Every intake metric starts counting after the business was already at risk, so the stretch that decides the loss ratio produces no data. One number brings it into the next pack.
A chief underwriting officer can state last quarter’s combined ratio to one decimal place. Ask how many submissions arrived in the same quarter, and how many of those were ever read, and in most underwriting packs we see, the number doesn’t exist in that form.
The intake metrics in a standard underwriting pack start counting after a submission has already been picked up, priced, or entered into a system. The stretch before that, where a risk either reaches an underwriter or goes unread, produces no data. That stretch decides the selection set, and the loss ratio is an average taken over it.
Rate has gone as a lever this cycle, which leaves selection. Selection is decided in a stretch of the funnel no carrier measures today, because nothing reads a submission until a person does. That stretch hides how much business never reached a desk, and one number brings it into view.
Written for chief underwriting officers, underwriting managers, pricing actuaries and MGA leadership.
01 The Funnel Has No InstrumentsThree numbers describe intake in most underwriting packs. Each starts its clock after the point where the business was already at risk.
Quote turnaround starts at underwriter pickup. The broker’s clock started when they hit send. The gap between those two moments doesn’t register in any system or report.
Hit ratio starts at quoted. It measures conversion on business the carrier chose to price. It says nothing about whether the carrier priced the business it meant to write.
Submissions received starts at entry to the workbench. Risks triaged out of an inbox before clearance never enter the count, and their decline reasons go with them.
Carriers do track submissions, carefully. They track what entered a system. Take a risk that arrived in a broker email, waited eleven days, and was answered too late to matter. It leaves no trace in any of the three, because by the definition each metric uses, it never happened.
So the measured funnel begins where an underwriter has already engaged. A carrier can run a disciplined portfolio review every quarter on a population it didn’t choose.
02 Why the Unmeasured Stretch Is the Expensive OneMcKinsey’s Global Insurance Report 2025 examined 24 commercial lines carriers to find where performance separation comes from. 60% of performance was explained by how a carrier operates. 40% by the lines it writes. The split held through hard and soft markets alike.
Top-quartile carriers hold administration expense ratios two percentage points below their peers. On acquisition cost, leaders cut two points between the 2013 to 2017 and 2018 to 2023 periods while laggards added two. Both gaps are real, and both are small. Lloyd’s own expense ratio rose to 36.4% in the first half of 2026 from 35.8%, on higher acquisition costs and profit-related commissions.
The 40% is the lines a carrier chooses to be in. The 60% is everything about how it writes them, selection and terms included. McKinsey found leaders almost twice as likely to have made significant modernization investments in underwriting. It also found position in the ranking holds: five of seven top performers in 2013 to 2017 were still top performers in 2018 to 2023, and only two of eleven middle-quartile carriers moved up.
McKinsey does not break that 60% down to intake, however. The case for intake rests on sequence: selection operates only on the submissions that reached a desk, so everything a carrier improves downstream works on a population that intake already fixed.
03 The Constraint Is Reading TimeAccenture’s 2021 P&C Underwriting Survey, published in February 2022 and part of a longitudinal study running with The Institutes since 2008, put underwriters at 30% of their time on underwriting, 40% on administrative tasks, and 30% on negotiation and sales support.
That 40% is the bottleneck: the extracting, re-keying and chasing that separates a submission arriving from anyone forming a view on it. It sets how far down the queue a desk gets in a week.
Whatever the desk can’t reach goes unread, with no record of what it held.
Queue order decides the rest. Arrival time, broker relationship and follow-up persistence all move a submission up the list. None of those track risk quality.
Reading is the constraint, and it is the part of the workflow AI now handles at volume. Once every submission is read, queue order stops deciding which ones reach an underwriter.
04 What Changes When Submissions Are Read on ArrivalReading a submission at the point of arrival produces three records a carrier didn’t have before:
Kinetic Insurance Services already reads submissions on arrival, in production. Convex Group, an international specialty insurer, launched Kinetic in June 2026 as an MGA platform for large, complex and middle-market specialty risks. Provectus built its AI-powered underwriting platform together with Quotech, whose underwriter-led team brings first-hand knowledge of how a London-market submission moves.
The front of that platform does the reading. Submissions arrive as broker emails, ACORD-style forms, SOV spreadsheets and scanned loss runs, rarely in the same shape twice.
API-connected data capture handles the multi-format ingestion and extraction. The platform normalizes submissions into the OASIS Open Exposure Data standard, then triages and routes them against appetite. Authority controls are built into the underwriting workbench, so the sorting finishes before a person opens the file and the decision stays with the underwriter.
None of those three exist until something reads the submission on the way in.
05 Three Numbers to Put in the Next PackStart with the one that shows the blind stretch. Submissions received at the point of arrival, against submissions cleared into the workbench, split by whether the risk was inside stated appetite. One specialty line, last four quarters.
The difference between those two figures is a selection decision nobody made.
A carrier that can’t split it yet has found the first thing worth fixing, because every ratio further down is computed from it.
Two more are worth pulling at the same time. Median days from submission arrival to quote issued, against the quote turnaround the pack reports: that gap is time no system owns. And hit ratio on risks inside stated appetite, against hit ratio overall: that gap is business written nobody meant to write.
06 ConclusionThe loss ratio is an average taken over the risks a carrier wrote, drawn from the risks it read. Carriers manage the first half of that closely and hold no instruments on the second.
The submissions that never reached a desk this quarter are already in next year’s result, and nothing in the current pack will show where they went.
Reading every submission on arrival turns the top of the funnel into something a carrier can report on, and then manage. Submission Flow, the Provectus blueprint for the front of the underwriting funnel, is how we build it with you.
Start with the three numbers above, on one specialty line and the last four quarters.
We pull them with you against the workbench you already run. The baseline is agreed before anything is built. Then the current workflow and the new one run on identical submissions through one renewal cycle. Two weeks to the first read.