Loss run analysis automation: five carriers’ loss runs, one loss picture, every number traced to its row
Loss summaries in minutes, not an hour per account, each claim traced.
See one case, screen by screen ↓Software that reads the loss runs prior carriers send with a submission and turns them into one loss summary an underwriter can price from. Loss Run Analysis Automation reads every run in any layout, maps each carrier’s columns to one set of fields, ties each claim to a policy year and line, shows reserve development, flags stale and missing runs, and drafts a cited loss narrative.
How to choose AI solutions for insurance →
Updated
Every commercial submission arrives with loss runs: one report per prior carrier, per line, for the last three to five years. No two carriers lay them out the same way. One sends a spreadsheet with paid and outstanding expense in separate columns, one a landscape PDF grouped by policy period, one a scan that shows loss only, without adjusting expense; the workers’ comp carrier splits every claim into indemnity, medical and expense. Each is valued on a different date, and some are years out of date.
So an underwriter or assistant re-keys them into a spreadsheet to see frequency, severity, open claims and large losses before pricing. “Incurred” includes expense on one run and not the next, status words differ, and the sheet adds them up as if they were the same. The detail that matters most on a casualty account — how far the reserve on an open claim has moved since the last evaluation — only shows when two evaluations sit side by side, so it is the one most often missed.
Estimated split for a five-carrier, three-line account, by hand and with the solution, in minutes. The underwriter’s own reading and acceptance come on top.
Six agents read, map, match and summarize every loss run, each number cited to its row; the underwriter decides.
Linden & Shaw sends the Calloway Cold Freight submission — refrigerated trucking and cold storage, auto, general liability and workers’ comp, quote due Friday 23 October. Here is what happens next, screen by screen, in the working solution.
Marisol opens her accounts in quote. Calloway Cold Freight is ready for her: 5 runs, 69 claims, $1,638,650 incurred over five years, a 44.2% loss ratio and 3 gaps. Brightwater Dairy is still being read, page 6 of 11. Each account carries a five-year incurred strip, red where a year has a large loss.
“31 loss runs read this week · median 4 min from email to a loss picture”
Across the top, the five runs as received on 9 Oct: Penmark’s spreadsheet export, Harbor Crest’s landscape PDF, Ironridge’s scanned PDF marked loss only, Northwind’s claim detail and Bluestem’s one page — each with its valuation date and age, red at 104 days and 3 years 7 months. Below, they become one loss picture by policy year: claims, open, paid, reserve, incurred, premium, loss ratio and large losses.
The run log shows how the picture was built. The loss run reader ties Penmark’s 18 rows to the sheet total of $674,550, reads Ironridge’s two scanned pages and checks two faint figures against the page total. The field mapper notes that Ironridge shows loss only, with no adjusting expense. The policy period matcher places 69 claims in 5 policy years × 3 lines and finds that auto 2021–22 has no run.
A faint figure is shown and checked against the page total — never filled in.
Every number has a thread. Marisol hovers the $412,000 auto chip in 2024–25, and a line runs to row 8 of the Penmark sheet: PC-AL-24-118407, paid $61,250 plus reserve $350,750, open. Underneath, what nobody had to look for: a year ago this claim stood at $115,000.
“At the Sep 2025 evaluation: $115,000 · +297,000 since”
The account’s two earlier submissions on file are lined up with the current runs. Policy year 2024–25 went from $252,850 at the Sep 2025 evaluation to $601,150 now, ×2.38, read from 18 rows on two runs. 2021–22 sits flat at $75,550 — because the same March 2023 Bluestem run has been sent three years running.
The claims view puts all 69 claims in one layout, with filters for open (10), large (4) and moved by $10,000 or more (3). The I-70 work-zone rear-end opens beside the table as Penmark sent it, with its history across evaluations: not yet reported in Sep 2024, $115,000 in Sep 2025, $412,000 now, in suit since 9 Jun 2026.
On the Ironridge loss runs tab, page 2 of the scan carries the carrier’s own stamp: loss only, ALAE not included. The column map reads Paid and Reserve as loss only and Incurred as no adjusting expense, so auto 2022–23 and 2023–24 are marked LO in every total rather than mixed with runs that include expense.
“Loss-only runs are marked, never grossed up.”
Gaps checks the runs against the prior policies on the application, the 90-day valuation window and the evaluations on file. Four items go in the request: the missing auto 2021–22 run for Ironridge policy IR-CA-7781302, a current Ironridge run, a current Bluestem run for open claim BLC-W-21-0337, and a status note on the I-70 suit. The email to Tom Albrecht at Linden & Shaw is drafted, with a follow-up date of Mon 19 Oct.
The valuation checker drafts; only the underwriter sends.
Five years, 69 claims and $1,638,650 incurred against $3,706,400 of premium. Four losses of $100,000 or more make up $858,450, 52% of incurred. Without the I-70 claim, 2024–25 would sit at a 23.5% loss ratio. Each citation opens the source row in the carrier’s own layout, and the data limits — the missing year, the stale runs, the loss-only scan — are stated plainly.
“No pricing view: that is yours.”
Marisol accepts the summary. Because one open claim is over $250,000, the account goes to Graham Okoye, the underwriting manager, before quote — the file checklist and the activity on the account record both. The agents never suggest a price, a credit or a decline.
The same account, seen by the four people who work it — what their week looked like, and what it looks like now.
Six agents read, map, match and summarize every loss run, each number cited to its row; the underwriter decides.
Reads every page of every loss run — spreadsheet, PDF or scan — into rows, and ties the rows to the run’s own page and period totals.
Maps each carrier’s columns and status words to one set of fields: paid, reserve, incurred, status, line, dates.
Ties each claim to a policy year and line by policy number and date of loss, and checks the years against the application’s prior-carrier list.
