Underwriter capacity in commercial P&C: how to grow GWP without growing headcount in a soft 2026 market

Commercial P&C is heading into a soft market with a workforce that is thinning faster than its book is shrinking. By the close of 2026, the US insurance sector is on track to lose roughly 400,000 workers to retirement and attrition, and the underwriting bench takes a disproportionate hit because the median tenure is concentrated above age 55. Submission volume, meanwhile, keeps climbing. The math no longer works the old way. Carriers that defend margin in 2026 will not do it by hiring. They will do it by raising the gross written premium their existing underwriters can responsibly handle.

That is the capacity question. It is the only operational question that matters at the moment, and most carriers are not framing it correctly.

Why capacity, not headcount, is the right question for 2026

For most of the hard market, growth was easy to manage. Rates were rising, capacity was tight, and any submission a carrier won came in at adequate or above-adequate pricing. The question was which accounts to write, not how many.

That has flipped.

S&P Global's 2026 US P&C outlook describes a market where competition is intensifying, pricing momentum is slowing, and brokers are gaining leverage on the better risks. Amwins's State of the Market notes early signs of softening across most major commercial lines, and Insurance Insider US has gone further, calling 2026 a market where deeper softening "may be closer than you think." Burns and Wilcox's 2026 webinar captured the operating reality more bluntly: capacity is genuinely back, investors are returning, and carriers are competing for the same well-performing accounts.

A soft market exposes any operational slack in the underwriting function. When premium is shrinking on a per-policy basis, the only way to defend the top line, never mind the bottom line, is to write more policies with the same team. The carriers we work with at Pibit.AI are already running this math privately. The ones doing it well are restating their internal targets in terms of GWP per underwriter, hit ratio, and quote turnaround, not just combined ratio. Our analysis of rising submission volume against flat underwriting capacity details how this gap has been widening since 2023.

The talent picture sharpens the urgency. Insurance Business reported that the US insurance sector is set to lose around 400,000 workers by 2026, much of that concentrated in the 55-plus cohort. The Independent Agent magazine's February 2026 piece on the talent crisis cited 538,000 employees aged 55 to 64 and 186,000 aged 65 or older still working in the industry, an unusually old workforce by US labor standards. That is not a problem you hire your way out of in 18 months. The pipeline of new commercial underwriters takes three to five years to develop into independent authority levels. The gap between submissions arriving and qualified underwriters available to work them is structural, not cyclical.

Hiring is not the answer. Capacity per underwriter is.

What the standard responses get wrong

When a CUO walks into the room with a capacity problem, three responses tend to surface. Each one solved a smaller version of this problem in the past. None of them are equal to the 2026 version.

Response one: hire more underwriters. The talent supply is the constraint. Most carriers we speak with describe filling a senior underwriter seat as a six- to nine-month exercise, sometimes longer for specialty lines. Even when the seat fills, the new hire is operating below full authority for another six to twelve months. The arithmetic does not work against a soft market that is here now.

Response two: route intake to offshore teams. This held up reasonably well during the hard market because volume was being squeezed at the front of the funnel by rate. It is breaking now for two reasons. The first is quality. Independent benchmarks and our own customer measurements show offshore extraction quality plateauing in the 90 to 95 percent range, which sounds high until you translate it into book risk. On a $500M book, a 5 percent error rate on extracted submission data exposes the carrier to silent loss of roughly $9M per year, a number we have walked carriers through in deal conversations. The second is broker speed. Offshore round trips add 12 to 24 hours to standard submissions. Brokers, sitting on more options, route to the carrier that responds first.

Response three: build it internally. A handful of tech-forward carriers tried this in the last cycle. The ones who shared their results candidly told us the same story: an internal OCR plus scripts approach hits a wall around 60 broker formats. After that, every new format is another six to eight weeks of rework, and the system regresses every time a broker tweaks a template. The IT team is now committed to a build that is not closing the underwriting capacity gap.

Each of these responses is rational on its own terms. None of them produce a 2026 capacity step change.

The capacity equation, restated

The arithmetic is straightforward once you see it.

