Claims leakage is the gap between what a claim cost and what it should have cost. Most carriers and IA firms go looking for it the same way: pull a sample of closed files, score them against a checklist, and report a number. That sample tells you leakage exists. It can't fix the files it never opened, and on most desks that's nearly all of them.
The answer isn't a bigger sample. It's reviewing every file against the carrier's own guidelines before it closes, and giving the sample a better job.
IRMI defines claims leakage as dollars lost through claims management inefficiencies that come from failures in existing processes, manual and automated. Put simply, it's the difference between what you did spend on a claim and what you should have spent.
Most of it isn't fraud. Fraud gets the headlines, but leakage is usually ordinary: a line item the carrier doesn't allow, depreciation applied to something that isn't depreciable, a scope that doesn't match the photos, a supplement approved without the evidence behind it. Nobody in those examples is trying to cheat. They're what happens when a person has to hold a lot of rules in their head at once.
It isn't small, either. EY's P&C claims team, which runs claims quality assessments for carriers every year, puts leakage at roughly 7 to 14 percent of carriers' total spend. Even the low end of that range is bigger than most line items on a claims budget.
The standard approach is a closed-file audit. A QA team pulls a handful of files per adjuster or per vendor each month or quarter, reviews them against a checklist, and scores them. The scores roll up into a report, the report shows a trend, and the trend drives coaching.
There's a good reason it works this way. A thorough manual review of a complete property file, meaning the photos, the estimate, the sketch, the field notes and the general loss report, takes most reviewers an hour or more. Nobody has the staff to do that on every file, so you sample.
A sample is good at one thing. It tells you whether you have a problem and roughly how big it is. It's a thermometer, not a treatment.
This is the part that gets underrated. The variability in how claims should be serviced, how estimates should be written and how coverage is applied is enormous. It hits IAs and TPAs hardest, because they work for many carriers at once, and carriers carry a version of it too.
One carrier wants specific photos in every file. Another wants particular headers and forms. A line item that's allowed on one program isn't allowed on the next. What's depreciable, and what has to be provided to the onsite contact, changes by carrier. Add it up and you get hundreds of combinations of rules and guidelines, and those combinations are what decide whether an estimate package is correct.
A sampled audit asks a reviewer to carry all of that for a handful of files. The field team is supposed to carry it for every file. When a file leaks, it's rarely because somebody didn't care. It's because nobody can hold hundreds of rule sets at once.
Review that looks only at the estimate catches pricing and line-item problems inside the estimate, and that's useful. But many of the reasons a file is wrong live outside it: a photo of damage with no line item for it, a line item with no photo to support it, a sketch whose measurements don't match the scope, a field note that explains an exclusion the estimate never mentions.
It cuts the other way too. A reviewer who reads only the estimate will flag a missing item that the general loss report already explains, and the file goes back to the field for nothing. Reading the whole packet together is what separates a real finding from a false alarm.
When every file is checked against the right carrier's guidelines before it closes, three things change.
The finding arrives while you can still act on it. A missing photo, a line item that isn't allowed or a depreciation error gets fixed on the file, not recorded in a report about files you already paid.
The standard stops depending on who reviewed it. Every file is checked against the same rule set for that carrier, so the result doesn't change with a reviewer's workload or experience.
Your reviewers spend their time on judgment. Instead of reading every page hunting for problems, they start from a short list of findings, each one naming the rule, showing the math and pointing to the evidence in the file. They still make the call.
That's what QA Complete does. It reads the full file, including the photos, estimate, sketch, notes and general loss report, against the carrier's own guidelines, and hands back findings for a person to sign off. The guidelines are set up during onboarding, so nobody on your team writes or maintains rules. A file takes four to five minutes to process, and reviewers get about 45 minutes back per review.
Your sampled audit doesn't go away. It gets a better job: checking that the review is catching what it should, instead of being the only check most files ever get.
You don't have to take any of this on faith. The cleanest test is closed files, because you already know how they turned out.
Pull a batch of closed files, including some you know had problems and some you think were clean.
Send the carrier guidelines those files were held to.
We review every file in the batch against those guidelines.
Your QA team grades the findings: what it caught, what it missed and what it flagged that it shouldn't have.
Compare the findings with what your sampled audit found on the same files.
That last step tells you how much your sample has been missing. Run it on your own closed files and see.
Claims leakage is money lost on claims through process failures: the difference between what an insurer paid on a claim and what it should have paid. Common causes include line items outside the carrier's guidelines, depreciation errors, scope that doesn't match the evidence and missed recoveries. Fraud is one cause, but most leakage is ordinary error.
A claim file review checks a claim file against the carrier's guidelines: whether the documentation is complete, whether the estimate is supported by the photos and scope, and whether the right rules were applied. It can happen before a file closes or afterward, as part of a closed-file audit.
Usually through closed-file audits. A QA team reviews a sample of paid files, identifies what should have been paid differently and projects the rate across the book. That gives you a rate. It doesn't correct the files outside the sample.
There's no single standard. It depends on volume, staffing and the risk on each line of business. The better question is what the sample is for. A sample can estimate your leakage rate. Only reviewing every file can reduce it on the files you haven't paid yet.
No. It changes where their time goes. Reviewers start from findings instead of blank files, make the call on each one and use sampling to check the review itself. The AI surfaces. Your team decides.