Loss Runs Automation: Stop Chasing Carriers for Claims History

June 12, 2026

Every commercial submission worth marketing starts the same way: somebody has to round up the loss runs.

Three to five years of claims history, from every prior carrier, current enough that the underwriter won't bounce it back. Which means a CSR requesting reports from two or three carriers, waiting days for each one, chasing the ones that don't show up, and then re-keying claim details out of PDFs that all look different.

Nobody got their insurance license to do this. Here's what the workflow actually costs, which half of it you can automate today, and how agencies are doing it.

What the loss run chase actually costs

Take one marketed commercial account with three prior carriers. The typical sequence:

• Request: 10 to 15 minutes per carrier. Some have a portal path, some want an email to the underwriter, some still want a signed LPR (letter of record) before they'll talk to you.
• Wait: days per carrier. Some states set deadlines for carriers to produce loss runs, and carriers still drag to the deadline.
• Chase: 2 or 3 follow-up touches per slow carrier at 5 to 10 minutes each.
• Re-key: 20 to 40 minutes per report extracting dates of loss, claim types, paid amounts, reserves, and open/closed status into your submission or ACORD package.

Call it 2 to 3 hours of touch time per account, smeared across one to two weeks of calendar time. An agency marketing 10 commercial accounts a month is spending 20 to 30 hours a month on claims history paperwork. At renewal season it's worse, because every remarket needs fresh runs and they all need them in the same 60-day window.

The calendar cost matters as much as the hours. A submission that waits two weeks on loss runs is a submission your competitor might beat to market.

Why loss runs resist automation

Loss runs are the least standardized documents in the P&C workflow. Every carrier formats them differently. Different column orders, different valuation date placement, different ways of showing reserves versus paid, subrogation credits buried in footnotes. Some arrive as clean PDFs. Some arrive as scans of printouts.

That's why "just OCR it" never worked. A template built for one carrier's format breaks on the next carrier's. We've written about why rigid automation scripts fail in insurance generally, and loss runs are the extreme case: the input format changes with every single carrier.

This is exactly the problem modern AI document parsing handles well, because it reads documents the way a person does instead of matching a template.

The workflow has two halves. One automates today.

Half one: the chase. Requesting and collecting the reports. This half is mostly process discipline, and the fix is boring but effective:

• Request loss runs the day you decide to market the account, not the week the submission is due
• Use a standing request template per carrier (portal path or underwriter email) so nobody reinvents it
• Track requests like a pipeline: requested date, expected date, chase date. A report nobody is tracking is a report nobody chases.
• At renewal, trigger requests at 90 days out automatically, the same way renewal workflows trigger everything else

Half two: the re-key. Getting the claims data out of the PDFs and into your submission. This is the half AI handles outright. Document parsing reads a loss run in any carrier's format and returns structured claims data: date of loss, claim type, status, paid, reserved, total incurred.

The part that matters for E&O: every extracted field carries a confidence score. A clean typed amount scores high. A smudged scan of a handwritten reserve scores low and gets flagged for a human to confirm. Your CSR verifies three flagged fields instead of re-keying three hundred. Nothing goes into a submission without a person having seen the doubtful parts. Same approach we use for ACORD extraction.

What the parsed data unlocks

Once loss history is structured data instead of a stack of PDFs, the downstream work gets fast:

• Underwriter summaries. Five-year loss summary with frequency, severity, and open reserves, drafted in seconds from the parsed claims, reviewed by your team before it goes out.
• Cleaner submissions. Claims history lands in the package consistently formatted, no matter which carrier produced the original report.
• Faster go or no-go. A producer can see loss frequency before sinking hours into marketing an account that no carrier will touch.

The agencies that automate the re-key half first report the same thing: the chase still takes calendar time, but the touch time per account drops from hours to minutes, and submissions stop stalling on the last loss run to arrive.

Key takeaways

• Loss runs cost a typical agency 2 to 3 hours of touch time per marketed account, plus one to two weeks of calendar wait
• The workflow has two halves: the chase (process discipline) and the re-key (automatable today)
• Loss run formats vary by carrier, which kills template-based OCR. AI parsing with confidence scores and human verification handles any format safely.
• Structured loss data turns underwriter summaries and submission prep from an afternoon into minutes