What underwriters actually want
Features do matter — but only the ones that convert admin time into premium.
Our first paper argued that an insurance system is worth exactly what it does to four numbers: cost to start, cost to run, cost to connect, and revenue per person. That can read as a case against features. It isn’t. Features are the mechanism; the numbers are the outcome. The apparent contradiction dissolves the moment you ask what an underwriter’s day is actually made of.
An underwriter does two kinds of work. The first is judgement: selecting risk, pricing it, negotiating with brokers, holding the line on appetite. That is the work they are paid for, the work that binds premium, and the work no system should touch. The second is administration: rekeying data out of proposal documents, drafting quote letters and schedules, chasing files, preparing endorsements, answering status queries, assembling audit trails. Ask underwriters how their time splits and the honest answer, in most agencies, is that well under half the day is spent underwriting. The rest is admin — and admin is the raw material a system converts into revenue.
So the features underwriters ask for are not indulgences. Read correctly, each one is a request to buy back judgement time — and every hour bought back has a market price, because it becomes quotes, and quotes become premium. The discipline is to map each feature to the number it moves.
The mapping
| What the underwriter asks for | What it actually replaces | The number it moves |
|---|---|---|
| A submission that arrives pre-filled | Rekeying risk data out of PDFs and email chains — often 30–60 minutes per risk before any judgement begins | Revenue per underwriter. Extraction with confidence scoring turns an hour of typing into a two-minute review. |
| Verify and issue | Drafting the quote letter, the schedule and the policy documents after the decision is made | Revenue per underwriter, and cost to run. The underwriter confirms the number and the analysis; the system issues everything else. |
| Appetite that answers instantly | Reading, working up and politely declining submissions that were never in appetite | Revenue per underwriter. Encoded rules triage on arrival, so judgement time lands only on winnable risks. |
| One file, one screen | Hunting through inboxes for the wording version, the referral note, the prior-year terms; rebuilding files for audit | Cost to run, and a safer binder. The record assembles itself as a by-product of trading. |
| Endorsements and renewals that run themselves | Mid-term administration and renewal re-keying — high volume, low judgement | Cost to run. The book services itself; people handle exceptions only. |
| Brokers who can self-serve | Status calls, document re-sends, and the quotes lost to a faster competitor while the file sat in a queue | Revenue per underwriter. Speed is a feature brokers reward with flow — the fastest quote wins a disproportionate share. |
Verify and issue, worked through
Take the second row, because it carries the theme. A submission arrives as a proposal document. The system reads it, fills the rater, scores its own confidence, and runs the risk against encoded appetite. What lands in front of the underwriter is not a task list but a decision: a number and the analysis behind it. The underwriter verifies (adjusting a loading, adding a subjectivity, or agreeing), and the system does the rest: quote letter out, schedule and policy documents generated from the wording version in force, bordereau row written, audit entry sealed. Submission to issued quote in minutes.
The economics follow in a straight line. Each underwriter handles multiples of their previous submission count, so premium per underwriter rises without hiring. The small end of the market — the $500 to $2,000 premiums that were uneconomic to quote by hand — becomes profitable at volume. Hit rate rises because the fastest quote wins. And the hours bought back do not vanish: they go into the market, into broker relationships and new facilities, which is where underwriters wanted to be all along. The feature they asked for and the number the business needed turn out to be the same thing.
There is a quieter economic effect worth naming: retention. Experienced underwriters are scarce and expensive to replace, and they increasingly choose employers by tooling — good underwriters do not want to spend a career rekeying. A workspace that respects judgement time is a recruiting asset with a measurable cost saving behind it.
The mapping rule
Every feature request should survive being rewritten in one sentence: this removes X minutes of administration per risk, which becomes Y more quotes a day, which is Z more premium per underwriter — or it reduces a named cost. If a feature cannot be written that way, it is a preference, not a requirement. Preferences are fine; they just should not drive the roadmap.
The conclusion
What underwriters want and what the business needs are the same thing seen from different chairs. The underwriter wants the admin gone so they can underwrite and sell. The owner wants revenue per underwriter to rise and cost per policy to fall. A system built on verify-and-issue serves both at once — which is why the features that matter are not the ones that look impressive in a demonstration, but the ones that give an underwriter their day back and send them into the market to use it.
Cuttleflow Systems · Perspective 02 · 33°53′S · 151°16′E · Sydney