Behavioral Science for Insurance Claims: Reducing Customer Stress and Improving Follow-Through

Last updated October 23, 2026

Just over a third of consumers, 36%, would accept a fully automated claims process in return for cheaper cover, according to KPMG research on consumer attitudes to AI in claims reported by Insurance Times in April 2026, even though about 3 in 4 insurance professionals in a separate Insurance Times survey say their firms already use or pilot AI in claims.

Insurers use behavioral science to improve the claims experience by redesigning the moments where stressed customers stall or lose trust. At The Decision Lab, we focus on five of those moments: the first step of a claim, the wait for news, document requests, follow-up tasks, and the explanation of the outcome. Automation can speed up every one of these steps, and it can also make a claim feel colder at the moments customers care about most.

We have worked inside this industry before. For one of North America's largest insurers, we analyzed more than 170,000 hours of agent sales calls to find where conversations went wrong, then tested new scripts and agent training against a control group. Sales in the first pilot rose 12% over the control group, as described in our insurance sales call case study, and the insurer now runs an in-house behavioral science team that we designed and helped staff. Claims conversations have the same kind of hidden turning points as sales calls, and a customer who has just lost a car or a kitchen ceiling has far less patience for getting through them.

1. Make the claim easy to start, including for customers who are unsure whether filing is worth it.
2. Keep uncertainty low while customers wait, by showing what is being done and what happens next.
3. Ask for one thing at a time, because stress shrinks the attention customers can give each request.
4. Turn document and appointment requests into specific plans with a date and a time.
5. Explain settlements and denials so the decision feels fair, with a person available when the answer is no.

Customer effort, the work a claimant has to do to move a claim forward, sits underneath satisfaction scores. It shows up in status-chasing calls, duplicate document uploads, missed appointments and claims abandoned partway through. The first four interventions reduce customer effort directly, and the fifth targets trust, since a claim that took little effort can still feel unfair.

Buying or renewing a policy is a decision customers choose to make, usually when nothing has gone wrong. A claim arrives after a loss, often at a bad time, and asks people to gather evidence and then wait on a decision they cannot control. Most customers file claims rarely enough that every step feels unfamiliar, and they judge the whole relationship with their insurer by how that one episode goes.

J.D. Power's claims studies show what customers weigh most. Its 2025 U.S. Auto Claims Satisfaction Study, based on 9,455 customers who had settled a claim, ranks trust first and fairness of the settlement second among eight dimensions, with ease of starting a claim and digital channels in the last two places. Its 2026 U.S. Property Claims Satisfaction Study, based on 5,093 homeowners who had filed a claim, puts fairness of the settlement first and trust second, with ease of starting and ease of resolving the claim at the bottom.

Ease still matters, because effort is where customers drop out or call back to chase. It matters less to how they judge the experience afterwards than whether they trusted the process and believed the outcome was fair. For an insurer deciding where to start, the rankings suggest that work on trust and perceived fairness should run alongside work on digital intake, not behind it.

Financial stress adds to the problem. In our research with Capital One on financial anxiety, the negative effects of financial stress held even after we controlled for household income and credit scores, so the strain did not simply track how much money people had. A claimant waiting on a settlement for a damaged home is usually under exactly this kind of strain.

In J.D. Power's 2025 auto claims study, 7% of auto insurance customers, about 1 in 14, said they had avoided filing a claim for fear their rates would rise. Some of those decisions are sound financial choices. Others rest on guesses about consequences the customer could have checked, if checking had been easy.

Research on why people fail to claim money they are owed shows how much the form of a first message shapes whether they act. Saurabh Bhargava and Dayanand Manoli ran a field experiment with the US tax agency on claiming a tax credit, published in the American Economic Review, sending redesigned notices to 35,050 tax filers who had not claimed a credit they qualified for, even after a first reminder. A more complex notice lowered take-up by 6 percentage points compared with the simpler version, and a longer worksheet lowered it by 4. Showing the likely credit amount raised take-up by 8 points compared with notices that left it out, while messages meant to reassure people about stigma or audits made no measurable difference.

For insurers, the lesson is to make the first step short and concrete. Let customers start a claim with the minimum information and add the rest later. Pre-fill what the policy record already holds, and tell them plainly what they stand to receive, including the deductible that applies and the expected next step. Where the rules allow it, explain what filing will and will not change about their premium, so the decision to file rests on facts.

Most of a claim is waiting, and customers judge the wait by what they can see. Ryan Buell and Michael Norton called this the labor illusion, in research on operational transparency published in Management Science. In their experiments, people valued a service more when it showed the work being done on their behalf, and some preferred a slower service that showed its work to an instant one that did not.

J.D. Power's 2026 property study points the same way. Just under half of the claimants it surveyed, 45%, received claim updates digitally, and customers who used digital tools during their claim reported higher satisfaction than those who did not.

We have tested the experience of waiting directly. For the sonic branding agency Made Music Studio, we ran a controlled experiment on how sound changes the experience of waiting with nearly 200 participants, comparing seven sound conditions, including ambient music and silence. Time felt like it passed fastest with ambient music, and the results suggested that sound can raise trust and lower stress during a wait.

For claims, this means status updates that name the task in progress and give a date range for the next step. The update should reach the customer before they have to call and ask, because a call asking where a claim stands usually means the customer has gone too long without news.

Stress narrows attention. Anandi Mani, Sendhil Mullainathan, Eldar Shafir and Jiaying Zhao tested how financial worry affects reasoning in a study published in Science, asking shoppers to think through a hypothetical car repair before taking reasoning tests. When the repair cost $1,500, lower-income participants scored noticeably worse than when it cost $150, while the scores of higher-income participants barely changed.