GWP per underwriter = (submissions worked per underwriter per week) × (hit ratio) × (average premium per bound policy)

A carrier wanting to grow GWP per underwriter has three levers: more submissions worked, a higher hit ratio, or higher premium per policy. In a soft market, premium per policy is moving against you. Hit ratio has a ceiling determined by appetite, market segment, and competitive pricing. The only lever the carrier fully controls is submissions worked per underwriter.

Where does an underwriter's day go today? Industry research and our own meeting intelligence converge on the same answer. Roughly 30 to 40 percent of the underwriter's hours go to actual underwriting judgment, the rest to data assembly: pulling loss runs apart, normalizing statements of values, validating ACORD form fields, chasing missing information from brokers, opening multiple internal systems to confirm appetite. Our piece on the real job of a commercial underwriter walks through where that time leaks.

If a carrier reduces the data assembly portion of the underwriter's day from 60 to 70 percent down to 20 percent, that single change roughly doubles the time available for actual underwriting. In practice, the gain is less mechanical because of context switching costs, but the direction is correct. An underwriter freed from data entry can work materially more submissions, with sharper judgment on each one.

That is what capacity means in 2026. It is not faster typing. It is not a bigger team. It is the deliberate removal of data assembly work from underwriters who should be evaluating risk.

What removing data assembly actually requires

Every CUO has heard a variation of this pitch before. The 2026 version is different because the underlying technology is different and the proof points are concrete. Three operational requirements separate a real capacity gain from a marketing claim.

1. Template-agnostic extraction across every broker format

A carrier writing through hundreds of brokers receives loss runs in dozens of layouts, statements of values in spreadsheet formats that vary by retail agency, and ACORD forms in scanned and fillable variants that differ subtly across years. Any extraction system that depends on per-template configuration breaks the moment a broker changes the layout, and the carrier ends up paying for a system that the underwriting team quietly stops using. Pibit.AI's DocumentCURE™ runs template-agnostic extraction across loss runs, SOVs, ACORDs, broker emails, schedules of vehicles, and policy documents, including handwritten and degraded scans. There is no per-format setup. Our piece on SOV processing in commercial property details how this plays out specifically for property submissions.

2. Verifiable accuracy, not benchmark accuracy

The difference between 95 percent and 100 percent accuracy is enormous in a production environment. A 5 percent error rate sounds small. Translated into a typical commercial book, it produces millions in silent annual loss exposure and a steady stream of avoidable claims disputes. Pibit.AI commits to 100 percent contractual accuracy, with field-level provenance linking every extracted data point back to its source document. Underwriters can audit any number, any policy, any time. We covered the procurement implications of this distinction in why 95 percent AI extraction is not production ready and in the real cost of inaccurate underwriting data.

3. Decision-ready packages, not just extracted data

Extraction without orchestration is not capacity. The underwriter still has to assemble the appetite check, the prior loss summary, the OSHA and SAFER pulls, the news and litigation searches, the broker history. The CURE™ platform delivers decision-ready submissions: extracted data is normalized, validated against carrier appetite rules, enriched with external risk signals via ResearchCURE™, and packaged into the underwriter's existing platform. Whether the carrier sits on Guidewire, Duck Creek, Insurity, or a homegrown system, the underwriting team works in their existing environment. Our piece on submission intake automation for commercial P&C explains the orchestration model end to end.

When all three operational requirements hold, the data assembly portion of the underwriter's day drops sharply. That is where the 32 percent GWP per underwriter improvement comes from in our customer book. It is not magic. It is the structural redesign of the underwriter's workflow.

The unit economics, modeled honestly

Numbers are useful here because the capacity argument depends on them.

Take a mid-market commercial carrier with $500M GWP, 25 underwriters, and roughly 200 submissions per underwriter per week arriving across primary lines. That carrier is doing $20M GWP per underwriter on average, with a hit ratio of around 18 percent on standard accounts and an average bound premium of $11,000.

If the carrier removes data assembly work and an underwriter's submissions worked per week rises by 30 percent, the second-order effects compound. Same hit ratio, same premium per bound policy, GWP per underwriter rises to roughly $26M. Same headcount, same appetite, same underwriting discipline. The book grows by approximately $150M without a single new hire.

