Selling Into Insurance Claims: Cycle Time, Leakage and Why Nobody in That Building Cares About Your AI

Nobody in that building woke up wanting new software

The claims floor is not a growth org. It is a cost center that pays out money it would rather not pay, under a clock, under a regulator, and under a set of internal metrics that were set before you dialed. Nothing about your product changes that. What changes is whether you sound like someone who understands the job or someone who read a page on the carrier's website.

I have not spent my career inside a carrier. I have spent it selling into large, regulated, procurement-heavy enterprises, including inside AWS and Dell, and the pattern is consistent: the more consequence attached to a buyer's decision, the less interested they are in your capabilities and the more interested they are in what breaks and who gets blamed. Claims is that pattern at full volume. Every dollar that leaves the building is scrutinized after the fact. So is every vendor who touched the process.

Here is the field guide I would want if I were picking up the phone into claims tomorrow.

The org chart, and what each person is actually measured on

The adjuster

The adjuster owns a caseload. They are handling files, documenting them, setting reserves, negotiating settlements, and closing. They are measured on how many open files they carry, how fast they close, how often files reopen, and whether their file documentation survives a quality audit. They are not a buyer. They will never be a buyer. But they are the person whose behavior your software has to change, and if the VP thinks adjusters will hate it, you are dead in the water regardless of ROI.

Adjusters are also the best free discovery you will ever get. They will tell you exactly which screens they hate, which system they have to alt-tab into, and which part of the process they route around. Talk to one before you talk to the VP if you can.

The claims manager or supervisor

This person runs a unit. They are measured on their unit's pending count, cycle time, quality audit scores, and whether they are losing adjusters. They live in the gap between what leadership promised and what a team of overloaded people can actually do. They are your best internal champion because they feel the pain daily and they have enough context to translate your product into operational language.

They do not have budget. They have influence and they have the data you need.

The VP of Claims

This is your economic buyer for most claims tooling. They own indemnity spend, loss adjustment expense, cycle time across the book, and the customer experience metric that the CEO reads in a board deck. They are managing a headcount plan they probably cannot grow and a book of business that keeps producing claims.

The VP of Claims is measured on outcomes that are legible to finance. If you cannot state your value in those terms, you are a nice conversation they will not fund.

Actuarial and finance

Most reps never meet them and then get confused when the deal dies. Actuarial cares about reserve adequacy and whether the numbers they set are holding up against actual paid loss. If your product changes how reserves get set, how severity gets predicted, or how quickly files close, you are touching their world whether you intended to or not.

Finance cares about whether the savings are real and whether they show up in a line item someone owns. "Efficiency" that does not reduce headcount, reduce indemnity, or reduce external spend on adjusters, attorneys and experts is, to finance, a story.

The people who can stop you

Compliance and legal review anything that touches a claim decision, a customer communication, or claim data leaving the building. IT security has a questionnaire that will take longer than you think. SIU cares if you touch fraud referrals. Vendor management will benchmark your price against whatever they last bought.

None of these people can buy. All of them can stall. Plan for them in your first call, not your fourth.

The three numbers that move a claims buyer

Cycle time

The elapsed time from first notice of loss to close. Every claims leader tracks it, usually by line of business, usually with a target. Cycle time drives customer satisfaction, it drives the cost of carrying an open file, and in some lines it correlates with severity because files that sit get worse.

When you ask about cycle time, ask about the distribution, not the mean. Claims leaders know the tail is where the money is. A book where most files close quickly and a stubborn minority drag for months has a very different problem than one that is uniformly slow. Asking the question that way signals you have thought about it.

Indemnity leakage

Leakage is the money paid out that should not have been paid, or should have been paid at a lower amount, under correct handling. Overpayment on a settlement, a missed subrogation opportunity, a benefit paid that policy language did not require, a coverage decision made with incomplete facts. Carriers run leakage studies to size it. Most claims leaders have a number in their head and are not thrilled about it.