Claims letters often bundle every outstanding requirement into one message, from photos and receipts to repair estimates and signed forms. Splitting a long request into a sequence, one action per message with a short reason for each, gives customers a task they can finish. Confirming receipt of each item also stops the duplicate uploads and chasing calls that come from not knowing whether something arrived.

Many claims stall on tasks the customer fully intends to do: booking the repair, sending the estimate, attending the adjuster visit or, for disability claims, taking the first step back to work.

Katherine Milkman and colleagues tested a simple fix for this gap with flu shots at a large firm, in a field experiment on planning prompts published in the Proceedings of the National Academy of Sciences. Every employee received a mailing with clinic times and locations, and about 1 in 3 of those who got only that mailing were vaccinated. Employees whose mailing also prompted them to write down the date and time they planned to go were vaccinated at a rate about 4 percentage points higher. A prompt for the date alone did not produce a statistically significant change.

The moment a plan gets made can decide follow-through months later. For American Financial Solutions, a nonprofit debt management program, we analyzed records from thousands of clients and traced early dropout back to the first onboarding call. After we redesigned the counselor call scripts, check-in texts, marketing collateral and mail warnings, the dropout rate fell by half.

In claims, the first notice of loss call plays the same role, since it is where customers learn what they will need to do. We would expect a call that ends with a booked date for the first task to produce better follow-through than one that ends with a promise to be in touch, and it is an inexpensive change to test.

Fairness sits at or near the top of both J.D. Power rankings, and people tend to remember an experience by its worst moment and its ending. Donald Redelmeier and Daniel Kahneman found in a study of how patients remember painful procedures, published in Pain, that memories of how painful a colonoscopy had been tracked the most painful moment and the final moments, with little regard for how long the procedure lasted. In a claim, the ending is the settlement or the denial.

A settlement letter that shows how the amount was reached and what the customer can do if they disagree gives the decision a chance to feel fair even when it disappoints. A short call from a named person before a denial letter arrives does the same work for the hardest cases.

Automation raises the stakes of this step. In the KPMG research reported by Insurance Times, 2 in 3 consumers aged 18 to 24 said they would pay more for guaranteed human involvement in claims handling, compared with about 3 in 10 of those over 65. The customers most comfortable with digital tools were the most willing to pay for a person. We would use AI to speed up simple, uncontested claims while keeping a named person visibly involved in denials and disputed settlements, and we would tell customers which parts of their claim were automated.

Claims teams also use behavioral design to encourage honest reporting, and here the evidence needs care. A widely cited 2012 paper in the Proceedings of the National Academy of Sciences reported that signing an honesty statement at the top of a form reduced dishonest reporting, including in a field experiment with an auto insurer. The journal retracted the paper in 2021 after independent analysts presented evidence questioning the validity of its data, and a large replication published in 2020 by Ariella Kristal and colleagues, including the original authors, found that signing first did not reduce dishonesty.

Insurers that built forms around that result should test them in their own setting before counting on them. Anti-fraud friction also lands on every honest claimant, so its cost belongs in the same test.

Satisfaction scores arrive late and blend many causes, so we pair them with behavioral measures that show where a claim actually stalls. The most useful are the share of started claims that are completed, the time between a document request and its arrival, the number of status-chasing contacts per claim, and the share of settlements that are disputed or reopened. Retention at the first renewal after a claim connects the claims experience to the revenue measures that sales and renewal teams already track.

Our claims work follows the same sequence we use across financial services. We start by mapping the journey from the claimant's side, through interviews and analysis of call recordings and drop-off data, to find the specific moments where customers stall or give up. We then design targeted changes for those moments and test them.

A test against a control group gives the strongest evidence, and we use one where claim volumes and the stakes allow it. For low-volume claim types, a careful before-and-after comparison can be the more sensible choice. Once a change works, we help build it into the scripts and systems your team already uses, so it outlasts our engagement. For insurers that want to run this work themselves, we can also design and help staff an in-house behavioral science team, as we did for the insurer in our sales call work.

How can insurers use behavioral science to improve the claims experience?

Insurers use behavioral science to find the moments in a claim where stressed customers stall or lose trust, then redesign those moments and test the changes. Typical targets include the first notice of loss call, document requests, status updates and settlement letters. The work is most useful when each change is measured against a control group or a clear baseline.

What does behavioral science consulting for insurance claims involve?

An engagement usually starts with research into the claimant journey, using interviews and operational data such as call recordings, followed by designing and testing changes at the points where customers drop off. At The Decision Lab, we also help build the winning changes into scripts and digital flows, and can design an in-house behavioral team for insurers that want to keep the capability.

How do insurance companies reduce customer effort during claims?

They let customers start a claim with minimal information, pre-fill details from the policy record, ask for one item at a time and send updates before customers have to chase. The number of status-chasing calls per claim shows whether effort is actually falling, and it is cheap to track from existing contact logs.

Should insurers use AI to handle claims?

AI can shorten simple, uncontested claims, but consumer surveys show strong demand for human involvement, especially in decisions about payment. We recommend keeping a named person involved in denials and disputed settlements, and testing whether automated messages change trust or follow-through before rolling them out widely.

Does behavioral science help reduce insurance fraud?

It can, but the best-known example needs caution. The finding that signing an honesty statement at the top of a form reduces false reporting was retracted, and a later large replication found no effect. Insurers should test any honesty prompt in their own claims flow and measure its cost to honest customers as well as its effect on fraud.