By Liam Richards, CEO, Arete Adjusting
I have spent the majority of my career on the claims side of the marine, cargo and transportation insurance market. In that time I have worked across a wide range of losses, geographies and lines of business, and I have seen first-hand what it means to be the person responsible for delivering on the insurance promise.
When someone submits a claim, something has already gone wrong. That is the starting point for every claims professional, every time. Brokers bring insureds to the table. Underwriters make the promise of protection. Claims professionals are the ones who have to deliver on it. Insurance is a contractual promise upheld by many different people, each with a distinct role. Claims is where that promise moves from paper to practice.
For most of my career, the conversations about what that actually feels like have happened informally. Between peers. Over coffee. In the margins of events designed for someone else.
That is what prompted this research.
Over the past several months I have had conversations with senior claims and insurance professionals across the United States, Europe and the Middle East. Adjusters, TPAs, delegated authorities, underwriting managers, brokers and freight professionals. People sitting with live files, managing real losses, making decisions under pressure with incomplete information and time constraints the people above them rarely see.
I did not go into those conversations with a thesis. I went in with questions. What came back was consistent enough, and urgent enough, that it became clear someone needed to write it down.
The people best placed to describe what is happening in marine, cargo and transportation claims are the professionals living it every day. Not analysts. Not market commentators. The people holding the file.
The six forces in this paper are not a list I constructed. They are what emerged independently, across geographies and specialisms that do not talk to each other nearly enough. Organised cargo crime came up in San Diego, in Qatar and across Europe. The talent gap came up in every conversation, whether I asked about it or not. AI was raised by almost every participant without prompting, and almost all of them said the same thing. It is being talked about far more than it is being used well.
What struck me most was how consistent the picture was despite how different the operations were. A TPA in the US handling last-mile freight and a delegated authority in the Middle East managing war risk losses are not usually in the same conversation. They are navigating more common ground than either of them realises.
What I want this paper to do is start that conversation properly. Not as a market report written about claims teams for an underwriting audience. As a document written by claims professionals, for claims professionals, that gives the people doing this work an honest account of what 2026 looks like from the inside.
My hope is that a claims professional reads this and recognises something on every page. And puts it down with a sharper sense of what their operation needs to be able to do in the next twelve months.
The findings are the participants' own words and experiences. I have contributed my own perspective where it is relevant, and where I have, I have put my name to it. But the authority of this paper comes from the people who gave their time and their candour to a conversation this industry has not prioritised enough.
I am grateful to all of them.
Most published research on this market is written for underwriters and brokers. It tracks premium growth, models loss ratios and forecasts market cycles. What it does not do is sit down with the people holding the file and ask what is actually happening.
This paper does that. It is based on in-depth interviews with senior claims and insurance professionals across the United States, Europe and the Middle East, spanning TPAs, delegated authorities, MGAs, carriers, reinsurers, freight operations and specialist brokers. No participant was compensated for taking part. No names and no company names appear in the findings. Participants are identified by role and sector only.
Six forces emerged consistently. They are not a framework imposed on the data. They are what participants described, independently, in their own words.
The forces are not six separate problems to be solved one at a time. They are a converging set of pressures landing on teams that are already stretched. The operations navigating 2026 well are the ones that built data visibility and retained expertise before the pressure arrived.
The six forces
"The forces ultimately pinpoint for us the increasing complexity of the claim and the diminishing ability to rely on traditional ways of managing it. The common thread I see is that the conventional claims model is becoming unsustainable.
Organised cargo crime, fragmented data, talent turnover, litigation pressure, evolving legal exposures and the rapid emergence of AI all point toward the same conclusion. The claims function must become more connected, more data-driven and more investigative, without losing the judgment and accountability that remains at the heart of good claim handling."
The geopolitical dimension
The six forces emerged across all geographies. The interviews also captured something harder to categorise: the speed at which a geopolitical event turns a theoretical claims scenario into a live operational crisis.
A senior claims broker in the UAE described managing a live claim on a feeder vessel struck by a projectile and subsequently sunk, with her client's containers on board, at the precise moment JCC watch list changes were being applied to the region. The insurer's position was that the watch list update created an exclusion. Her position was that the coverage should respond.
Her complaint was not about the incident. It was about the notice:
Why didn't the insurers at least tell us? We are finding ourselves in a situation where we've got our client's claim declined based on this JCC updated list.
Coverage did not change. A watch list did. The broker found out when the claim was declined, which means the exclusion had existed for months inside a document nobody was obliged to send her.
That scenario did not become a force of its own, because the research did not find it universal. It is documented here because it captures something the six forces individually do not convey. External events are now moving faster than the systems claims professionals rely on to manage them, and the data visibility problem in Force 2 and the talent gap in Force 3 both look different when a geopolitical event lands on a team mid-claim without warning.
How the forces interact
Cargo crime, the talent exodus, nuclear verdicts and the Montgomery ruling arrive from outside the claims operation. Data visibility and AI adoption are the internal capability that determines whether a team can meet them.
Several compound directly. Montgomery expands freight broker liability at the exact moment litigation has become the default outcome. The talent gap acts as a multiplier underneath everything, because a green adjuster facing cargo fraud or a hostile jurisdiction is a fundamentally different risk from an experienced one facing the same file. And data blindness prevents the portfolio visibility that would make AI useful and cargo crime patterns detectable before they become expensive.
The operations succeeding in 2026 have built two things that make every force survivable: real-time data visibility and retained expertise. The ones struggling are attempting all six without either.
Organised Cargo Crime Has Gone Industrial
The loss that leaves no scene, and what that does to an operation built to investigate one
The load was collected on time. The paperwork was complete. The carrier's authority checked out. The freight arrived at a warehouse, was signed for, and disappeared.
As a transportation insurance broker with twenty-two years in the market described it:
It's just full trailer loads getting moved to the wrong warehouse, delivered to the wrong warehouse, looks all legit, and that freight is gone.
No forced lock. No cut seal. No scene.
That is the change. Not the volume, though the volume has moved. The crime itself has stopped producing evidence.
What actually changed
Cargo theft is not new. Every experienced practitioner in this research had handled it for years. What they described was a shift in method that has quietly made their existing process obsolete.
A senior claims specialist at a major US freight operation put it plainly:
We've gone from opportunistic thefts where they break into the back of a trailer and steal some freight to organised groups operating overseas using electronic methods, AI, and hiring groups within the continental US to perpetrate these on their behalf.