Leakage is the single best hook for most claims software because it is direct indemnity dollars, not soft efficiency. If your product finds a recovery opportunity, prevents an overpayment, or catches a duplicate, you are speaking the language of the money that actually matters.

Adjuster caseload

How many open files sit with each adjuster. Caseload is the pressure gauge for the whole org. When it climbs, cycle time climbs, quality drops, leakage grows, and adjusters quit, which pushes caseload higher for everyone left. Claims leaders are hiring against attrition and rarely winning.

If your product removes work from the adjuster's day, quantify it in files, not in minutes saved. "Each adjuster can carry more open files at the same quality level" is a sentence a VP can take to a headcount conversation. "Saves fifteen minutes a day" is not, and I have watched reps in every industry lose deals by pitching the minute instead of the file.

Why "AI" in your opener now reads as noise

Carriers have been running pilots for a while now. Document ingestion, severity prediction, fraud scoring, summarization of adjuster notes, automated FNOL triage. A lot of those pilots produced a slide deck and nothing else. The claims leader you are calling has probably sat through several, has explained to someone why the model's output could not be defended to a regulator, and has watched a vendor disappear when the integration got hard.

So when you open with "we use AI to transform the claims process," you are not differentiating. You are joining a queue of people who said the same thing and left. You are also, quietly, triggering the compliance reflex, because AI in a claims decision path means explainability, adverse action, bias testing and state filings.

The fix is not to hide the technology. It is to lead with the outcome and the mechanism, and let the technology be an implementation detail that comes up when they ask. "We read the medical bills and flag the ones that do not match the treatment codes on the file" is a sentence a claims manager can picture. It also happens to be AI. Nobody needs you to say so in the first fifteen seconds.

The other thing that lands: say out loud that you know they have been through pilots that went nowhere, and ask what killed them. That question does more for your credibility than any capability slide. The answer is usually integration, data quality, or an internal owner who moved teams, and all three are things you can then address on purpose.

Learn the vocabulary before you dial

You do not need to be an adjuster. You need to not sound like a tourist. A few terms that show up constantly:

Severity is the average cost of a claim. Frequency is how many claims come in. Leaders talk about both, and a change in one without the other tells a story. Reserves are the money set aside on an open file against expected payout, and reserve accuracy is a genuine sore spot because setting them too low creates ugly surprises and setting them too high ties up capital. Development is how a claim's estimated cost changes over its life.

Loss adjustment expense, usually LAE, is what it costs to handle the claim, split between allocated expense tied to a specific file, like an independent adjuster or defense counsel, and unallocated overhead. When you pitch efficiency, you are usually pitching against LAE. When you pitch leakage, you are pitching against indemnity. Know which one you are talking about, because they are different budgets with different owners.

Subrogation is pursuing a third party who caused the loss to recover what the carrier paid. Recovery is the broader bucket including salvage and deductible recovery. Subro is a place where money is provably left on the table, so if your product touches identification of subro opportunities, you have a very clean story. Reopens are closed files that come back, and they are a quality embarrassment. Pending is the open file inventory. FNOL is first notice of loss, the intake event, and straight-through processing is a claim that closes without human touch, which is the holy grail in high-frequency, low-severity lines and a non-starter in complex ones.

Use two or three of these correctly and the tone of the call changes. Use them wrong and it changes the other way, so if you are not sure, ask instead of guessing.

Translate your feature into a per-claim dollar

This is the work most reps skip. Claims buyers think in per-claim economics because that is how their book is modeled. Your job is to convert whatever you do into a figure attached to a single file, and then let them multiply it by their own volume.

You cannot do this with numbers you invented. You do it with theirs. Ask for claim volume by line, the severity on that line, the current caseload, and their own leakage estimate if they have run a study. Then build the arithmetic in front of them: if we prevent this category of overpayment on the share of files where it occurs, at your severity, that is this many dollars per claim, times your volume. Leave the assumptions visible and conservative. Claims people audit for a living. A model they can poke holes in and still believe is worth more than a polished number they distrust.