Matches claims to the evaluations on file, measures reserve movement, and finds one accident opened under two coverages.
Ages every run against the valuation window, flags missing years and stale runs, and drafts one request to the broker.
Builds the summary by year and line, the development view, large losses and loss ratios, and writes the cited loss narrative for the file.
Accepts the summary, sends the request. The agents propose; a named person decides.
Underwriters ask in plain words — what moved, what is missing, what a change would do — and every answer is cited to the loss runs and the evaluations on file.
Why did 2024–25 more than double in a year?
Policy year 2024–25 went from $252,850 at the Sep 2025 evaluation to $601,150 now, ×2.38. $297,000 of the $348,300 is one claim: PC-AL-24-118407, the I-70 work-zone rear-end, now in suit. Six claims reported after Sep 2025 add $23,100 and six others developed by $28,200. Without that claim the year would sit at $189,150, a 23.5% loss ratio.
Which open claims moved most since last year?
Three claims moved by $10,000 or more since the Sep 2025 evaluation: PC-AL-24-118407, auto, $115,000 → $412,000; NW-WC-2304512, workers’ comp lumbar fusion, $96,000 → $186,500, still open; and IR23A-01377, auto, $104,000 → $121,750, closed in Feb 2026. The two open ones carry $400,750 of the $561,850 outstanding.
What do we still need from the broker?
Three things before quote: the 2021–22 auto run — Ironridge policy IR-CA-7781302 is on the application but not in the runs; a current Ironridge run, since the one sent is valued 30 Jun 2026, 104 days ago, over your 90-day window; and a current Bluestem run — it is valued 1 Mar 2023 and BLC-W-21-0337 was open at $53,000. I would also ask for a status note on the I-70 suit. The request to Tom Albrecht is drafted for you to send.
Set the large-loss threshold to $150,000
Done. At $150,000, two Calloway losses are large: $412,000 auto and $186,500 workers’ comp — $598,500, or 37% of incurred. HCM 0022-6189 ($138,200) and IR23A-01377 ($121,750) move back into the base. The change is version 7 of your settings; accounts read from now on use it.
13 screens from the working solution, on its sample data. Pick one to see it large.
How loss runs are read and summarized is a set of versioned settings — the underwriting manager changes them, and accounts read after the change use them.
| Setting | Default | Choose from |
|---|---|---|
| Large-loss threshold | $100,000 | $50,000 to $500,000, in $25,000 steps |
| Valuation window | 90 days | 30 to 180 days, in 30-day steps |
| Look-back | 5 years | 3 · 5 · 7 years |
| Include adjusting expense in incurred | On | On · Off; loss-only runs are marked LO, never grossed up |
| Count an accident on two coverages once | On | On · Off |
| Count closed-without-payment claims | On, in frequency | On · Off |
| Development view | All evaluations on file | All evaluations on file · Last two evaluations |
| Reviews before quote | Underwriting manager, Graham Okoye | Any open claim over $250,000 |
Every figure is labelled: a target the solution is built to, an estimate, a typical published result, or a proven one.
“demo” = seen in the working solution, on its sample submission · “target” = the design goal, measured in the live solution · “estimated” = our estimate · “typical” = published figures (vendor and consultancy ranges for manual loss-run preparation — Decisions, Perceptive Analytics, Nomad Data — one to three hours for multi-carrier casualty accounts; The Institutes and Accenture, P&C Underwriting Survey, 2021 — up to 40% of underwriters’ time on administrative and non-core work; Fla. Stat. §627.444 — 15 days to send a loss run on written request). People, companies, carriers and claims named on this page are fictional — characters and sample data in the working solution.
Software that reads the loss runs prior carriers send with a submission and turns them into one loss summary an underwriter can price from. Loss Run Analysis Automation reads every run in any layout, maps each carrier’s columns to one set of fields, ties each claim to a policy year and line, shows reserve development, flags stale and missing runs, and drafts a cited loss narrative.
Yes. It reads spreadsheets, PDFs and scans. In the working solution it reads a Penmark spreadsheet export, a Harbor Crest PDF grouped by policy period, a two-page Ironridge scan, Northwind’s claim-detail blocks and a one-page Bluestem PDF. A new layout is read on first sight and kept once someone confirms its column map; 37 layouts are learned in the demo.
Yes. The continuity checker lines up the account’s earlier evaluations on file with the current runs and measures what moved. In the demo it finds PC-AL-24-118407 up $297,000 since the Sep 2025 evaluation, and shows policy year 2024–25 developing ×2.38.
A run that shows loss without adjusting expense is marked LO in every total and never grossed up. One accident opened under two coverages is counted once, and both coverages’ reserves are kept as reported.
The valuation checker ages every run against your valuation window and checks the years against the prior carriers on the application. It lists each gap with its reason and drafts one request to the broker; the underwriter edits and sends it.
No. The agents read, map, match and draft. They express no view on price, credit or declination; the narrative says so, and those decisions stay with the underwriter.
Yes. The underwriter accepts every summary and sends every broker request. Any open claim over $250,000 goes to the underwriting manager before quote, and every correction, acceptance and message is recorded with the person’s name.
The Agentic Solution Engine builds and deploys it from your requirements — your underwriting guidelines on loss runs, your file template and a sample of past submissions — and it goes live once every quality gate has passed. We will walk you through it on your own loss runs first.
Submissions ready to price the same day, poor fits declined in minutes.
1-min demoSchedule of Values ValidationEvery location geocoded and its COPE gaps flagged before the property is priced.
1-min demoUnderwriting Authority & Referral ManagementAnswers in seconds, each referral tied to its rule and authority letter.
1-min demoWe’ll run Loss Run Analysis Automation on a sample of your own submissions’ loss runs.