And the part that matters most for anyone investigating:
Sometimes those groups know they're doing it, sometimes they don't.
The truck that collected the load may belong to a legitimate carrier with a clean record and no idea it is the final link in a fraud chain. The transportation broker described the same pattern from the other side of the transaction:
There are all these authorities that are getting purchased up by bad actors and then they're turning around and double and triple brokering loads. And there's legitimate trucking companies hauling loads that they don't even realize have been double or triple brokered. They're actually the ones moving the product to the thieves.
Twenty-two years in, he had not seen anything like it. Strategic theft rose across the US between 2021 and 2025, and food and beverage is now the largest theft category by incident count.* But the statistics describe the outcome. The practitioners describe the mechanism, and the mechanism is why investigations are failing.
Verisk Analytics / CargoNet: Supply Chain Risk Incident Frequency, January to December 2025. Data updated March 2026.
No vendors sponsored this research. Platforms are cited solely because participants raised them organically.
The investigation arrives at nothing
An investigation designed for physical theft begins at the scene. Somebody inspects the trailer. Somebody pulls the yard footage. Somebody interviews the driver who found the lock cut.
None of that exists here.
By the time the loss is reported, the entity that collected the cargo has dissolved. The phone number was live for one load. The carrier identity was constructed for one load. There is nothing to photograph and nobody to interview, because the person who signed for the freight was doing his job.
So the investigation starts late, on paper, chasing a company that no longer exists.
That has a price, and it does not appear under theft. Allocated loss adjustment expense climbs, because external investigators spend weeks reconstructing a documentation trail built to survive exactly this kind of review. Reserves cannot be set accurately, because liability cannot be assessed until the brokering chain is mapped, and mapping it takes months. Claims that should have closed inside thirty days are open at six. And legitimate cargo owners sit in extended dispute, not because their claim is complex, but because the investigation arrived too thin to tell their loss apart from a fraudulent one.
The loss is the freight. The cost is everything that happens afterwards.
Somebody is telling them what is on the trailer
Practitioners kept arriving at the same uncomfortable conclusion without prompting. These loads are not being selected at random.
There's obviously intel that's going on at the shipper, somewhere at that level, where they know what's on that trailer and they're directing it to the wrong place.
The broker was careful about how far he would go. He had no documented case originating inside a transportation company. But the targeting is too precise to be anything else, and he named the coverage consequence, which nobody in this market seems to be discussing:
A lot of our cargo policies have employee theft and dishonesty exclusion. So you've got to be buying that back or you've got to be buying a separate crime policy to cover employee theft.
Read those two statements together.
The industry is fairly confident these losses begin with someone on the inside. Most cargo policies exclude exactly that. The mechanism practitioners describe most often is the one the wording was drafted to avoid, and almost nobody is buying the cover that would respond to it.
Underwriters have already reacted in one place
The load board exclusion is now appearing directly in cargo wordings on freight broker business. Not as an endorsement. In the body of the policy.
It's an exclusion that's being written into the coverage. Basically says, do not do that, because those are the loads that are going to get targeted.
It is worth noticing what that admits. The market has concluded that an ordinary, everyday commercial practice is now so reliably a route to loss that it is cheaper to exclude it than to price it.
The proportion nobody says out loud
Strategic theft dominates the conversation. It does not dominate the desk.
Financially, yes, strategic cargo theft is the biggest loss indicator we have. But if you're talking about volume, I would say well over two thirds, three quarters of my claims are still due to the traditional loss or damage claims.
An adjuster's day is still mostly wet cargo, crush damage and shortage. Strategic theft is a small number of files consuming a disproportionate share of the loss fund and almost all of the investigative capacity.
Which is precisely why it keeps getting handled inside a process built for the other three quarters.
Where the response is actually working
Two things shifted in the last year and both came up unprompted.
The first is regulatory. The FMCSA transition from SAFER to MODIS has made it materially harder to buy an established DOT authority with a clean history, which was the standard workaround for anyone needing a track record they had not earned. Practitioners describe it as too new to judge but pointed in the right direction.
The second is intelligence rather than investigation. A senior freight claims professional described using a supply chain visibility platform not to find one missing truck, but to connect several:
They've got the cameras all over the country, so not only do they provide a live view of where trucks are located, but if a truck goes missing you can send that in and it's like having eyes on for a BOLO. And then aside from that, you can back engineer that.
The outcome was not recovery of that shipment. It was better:
We were able to determine that it was the same group.
That sentence is the whole argument. One file tells you a load was stolen. Several files, cross-referenced, tell you who is stealing them and where they will be next week.
"Speed decides whether you recover anything, and most operations are not built for it. If you don't keep the trail hot within days, by the end of the week, there is no point looking any further. Within a week the cargo could be absolutely anywhere. The evidence we need most is also the evidence clients give us last. Take statements while it is fresh and you learn how they are doing it and sometimes what they will try next. Take them six months later and you get an account that has hardened into the minimum information necessary to get the claim paid.
There is a structural reason this keeps happening. In most operations there is not one person dedicated to figuring out theft losses. That task gets attached to an operations manager who already has a day job, and it is never the most important thing on his desk that morning.
This is not a new type of loss. Fictitious pickup has always existed. What has changed is the speed at which the schemes evolve. Criminals adapt faster than we do. The operations that stay ahead are not the ones with the best technology. They are the ones that keep changing their own process and refuse to let it settle. A process that stops evolving becomes a pattern. Patterns are exactly what organised groups are looking for."
Fraud detection is a first notice function, not a post-intake one. The adjuster has authority to mobilise a specialist investigation the moment a fictitious pickup is suspected, without waiting for triage. The clock starts when the loss is reported, not when the file is allocated.
Cargo crime is treated as intelligence, not claims. Incidents are cross-referenced across the portfolio. The same constructed identity, the same corridor, the same documentation structure. Invisible in one file, obvious across twenty. That intelligence goes back to underwriting and carrier approval.
The variable is not budget. It is authority and pace. The operations that get something back are the ones where the adjuster does not have to ask permission to move.
Take the last three cargo theft losses your operation handled. How many hours passed between first notice and somebody being instructed to investigate? Not allocated. Instructed.
And has anyone compared those three files against each other?
If the answer to the second question is no, the operation is handling cargo crime one claim at a time. That is not a cargo crime problem. It is the subject of the next force.