The same logic applies on the expense side. If a step you remove is currently done by an outside vendor, that is an invoice with a real number on it and the finance conversation is easy. If it is done internally by an adjuster, the value only becomes real if it converts into more files per adjuster or fewer new hires, and you should say that plainly rather than pretending saved minutes are cash.

One caution: never present savings that imply people get fired unless the buyer said it first. Claims leaders are usually understaffed and are pitching redeployment, not reduction. Frame the number as capacity.

Regulatory review will eat your timeline unless you get ahead of it

Claims is regulated at the state level in the US, and market conduct exams look at how claims were handled, how fast, with what communications, and on what basis decisions were made. Anything that touches a coverage decision, a payment, or a policyholder communication draws scrutiny. Some changes require filings. Model governance adds its own review layer if you are scoring anything.

The mistake is treating this as legal's problem to be discovered in month four. By then your champion has built a business case on a timeline that is about to slip two quarters, and the deal goes cold not because anyone said no but because momentum died.

Pre-empt it in discovery. Ask directly: does something like this go through your compliance review, who owns that, have you put a vendor through it recently, and how long did it take. Ask whether the use case sits inside or outside the claim decision path, because that distinction usually determines the depth of review. Ask if there are states where the answer is different. You will occasionally get a surprised pause on the other end, and that pause is the sound of you becoming a different kind of vendor. Then build the timeline with those steps in it and hand it to your champion so they look prepared to their own leadership.

The call sequence

Cold call

You get about ten seconds of patience. Do not spend them on your company. Open with the metric and a specific, unglamorous mechanism, then ask a question that only someone in claims can answer. Something like: "I work on subrogation identification. Most carriers I talk to find that referral rates vary a lot by unit rather than by file type. Is that true on your side or have you already normalized that?"

That opener assumes competence, names a real problem, and hands them the floor. It also invites a correction, which is fine, because a claims leader correcting you is a claims leader still on the phone. If you want a fuller structure for holding that conversation past the first objection, I laid out an approach in the insurance cold call script that is built around exactly this problem.

Discovery

Do not run a features interview. Run a measurement interview. What is your cycle time target and where does the tail sit. What does your leakage study say, and when was it last run. What is caseload now versus a year ago and what is your attrition doing. Where are files sitting waiting on something rather than being worked. What is currently outsourced and what does that cost per file. Which of these did you already try to fix, and what happened.

That last question is the one that surfaces the pilot graveyard. Ask it gently and listen for the reason it failed, because that reason is your implementation plan. The 25-minute discovery playbook for claims walks through the sequencing so you get the money questions in before your time runs out.

Demo

A claims room is not watching to be impressed. They are hunting for where it breaks. Expect the complex file, the coverage edge case, the state where the rule is different, the adjuster who works around the system. Bring those yourself. Show the exception path before they ask for it, show what happens when the data is bad, and show the audit trail, because the audit trail is what makes the whole thing defensible in an exam.

Say "it does not do that" when it does not do that. In a room full of people who evaluate credibility for a living, one honest no buys you three yeses. I go deeper on that posture in the claims demo script, including how to handle the senior person who stays quiet until minute twenty and then asks the only question that matters.

What I would do next

Pick one line of business, learn its claim lifecycle end to end, and write your per-claim math on a single page with the assumptions exposed. Then say the whole thing out loud until the vocabulary stops sounding borrowed. That last part is the step everybody skips, and it is the one that decides whether a claims leader keeps listening. It is why we built DrillCall — so reps can rehearse the claims opener, the leakage questions and the hostile demo room against a buyer who pushes back, before spending a real dial on it.

Get the language right and the rest of the deal gets easier. Get it wrong and no amount of product will save the call.