That is the headline number. The undercurrent is more interesting. Underwriters spending less time on data assembly tend to make sharper risk decisions, because they have time to actually review the loss history, evaluate the operational risk narrative, and engage the broker substantively on borderline accounts. The hit ratio on the right accounts often improves modestly. Loss ratios on bound business improve more visibly because adverse selection drops. We have measured combined ratio improvements in the 400 to 700 basis point range across customers, which on a $500M book means approximately $20M to $35M in improved underwriting margin annually.

The capacity decision shows up in two lines on the income statement: more premium written and a better loss ratio on what gets written. That is the math a CUO actually cares about.

What changes operationally

A carrier moving on capacity in 2026 will notice five operational shifts in the first quarter after deployment.

Quote turnaround compresses. Standard submissions move from 24 to 48 hour turnaround to same-day on most accounts. Brokers notice within two weeks. Hit ratios on first-look submissions tend to improve because the carrier is no longer late.

The intake bottleneck disappears. Submissions stop accumulating in the inbox. Underwriters open the morning to a queue of decision-ready files, not a wall of unsorted email and PDFs.

Underwriter time reallocates. The 60 to 70 percent of the day previously consumed by data assembly drops below 20 percent. The reallocation goes to risk evaluation, broker conversations on complex accounts, and proactive book management.

Hiring plans adjust. Open senior underwriter requisitions can be held without harming throughput. Carriers running multiple LOBs can move planned capital from headcount expansion into book quality investments, into underwriting technology, or into new program launches.

Broker relationships strengthen. Brokers route more business to the carrier that responds first with clean, accurate quotes. The compounding effect over 12 months is meaningful, especially in a soft market where placement is more competitive.

These shifts are observable, measurable, and verifiable in the underwriting team's daily work, which is what differentiates real capacity gains from theoretical ones.

The execution path for a CUO

A CUO running this play in 2026 has a relatively short critical path.

First, baseline the current capacity number. GWP per underwriter, submissions worked per week per underwriter, average quote turnaround, hit ratio by LOB, and the percentage of submissions that arrive cleanly versus those that require data assembly. Most carriers do not have this data assembled in one place. Two weeks with the underwriting operations lead is usually enough to build it.

Second, identify the LOB or program with the highest data assembly burden relative to underwriter time. For most carriers we work with, that is workers compensation loss runs, commercial property SOVs, or commercial auto vehicle schedules. Start there.

Third, run a structured pilot with clear success criteria. The metrics should be capacity-focused: submissions worked per underwriter per week, average turnaround time, accuracy holding above the contractual standard, and broker satisfaction scores if available. A reasonable pilot window is six to eight weeks on a single LOB.

Fourth, measure the second-order effects honestly. Hit ratio, loss ratio on bound business, underwriter retention. These tend to lag the capacity number by a quarter, but they are where the actual margin shows up.

Fifth, expand methodically across LOBs and programs. Modular deployment works because the platform is modular. Carriers can stand up DocumentCURE™ for loss runs first, add ClearCURE™ for clearance and triage next, then layer in ResearchCURE™ for external data enrichment. Our piece on why most AI rollouts in underwriting stall covers what the successful rollouts do differently.

The carriers that have run this path in the last 18 months are not running pilots in 2026. They are expanding programs and growing book size while the rest of the industry argues about hiring strategy.

What we tell CUOs who are still on the fence

The honest answer is that 2026 is a year where the capacity question becomes unavoidable. The talent gap is structural. The soft market reduces the price lever. Submission volume keeps rising because brokers have more options and route to more carriers. The status quo is not staying still; it is degrading quietly.

There is also a fair question about vendor risk. Carriers who have been burned by extraction vendors that promised 95 to 96 percent accuracy and underdelivered are right to ask hard questions. The protection there is contractual. Pibit.AI commits to 100 percent accuracy with financial penalties tied to the standard, model isolation per client so underwriting guidelines stay proprietary, SOC 2 (AICPA), ISO 27001, and NIST AI RMF certifications, and a 3 to 6 week implementation that delivers measurable capacity in the first quarter. Our piece on data security in underwriting automation lays out the full vendor evaluation framework for carriers in active procurement.

Capacity is the question. The answer is operational, measurable, and available now.