* Verisk Analytics / CargoNet: Supply Chain Risk Incident Frequency, January to December 2025. Data updated March 2026. No vendors sponsored this research. Platforms are cited solely because participants raised them organically.
Claims Teams Still Cannot See Their Own Book in Real Time
Data blindness as the structural problem underneath every other force, and the most fixable one
A reinsurance executive joined a firm with a substantial book in Turkey. In his first weeks he asked what should have been a simple question.
How much risk do we have in Turkey? And we didn't know. And then on top of that, we didn't know how much premium we had coming out of Turkey. So it was really hard to gauge how much reinsurance do we need in this one area and how much are we willing to pay for it. Because we didn't have any data. It was just kind of like hopeful best.
That is a reinsurer. Not a small MGA. And the consequence is not reporting inconvenience. It is that the business could not work out how much protection to buy.
The most dangerous vulnerability in a claims operation is not what the team does not know. It is how long it takes them to find out.
The problem starts before the file arrives
Ask a claims professional in any market to design their ideal operation and almost nobody describes technology or headcount. They describe a claim that arrives complete.
A claims manager at a US carrier writing motor truck cargo:
It's very important on the front end to have qualified people to ask the right questions, to get the right information, to get started off on the right foot. Some of these that come in through our portal could be more detailed.
A claims director at a major carrier's inland marine team:
Sometimes there might be a delay from when I received the claim to when the documents the insured sent in got to the file. So by the time I'm contacting them, I'm asking them for something that they've already sent in.
A claims specialist inside a major freight operation, describing what she would change first:
Instead of just crush damage, okay, how many are crushed? Where was it identified? Who identified it? Who do I reach out to at the customer?
An aviation reinsurance broker in the Middle East, arriving at the same answer from an entirely different market:
Providing a checklist, perhaps. You get a claim that's under this policy, boom, boom, these are the documents you need to have.
This is the most consistent finding in the research. Different countries, different lines, different sizes of operation, same answer. The industry's biggest operational problem happens before the file reaches anyone qualified to handle it.
Everything downstream inherits that. A portfolio cannot be queried on cause of loss if cause of loss was never captured properly. The data is not missing because the technology is expensive. It is missing because capturing it was nobody's primary job, so it became nobody's job at all.
The lag nobody has priced
There is a second gap before the file even reaches intake, and one MGA claims leader has put a name to it.
One of our biggest is what we like to call lag time. The space between the occurrence of the incident and when it's reported as a claim.
He explained why it happens by walking through it:
Imagine if you will. I'm a landscaping company. My guy calls me and tells me he got into an accident. He's got three lawns to mow and it's like eight in the morning and that's money in my pocket. So I tell him, exchange information, get it all down, and when you come in we'll do an incident report and then we'll report it to the insurance company at the end of the day. Which is not horrible, but the problem is that the guy gets through his day, he's exhausted, he blows it off. Goes back to work, does his next four lawns, and then by the time the boss corrals him back in, you're two, three days down the road. And the claimant's already called it in. And now we're asking them, why haven't you reported this claim?
Lag time is not carelessness. It is a rational decision by somebody who is paid to mow lawns, not to file claims. Every operation treating it as an insured behaviour problem is trying to fix the wrong thing.
By the time the file opens, the claimant has been in contact with the other side first, and the operation is already reacting.
The cost is distributed, which is why nobody owns it
Data blindness never appears on a single line. It shows up everywhere else.
A claims team without real-time theft pattern data cannot see that three losses in one month share a carrier identity, a corridor and a documentation structure. Each claim is handled individually. The criminal operation continues. The cost is not the claims that were paid. It is every claim that followed because the pattern was invisible.
In a hostile jurisdiction, data blindness means concentration risk is identified after a file is already in litigation. A team that cannot quickly tell underwriting which transportation risks sit where cannot have the conversation about reserve adequacy or programme limits until it is too late to matter.
The pricing consequence is more insidious. A claims lead at a Gulf insurer described it as a market condition rather than a failure:
Sometimes it's not the technical price who is ruling, it's the market price, the competitive price.
And then the trap:
But if you follow that strategy, maybe you're out of the business.
Underwriters are not mispricing risk because they are bad at maths. They are mispricing it because claims cannot prove them right in time. The technically justified rate loses to the competitive rate every time the evidence arrives after the quote.
A director at a US captive described what happens next to the buyer who took the cheaper option:
They receive mediocre, if not less than mediocre claims management. The claims escalate, fires are happening and the reserves are outrageous.
Size does not solve this
The same reinsurance executive described his partners:
Two of our partners are twice the size of us, or one's even four times the size, and they have no systems. It's all still on Excel. They don't know how much aggregation they have in different geographical regions. They don't know what their claims loss looks like.
His own firm was the same until a year earlier, and was the first in the group to fix it.
The smallest firm in the group was the first to know what it owned. Size does not produce visibility. It usually delays it, because there is more to unpick and more people who have learned to work without it.
He also described what the absence costs in senior time:
There was a lot of, hey, wait, why did we pay this claim? And then we have to get them on the phone and then we have to set up a Zoom and then we all have to be available for the Zoom. And next thing you know, you're a month down the line and now you're talking about it and it's like, what's the point talking now? We can't even go back and change it.
A month of executive attention, to answer a question a notes field would have answered in a sentence.
One disagreement worth reporting
The industry has broadly settled on the view that claims and underwriting should sit closer together, and the strongest advocate in this research put it as forcefully as anyone:
There's no distance between underwriting and claims. It's not like the production side is glorified and claims is in the back office. That's the exact opposite of what we are. Claims has the leverage to affect 60 cents of every dollar.
But not everyone agrees, and the dissent is deliberate rather than accidental. A claims director at a US captive:
I like the fact that there's a little bit of silo between claims and underwriting, because I'm not managing my claim based on trying to get you the cheapest rate in your premium. I am managing my claim based on the facts that are presented to me.
A carrier claims manager said something similar, describing limited collaboration because his team wants to control the coverage decision.
The disagreement is not about whether data should flow. Everyone agrees it should. It is about whether influence should flow back the other way. If claims sits close enough to underwriting to feel the renewal, it sits close enough to be affected by it.
"The conversation I have most often with clients about their own portfolio goes like this. How many claims do you have where the cause of loss is X, in this country? I don't know, I'll have to come back to you. Cause of loss with a certain commodity, in a certain region, with a certain carrier, where you have not been able to recover because of limitations in the bill of lading or time bars? I don't know, I'll have to come back to you.
