partly tooled. jasmyne-mcdonald enters commercial risks into first-connect and gets back a list of carriers likely to be in appetite. That is an appetite-matching product in production. Two qualifications keep the hypothesis alive rather than killing it: the results are “sometimes incorrect and eligibility still fails,” and she still falls back on remembered per-niche defaults (K&K for events, Progressive for transportation). This reframes the wedge from existence to reliability — a narrower and more testable claim.
  • 2026-08-20-directions-jasmyne-mcdonald-farmers-captive-interview — appetite is also finer-grained than a carrier-level lookup can express: in transportation it depends on cargo type (car haulers nearly unplaceable, consumer goods widely written), and in habitational on building age and occupancy. A matching engine has to model risk attributes, not carrier categories.

Implications

  • An appetite-matching layer would be a genuinely differentiated capability, and one that compounds: every decline is training data, which is exactly the kind of asset insurance-technology-vision argues for.
  • It supports entering as a broker rather than as tooling — you only see decline outcomes at scale if you are placing the business.
  • It sharpens the junior-broker economics inside ai-collapses-smb-brokerage-labor-cost: if appetite knowledge is what takes years to build, an AI that holds it lets a cheap junior operate at senior throughput, which is the labor-cost collapse stated concretely.
  • It raises a channel question: the decisive escalation runs through carrier marketing reps. A new brokerage without those relationships may be structurally slower regardless of software.

Ideas this favors

Ideas this weakens

  • Any wedge whose core value is faster form-filling or document parsing — already discounted by H-INS-2, and this adds an operator-side reason.
  • insurance-checkup-funnel-design to the extent it assumes the hard part is getting the customer’s existing policy data in; that is the other seam, and this one sits downstream of it.

Confidence

Medium. Two independent operator interviews on the same day converging on the same mechanism is real signal, and Jim’s account is specific and unprompted. But it is two brokers, both self-reporting, neither asked to quantify how much of the week appetite actually costs them — and Greg’s “service and billing dominate my day” is a live tension with it. Not yet corroborated by any market or vendor evidence.

What would change our mind

  • Broker interviews where appetite matching is described as easy, already solved by published appetite guides, or trivial relative to service load.
  • A working product (bold-penguin, coverforce, herald) demonstrating that appetite routing is already commoditized in commercial submission APIs.
  • Time-allocation data showing appetite/market-selection is a small share of broker hours versus service, billing, and renewal work.
  • Evidence that carrier appetite changes fast enough that any modelled version decays below the value of a phone call.
  • A measured accuracy rate for first-connect-style appetite suggestions. If it is already high the bottleneck is gone; if it is low the wedge is accuracy rather than existence — either way this is now the decisive number, and it is obtainable from a vendor call.

Graph View

Backlinks