And when the answer does arrive, it arrives with a caveat. It is looking like this at the moment. Take it with a grain of salt. It could be a bit different when things are said and done. That language has become so normal nobody notices it. It is an admission that the data does not give the granularity required, and decisions on renewals, reserves and coverage are being made without it.
The timing is what does the damage. In the actual middle, where you are living, where everything is live and you need to decide whether to renew this account or keep this coverage, the information is almost always too late to react to.
When a new client comes to us the state of their claims data almost always leaves a lot to be desired. That is not incompetence. TPAs are under pressure to process the claim, pay it, recover, send the report. Data quality is what gets sacrificed to throughput. The fix does not start with a new system. It starts at first notice of loss. Lock the fields before the file advances and make that non-negotiable. The data is there in every operation I have ever worked with. The commitment to capturing it at the right moment is what is missing."
The operations closest to this have solved it at the source. No file advances past day one without the structured fields locked in. Cause of loss. Commodity. Carrier identity. Region. Time bar status. Bill of lading reference. Not optional fields. A hard gate.
The monthly data audit that many operations use as their primary control is a backstop, not a solution. By the time it surfaces a missing carrier identity, the cargo is gone.
Questions every claims operation should be able to answer right now.
If any of these takes more than two minutes, the data infrastructure is not managing the book. It is being managed by it.
Pick one of those six questions and ask someone to answer it today.
Note two things: how long it takes, and whether the answer arrives with a caveat attached.
The Talent Exodus Has Not Stabilised
What the post-Covid workforce gap looks like from inside a claims operation, told by the people still standing
Two claims professionals in this research, working in different sectors on different continents, described the same moment without any knowledge of each other.
A senior professional in US excess and specialty:
Before Covid everybody was there. Everybody's in office five days a week. When Covid hit, a lot of insurance companies either did layoffs or they did early retirement opportunities. So a lot of that knowledge, a lot of that expertise, people chose to retire. They either took the retirement, went to other carriers, or just got out of the industry altogether. So it hasn't stabilised since.
A director at a US captive:
We lost a lot of talent and we needed bodies to fill seats. And I don't think that we had access to talent that was capable of doing what 25 years of experience offers you. You look at a situation with 25 years experience, you know exactly what to do.
The industry did not lose headcount. Headcount recovered. It lost the ability to recognise a bad file on sight, and that is not a line item anybody reports.
Nobody chose 300 files
The most useful sentence anyone offered on caseload was not about caseload at all.
A lot of where companies are now in the high claim counts is not so much to do with company philosophy and a lot to do with the struggle of employee retention.
She then described the mechanism, from the adjuster's side:
What you paid pre-Covid and what you pay now, people are jumping from company to company because the pay increases are so large. And it's like, if I'm gonna have 300 claims, I'd rather go here and get paid 15,000 more to get 300 claims than stay here.
Read that carefully, because it reverses the usual framing.
Nobody chose 300 files. No operation sat down and decided that was the right number. It is what is left over when you cannot keep people, and it is now being managed as though it were a policy.
The caseload figures across this research vary enormously by complexity, and the spread itself is the finding.
| Sector and role | Sustainable open files |
|---|---|
| Aviation reinsurance broker, complex technical claims | 12 to 20 |
| Inland marine claims director, major carrier | 30 to 40 |
| Commercial auto MGA, in-house claims | 40 to 60 |
| Cargo adjuster, US TPA | 75 to 80 |
| Excess and specialty, most complex files | cap at 110 |
| Marine and Jones Act, international insurer | 125, difficult above 135 |
| Excess and specialty, mid complexity | 125 to 130 |
| Railroad claims agent | 120 per year, 170 was "fighting fires" |
What the gap actually costs
It does not show up on a headcount report. It shows up in the space between a technically correct decision and the right one.
An adjuster with two years of experience and 130 open files is not making bad decisions. They are making decisions without the experiential context that tells them which files need attention today, which clients need to hear from now rather than Friday, and which coverage positions are worth pushing back on.
High junior turnover compounds it. When a complex file is handed off three times in two years, the defence strategy fragments. The new adjuster inherits a file without the memory of why decisions were made, what was already said to the client, and what the other side has already been told. Plaintiff counsel is structured to exploit exactly that discontinuity, and the cost shows up in settlements rather than in HR reporting.
The market stopped training and started recruiting
A recently promoted claims director at a major carrier, asked about training:
You kind of have to fend for yourself. I've built relationships with more senior employees who have given me guidance whenever I have questions. But I don't think there's ever really a set training program when someone comes into the department. And I think that could be better.
Another participant described the two ends of the same market. One large carrier is known as the place that teaches you the job properly. The one she works for now hires people who already know it.
That works for any individual carrier. It fails for the industry, because the firms that still train are now subsidising the ones that do not, and there are fewer of them every year.
The remote working bind
Claims is learned by proximity. You hear how a senior adjuster handles a difficult coverage conversation. You see how they read a policy for subrogation potential. You absorb their judgment about when to push and when to hold.
Being in a claims office with other individuals to mentor you, to provide you on the spot answers, to hear other scenarios being discussed, it's invaluable. Working from home has a huge impact and part to play in that.
She was clear it is not an age problem:
It's not always age, because actually we're having a hard time getting the younger talent fresh out of college and training them up. I think it's mindset, entitlement, lack of mentorship and resource.
And employers are caught:
It's really hard for employers to not offer work from home right away, because that's what people are looking for. You need talented, experienced individuals, but they give you some hard ultimatums.
The most useful reframe came from a practitioner who splits it by career stage rather than by policy:
For somebody just starting out, I think office culture is a requirement. You don't just learn from a book. Now for somebody who's been doing it for 10, 15, 20 years, you could work from home and be fine.
If juniors are required in the office and seniors have earned the right to work from home, the office fills with people who need training and empties of the people who can provide it. That paradox does not resolve through attendance policy.
The barrier nobody mentions
One participant described spending years proving she could handle work she was already capable of, because of a perception that claims skills do not transfer between lines.
I've had to take some lateral moves and I've had to volunteer for things to show that I could do the job before you would give me the job. So it's not necessarily easy to move around as it technically should be.
The industry says it cannot find experienced claims people. It also makes experienced claims people prove themselves twice before letting them move between lines. Both are true, and only one gets discussed.
"There is a second pattern underneath the exits that gets less attention. A growing number of junior professionals are cycling through roles every six to twelve months, moving for incremental salary without staying long enough to build a base. Six months here, a year there. They are not really learning. And the financial consequence is identical to outright turnover: fragmented files, broken client relationships, and a portfolio where the institutional knowledge needed to manage complex losses does not exist at the adjuster level.
I also see the shortage most acutely in a specific bracket, the professionals who would normally be stepping into senior roles about now. When we hire, that is the hardest seat in the market to fill.
The industry has been treating claims as an entry point into insurance rather than a technical discipline in its own right. The result is a pipeline of people who spend two or three years in claims and then move into underwriting or broking, because that is where the status and the salary progression are seen to be. Fixing that means compensation that reflects the complexity of the work, progression that does not require leaving claims to advance, and leadership that understands the difference between a team that is coping and one that is performing. That difference is worth millions in loss ratio terms."
Caseload is a strategic variable, not a capacity constraint. One MGA in this research runs its in-house team at 40 to 60 pending files, deliberately, and treats it as a competitive weapon:
If you're going to staff that low, then you're going to have to be very demanding of your people who now have the time to do what they need to do to bring these claims to closure.
It is also the only operation in the research that volunteered its reserve adequacy figures unprompted: 100% at 30 days on both physical damage and third party property damage liability. Those two facts are the same fact.
Remote working is earned rather than assumed. Not through attendance policy, but through demonstrated competence, with senior presence structured around mentorship rather than personal file work.
Claims is compensated as a specialism. The operations with the lowest junior turnover are not the ones with the most perks. They are the ones where a junior adjuster understands that what they are learning is genuinely hard to learn anywhere else.
What is the median open file count per adjuster on your team right now?
And more importantly: did anyone decide it, or is it simply what is left over?
Nuclear Verdicts Have Made Litigation the Default
How the litigation environment changed at the small end, not the large one
The published conversation about nuclear verdicts is about severity. The change practitioners describe is frequency.
A transportation insurance broker with twenty-two years in the market:
Every single accident it seems like has attorney involvement. It does not matter the amount of severity, how severe it is, the amount of property damage. There is attorney involvement in almost every single accident with a truck right now. So that right there is driving claims costs on every single claim.
It is not that the large claims got larger. It is that the small ones now arrive with counsel attached.
He explained why in one sentence that summarises the entire environment:
I feel like the truck is inherently guilty until proven innocent. That is the way our world works right now.
The severity numbers, second
The median nuclear verdict in US transportation cases rose from .* Third-party litigation funding has removed the plaintiff's traditional incentive to settle, because the cost of a multi-year battle is now underwritten by someone else.
American Transportation Research Institute and Insurance Research Council, median verdicts in US transportation cases 2020 to 2024.
Those numbers matter. But they describe the tail. The frequency shift is what has changed the economics of an ordinary week.
Every single accident it seems like has attorney involvement. It does not matter the amount of severity, how severe it is, the amount of property damage.
Those numbers matter. But they describe the tail. The frequency shift is what has changed the economics of an ordinary week, and it is why a claim that might once have been reserved at $500,000 now carries a realistic pathway to policy limits.
The conditioning is not accidental. The broker again:
If you go through any big city, just look at the billboards. There are attorneys everywhere saying call me before you work directly with an insurance company. Everybody sees the big dollar cases. So that is what we've all been conditioned to hear.
What a jury walks in with
A senior in-house claims professional at a major US transportation operator described running mock trials specifically to test whether his team was arguing the right things.
The jury's just like, $5 million. Did you want to hear the evidence? Yeah, we'll listen, but we're already coming in with $5 million in our head.
The mock juries were not kind:
They burned us at the stake. And you take that info and it's like, you better use it for the good, because it's like, hey, we need to put some money on this case, make it go away. Well, we've got a strong case. Not really.
Then, in seven words:
If the jury hates you, you don't have a good case.
That is the environment, described by someone who paid to find out.
Which leads to the thing everyone in this research described and nobody quite says.
Very few insurers are still trying to win these files. They are trying to leave them. The strategy every practitioner described as getting out fast is not a strategy. It is an acknowledgment that the venue decides the outcome before the evidence is heard.
The plaintiff bar industrialised first
In Jones Act and maritime liability, the approach has gone further. A claims adjuster at an international marine insurer described medical build-up networks operating in specific jurisdictions, directing injured parties toward surgery and treatment designed to inflate claim value long before litigation begins.
It's definitely attorney driven, exaggerated. Like go get that surgery. Put this neck brace on.
Specific firms were named by participants. The pattern is not unique to any one of them. It is a market condition.
The courtroom method has a name in plaintiff litigation circles. The Reptile Theory frames the defendant's conduct as a threat to community safety, activating a jury's protective instincts before a single exhibit is shown. Transportation cases are among its most effective applications.
What a thin defence file actually costs
Operations still applying a pre-2024 approach to serious losses are arriving at litigation without the evidence that would have constrained the plaintiff narrative had it been collected in the first hours.
A thin file has a specific profile. Telematics overwritten before anyone asked for it. Dashcam footage not pulled the same day and no longer available. Driver interviews conducted weeks later, by which point the plaintiff's attorney has already taken a recorded statement and shaped the account. Radio records, maintenance logs and inspection histories never assembled into anything coherent.
A thin defence file does not increase the risk of settling. It changes what you are negotiating. You are no longer arguing about liability. You are arguing about the price of surrender.
Against the last serious injury file your operation opened, tick each of the four that happened.
You are no longer arguing about liability. You are arguing about the price of surrender.
The countermeasure almost nobody discussed
Only one participant offered something that changes the outcome rather than documenting it.
What I've really been encouraging everybody to do is put those cameras in the trucks, because now you have an independent eyewitness of what actually happened. Both forward facing, side view mirrors, not necessarily driver facing. You've got the defence attorneys telling them to give it to the officer who's writing the police report. Because now we don't have a witness that didn't really remember what happened, or word against word. The video just shows what happened.
Evidence preservation is a claims discipline. Evidence creation is a risk management one. In this environment the second is worth more than the first, because it exists before anyone knows they will need it.
He added a second variable that almost never appears in a buying decision:
You're better off going with an insurance company that works with trucking companies, because it's going to have a very good defence panel in every different state that is focused on defence for trucking companies.
Buyers compare price and limits. The variable that decides the outcome of a serious file is whose defence counsel picks up in that state, and almost nobody asks about it at renewal.
"There is a reporting problem inside this that nobody has fixed. Legal costs and potential settlement are routinely combined into one reserve figure and presented to underwriters and clients as a single number. It is not deliberate. It is the product of systems built before litigation funding changed the economics.
But a two million dollar reserve tells a client nothing. One million of settlement with one million of defence cost is a completely different file from a two million settlement with costs still running, and those two files require opposite decisions. The moment that distinction matters most is the monthly conversation about whether to settle or hold, and that is exactly when the reporting is too blunt to help.
What most people outside the claims room do not understand is what happens at board level when a nuclear verdict reserve lands on a balance sheet. The number is not just a liability. It is a question about the organisation's risk culture. Did we know this was coming? Why didn't we act sooner? The claims professional who can walk into that conversation with a documented strategy, a monthly reviewed reserve and a clear account of why the file looks the way it does is the one who survives it. The one who cannot is the one who reserved conservatively to avoid a difficult conversation and is now having a much worse one."
The operations ahead of this stopped treating litigation as the outcome of a failed negotiation and started treating it as a parallel track that begins at first notice. Every serious file in a hostile jurisdiction has a litigation strategy from day one, even when settlement remains the goal.
- Reserve the worst realistic outcome from day one. Not a comfortable middle number.
- Separate legal cost from potential settlement in every report. They are different problems requiring different decisions.
- Build a likelihood spectrum. A single number tells you nothing about probability.
- Review monthly, not when new information arrives. Information in these files arrives in increments that individually look insignificant and collectively change the picture.
- Never let the reserve conversation happen separately from the strategy conversation. They are the same conversation.
Think about the last serious injury file your operation opened.
How many hours passed before somebody preserved the telematics? And had the plaintiff's counsel already spoken to the driver?
* American Transportation Research Institute and Insurance Research Council, median verdicts in US transportation cases 2020 to 2024. The Reptile Theory is referenced as an established plaintiff strategy. It was not named by participants but describes the approach they documented.
Montgomery v. Caribe Transport
What actually changed, and what had already changed before the Court ruled
A note before this force begins. Montgomery is a United States decision with no direct application outside it, and this is the one force in this paper that is jurisdictionally bounded rather than universal. It is included because the exposure it creates runs through freight brokers into international cargo movements, and because the principle underneath it, that documented vetting has replaced legal protection, is being adopted well beyond the country that ruled on it.
It is also worth being clear about where the case came from. Montgomery was a bodily injury and safety claim, not a cargo claim. The negligent selection argument reached the Supreme Court through personal injury litigation arising from a road accident. Whether the same reasoning extends into cargo negligent selection is an open question, and one this force returns to at the end.
On the Supreme Court ruled unanimously in Montgomery v. Caribe Transport, eliminating the FAAAA pre-emption defence that freight brokers had relied on across multiple circuits to deflect negligent carrier selection claims. Justice Barrett wrote for the Court. Within four days the Fourth Circuit vacated a summary judgment entered on exactly that ground.
Montgomery v. Caribe Transport, US Supreme Court, decided 14 May 2026, unanimous, Justice Barrett writing.
Montgomery was a bodily injury and safety claim, not a cargo claim.
The headlines described a transformation. The one practitioner in this research who insures freight brokers for a living described something more specific.
A lot of the freight brokers I insure were already buying insurance for exactly this scenario, where they could get named in a big auto liability suit.
And on vetting:
There's already those third party vetting services that were out there. A lot of the insurance companies I work with on the freight brokers already wanted them to be using one. This is all pre-Montgomery. So this stuff was already out there.
The Court removed a defence that most of the market had already stopped relying on. What changed is not the exposure. It is the price of it, and who now gets to read the file.
What the ruling does
Before Montgomery, brokers in circuits that accepted the pre-emption argument could move to dismiss negligent selection claims on the grounds that federal statute pre-empted state tort law. Montgomery closes that door permanently and uniformly across all fifty states.
Every US freight broker is now directly in the liability chain for the carriers they select. The protection available post-Montgomery is not legal. It is a documented vetting process, and the practitioner reading of the Court's commentary is that this was the point:
As long as you can document that you're doing this, you should be good.
That is a practitioner's interpretation, not a holding. But it is how the market is behaving.
The consequence is priced, not litigated
The most predictable effect of Montgomery is not a wave of verdicts. It is a wave of capacity.
What I think that case is going to do is just increase litigation on the freight brokers, which is going to increase the premiums on the freight brokers. So they were getting that insurance for pennies on the dollar. It's going to start costing more.
And then:
There just are not a ton of options for freight brokers, especially before Montgomery. But post-Montgomery now there's a lot more premium there, so there's going to probably be more programs.
Brokers who were buying this cover for very little are about to find out what it is worth, and new programmes will arrive to sell it to them.
The exposure landed on somebody else
Montgomery is discussed as a freight broker problem. In practice the pressure has been transmitted straight down to carriers.
The motor carriers that have poor DOT scores or high accident history, lots of violations, they're the ones who are having the biggest issue, and you don't necessarily just fix that right away.
I've talked with clients that are like, yeah, we can't get loads from that broker now. Or we're having to get additional screening and answer additional questions about specific violations and what we've done to address them.
And the timing problem:
The way those scores are set up, you can't just erase them. Those scores stay what they are for a year or two years at a time.
Montgomery did not create a liability problem for freight brokers. It created an access problem for carriers, and the carriers least able to fix it are the ones already at the bottom of the market.
What claims professionals inherit
Carrier vetting documentation is now a day one investigative requirement on any loss where a freight broker sits in the chain. Not whether the broker used a vetting service, but whether they followed it, on this load, with this carrier, on this date.
The double brokering documented in Force 1 compounds it. Where a load has passed to a second broker without the shipper's knowledge, the liability chain extends further up, and mapping it is now a standard step rather than an escalation.
There is also a speed problem that will make all of this harder.
When they are dealing with a high dollar value claim with one of their shipping customers, the shipper is demanding that this gets handled right away. So the broker is doing whatever they need to do, hammering the adjuster, what do we need, what else do you need. So it's just how fast can we push this claim through?
The speed at which a high-value freight claim moves from first notice to coverage position is often the speed at which the vetting file is never properly examined.
The variable nobody publishes
Something else has started happening in vetting, and it points in an unexpected direction.
You may want to look at talking to a third-party vetting service and saying anybody with that particular insurance is a red flag for us, because look at how slow they are at processing this claim. It's 30 or 60 days, and now you've got your customer banging on your door.
Every experienced claims professional keeps an informal list of which insurers actually answer. Nobody publishes it. Everybody has one. And it has now started to affect carrier selection, which means an insurer's claims service has quietly become an underwriting characteristic of the risks it gets shown.
"The strategy does not fundamentally change. Get out of the suit as fast as possible was true before Montgomery and it remains true after it. Very rarely does anyone fight one all the way through, and the reasons for that are in the previous force.
What changes is the investigation that informs the strategy. You need to understand the vetting process far more deeply than you did before. Not just did they use a vetting service, but did they follow it on this load, with this carrier, on this date. That is a different question. And if the answer is no, you need to know that before working out the next steps, not after you have taken a position.
The ruling is a few months old. Whether it extends from bodily injury into cargo negligent selection will take a year or two of litigation to become clear. What is already clear is that a claims professional walking into a freight broker liability file without a complete vetting record is walking into a room where the insurer's best option is to settle immediately, and the plaintiff's counsel already knows it."
On the last freight broker file you handled, how long did it take to get the complete vetting record?
And did you communicate a coverage position before it arrived?
Montgomery v. Caribe Transport, US Supreme Court, decided 14 May 2026, unanimous, Justice Barrett writing.
AI Is Everywhere in the Conversation, Nowhere in the Operation
What claims professionals are actually doing with AI, and what the market is selling them
Every conversation in this research touched on AI. Most participants raised it without being asked.
Not one described an AI setting a reserve, deciding coverage, or conducting an investigation.
Not one.
What it is actually doing
Across six countries and every size of operation, the description was remarkably consistent.
A senior professional in excess and specialty:
It's more of like a resource, kind of like a secretary of sorts. Something that can collate your documents for you and put them in a certain way. But the adjusting itself is still the adjuster.
A claims director in inland marine:
I use it a lot of times for status letters or declination letters. Put in a mock-up of the letter and have it write it for me.
A carrier claims manager, on the one place it earns its keep:
The way we receive those, we don't get all the information we need. So AI has been quite the enhancement of helping us get what we need to get started on these claims on the right foot from the start.
A claims lead at a Gulf insurer, on grey-area coverage:
Sometimes in insurance you have grey areas where you are not sure, is it covered, not covered. So I go through a discussion with AI. I put the wording in to have more understanding about the interpretation.
An MGA claims leader, on contesting a declination:
I may ask AI to review that section of the policy and summarise it for me, giving the examiner the highlights. Once we've gone over it, then we will prepare something to go back to the carrier.
Note what every one of those has in common. The output is reviewed by a human before it influences anything.
The market is selling autonomous claims handling. Most claims teams have bought a very expensive drafting assistant. That is not a criticism of the tool. It is a criticism of the pitch.
Why the gap exists
Two reasons, and both came directly from practitioners.
The first is the subject of Force 2. A system asked to analyse loss patterns across a portfolio is only as capable as the data it can reach. In operations running quarterly loss runs and disconnected systems, that capability is close to zero. Data visibility is not adjacent to the AI problem. It is the precondition for it.
The second is accountability. A coverage decision, a reserve figure, a settlement position: these carry professional and sometimes legal consequences for the person who makes them. Handing that to a system that cannot be cross-examined in a coverage dispute is a risk most experienced practitioners are not yet willing to take. That is not conservatism. It is professional judgment.
The tell
One MGA claims leader described something every experienced practitioner in this research recognised immediately.
You can kind of identify AI tone. It has a certain tenor to it. So you can see where many of the summaries that the adjusters are now using have probably been put through a copilot or a chat.
Every experienced adjuster can now tell when a submission was drafted by a machine. None of them said so to the sender. The tell has become a routine part of reading correspondence, and nobody has decided what to do about it.
The failure mode is already here
The most documented problem appeared on the other side of the file first.
People cite cases that don't exist. They just put it in AI and they're not checking. So that's probably the gap.
Independently, from a different sector:
You're seeing where people are using AI to do things and they're not necessarily looking at what's being sent out. Take the issue in New York where attorneys got sanctioned for having AI write their motion.
Two practitioners, no connection, same failure.
The operational consequence is a verification burden that did not exist two years ago. Legal arguments that look authoritative but rest on fictional precedent create exactly the work they appear to save. Every citation from an AI-drafted submission now needs independent checking, and in operations already running above sustainable caseload, that is not a minor addition.
"There is something happening in correspondence that nobody has named yet. Several large carriers and brokers have started adding AI addresses to claims correspondence. Which means every letter an adjuster writes is now being read by a system that does not forget, on behalf of a counterparty.
Over hundreds of claims, that system learns how you handle files. Your response times. Whether your document requests are consistent. How quickly you move toward settlement under pressure, and which arguments you eventually stop resisting. That is possibly a good thing. It can also be used against you.
And the AI reading your correspondence may not be neutral. It may have been given information that favours the broker's position over what is actually in the claim documents. An adjuster receiving a confident, detailed AI-generated coverage argument cannot assume the system was trained to find the correct answer. It may have been trained to find the answer that serves the party operating it.
The risk is what we do in response. Rushing communications to avoid triggering a timing flag. Moderating document requests to avoid looking difficult. Both produce worse outcomes on the claim. What I keep coming back to is not whether AI will change claims. It will. It is whether we let it change claims in a way that produces better outcomes for the people at the end of the claim, or one that produces faster throughput and thinner expertise. My fear is that we end up with AI on both sides of a file, arguing with each other, while the humans debate whose system is smarter rather than whether the claim is covered."
The gap that matters most
A green adjuster who uses AI to draft their letters, review their policies and structure their coverage arguments is not developing the judgment that doing those things manually would produce. The adjuster learns to manage the AI rather than the claim.
One practitioner put the problem in a single line: if a new adjuster uses AI from day one, they may never develop the baseline competence to know when the AI is wrong.
In claims, knowing when the AI is wrong is not an edge case. It is the job.
That judgment is built by doing the work the machine is now doing, and nobody has worked out how to develop it any other way.
Ten claims tasks, all taken from the framework below. Place each one, then see where the research puts it.
| Task | You placed it | The research puts it |
|---|
- Appropriate: first-pass summarisation of large files, flagging missing documentation at intake, coverage summary drafts for human review, routine status letters, meeting notes.
- Human required: coverage decisions, reserve setting, settlement positions, client communication on complex or sensitive losses, any task where the adjuster's judgment is the output.
- Never without verification: legal research and case citation. Every citation from an AI-generated submission checked independently before it influences anything.
- New employees: restricted access for a defined period. The judgment required to use AI well is developed by doing the work without it first.
- Correspondence: treat every written communication as data that may be profiled by the counterparty's system. Deliberate, not reactive.
Ask the newest adjuster on your team to draft a coverage position without using any AI tool.
Then read it. That document tells you what you are actually building.
Sedgwick State of the Claims Industry report, 2025: approximately two thirds of carriers report a gap between their AI vision and operational reality.
What the Six Forces Tell Us
Six forces. Not separate problems to solve in sequence, but pressures that arrive simultaneously and compound in ways individual solutions cannot address.
Cargo crime became harder to investigate at exactly the moment the experienced investigators left. Litigation became the default outcome at exactly the moment freight broker liability expanded to expose a new class of defendant. AI is being marketed as the answer to a talent gap it may actually deepen. And all of it is happening inside operations that cannot see their own portfolio clearly enough to know where the exposure sits.
That picture did not come from any single conversation. It came from the accumulation of conversations across geographies and specialisms that rarely talk to each other, each arriving at the same reality from a different direction. These professionals are not reading the same reports. They are experiencing the same conditions.
The two variables
Across all six forces, two characteristics separate the operations that are ahead from the ones that are behind. Neither is the obvious one.
The first is retained expertise. Not headcount. Not technology. The experiential judgment that comes from handling complex claims over years, in conditions that cannot be simulated in training or transferred through documentation. The operations navigating 2026 well have it. The ones that do not are discovering, file by file, what its absence costs.
The second is data visibility. The ability to answer a specific question about a specific risk in a specific geography within minutes, and use that to shape decisions rather than react to events. The data exists in most operations. What is missing is discipline at intake and investment in surfacing it.
These two are connected in a way the structure of this paper does not fully capture. Expertise tells you what to ask of the data. Data gives experienced judgment something to work on. An operation with one and not the other is performing below its potential. An operation with neither is managing a pile of individual files and absorbing losses it should have seen coming.
How they compound
The talent gap (Force 3) makes every other force land harder. A green adjuster navigating a fictitious pickup (Force 1) without pattern recognition is a different proposition from an experienced one facing the same file. A junior handling their first hostile-jurisdiction injury claim (Force 4) without senior oversight is not just making errors. They are making errors inside a system plaintiff counsel has spent years learning to exploit.
Montgomery (Force 5) expanded freight broker liability at the exact moment the litigation environment became least forgiving. A broker fully in the negligent selection chain, in a hostile jurisdiction, with a thin vetting file and a junior adjuster on the claim, is carrying a risk profile that did not exist eighteen months ago. The forces did not each get worse independently. They got worse together.
The data force (Force 2) and the AI force (Force 6) are the infrastructure layer underneath all of it. Without real-time data, AI adds marginal value. Without experienced judgment, AI becomes a substitute for the expertise it was meant to support.
Select a force to light the ones it amplifies. Every connecting line is drawn from a sentence in this paper.
Select any of the six forces above.
"Two participants in this research criticised third party administrators without being asked, so it belongs in the findings rather than in a footnote. Both said TPAs add a layer, move slower than going direct, and underinvest in the data they are sitting on. One made the sharper point: TPAs are not measured on loss ratio, which is the only number that decides whether a programme survives.
I run a TPA. They are describing the category accurately. Any TPA that reads this and feels defensive rather than recognised has proved their point.
The wider version of that argument is the one I would leave people with. Claims has been treated as a cost centre for too long. That framing is wrong and it is expensive. The claims department is not the place where money goes out. It is the place where the value of the insurance relationship is either confirmed or destroyed. Every carrier, every MGA, every programme administrator that continues to underinvest in claims talent and claims data is making a bet that nothing systemically difficult will happen on their book. That bet is getting worse every year."
The question
Before that question, one more voice, because it locates where the answer has to come from.
"My most significant takeaway is that technology alone will not solve these challenges. We certainly need better data, better processes and better tools. But we also need experienced, flexible, motivated people who know how to question the facts, recognise inconsistencies, understand coverage and make sound decisions.
I see AI as an opportunity to reduce the time spent on administrative tasks and create more capacity for relationship building, critical thinking, knowledge development and professional growth, to address the existing talent gap.
I believe the talent exodus relates heavily to the burnout rate of claim professionals as much as anything, and that makes managing the other forces incredibly difficult. The focus must shift, in the office or working from home, toward consistent training, greater job satisfaction and meaningful mentorship, to develop, retain and strengthen the next generation of claims professionals. To make claims a destination, not simply a stepping stone."
The claims department is where the value of the insurance relationship is confirmed or destroyed. When a cargo owner suffers a loss, when a freight operator faces a liability claim, when a maritime injury lands in a hostile jurisdiction, it is the claims operation that delivers on the coverage or does not. That is not a back-office function. It is the moment the contract either means something or it does not.
All six forces are landing on it at once.
So if there is one question to sit with: if the five most consequential pressures on your operation arrived simultaneously tomorrow, would your team have the expertise to navigate them and the data to see them coming?
Most operations can answer one half of that with confidence. Very few can answer both with evidence rather than assertion.
The ones that can are distinguished not by size or longevity but by a set of deliberate decisions. To manage caseload as a strategic variable rather than an outcome. To invest in mentorship rather than assume it happens. To treat data integrity at intake as non-negotiable. To keep human judgment in the decisions that determine outcomes.
Those decisions compound. The gap between the operation that made them and the one that did not widens every year, and it widens fastest when the external forces are accelerating.
This paper is a diagnostic. Every claims leader who reads it now has something they did not have before: a documented account of what is actually happening on the ground, in the words of professionals across the United States, Europe and the Middle East, that they can measure their own operation against.
What happens next is their decision.
The six forces in this paper are not predictions. They are conditions that already exist. The question is not whether they are real. It is whether your operation is ready for them.
Every participant quote in the paper
Each one exactly as it appears above, with the line that introduces it where the paper gives one. Filter by force, or open the passage it comes from.