Why Going Digital Made Premium Leakage Worse
There’s a stat that doesn’t get talked about enough in insurance circles: application misrepresentation, or put another way, application integrity — how accurately applicants represent their actual risk — is down more than 20% over the last decade.
Twenty percent.
And the timing isn’t a coincidence. That decline tracks almost exactly with the industry’s shift toward digital-first applications, frictionless quoting, and instant-issue policies. The same decade in which carriers made getting insurance faster and easier than it’s ever been is the same decade in which the accuracy of what people submit has quietly deteriorated.
That’s not an argument against going digital. The customer experience improvements are real, and the conversion gains that came with them were real too. But here’s the thing nobody was talking about at the time: friction wasn’t just slowing people down. Some of it was catching application misrepresentation.
When you removed it, the misrepresentation came with it.
The Trade-Off Carriers Made (And Are Still Making)
For most of the last decade, the mandate from carrier leadership was straightforward: make the experience faster, simpler, more competitive. Reduce underwriting questions. Lean on prefill to populate fields automatically. Compress quote-to-bind from days to hours to minutes.
Every one of those changes was rational in isolation. Together, they created a structural problem.
The old application process — the one with a live agent, 40 underwriting questions, and a three-day turnaround — had obvious friction. It also had a meaningful side effect: it made application misrepresentation harder. A longer process gave people more opportunities to reconsider a lie. A human on the other end raised the psychological stakes. Reflexive questions triggered by earlier answers forced applicants to be consistent.
Take all that away and replace it with a four-minute digital form, and the friction that was catching misrepresentation goes with it.
What we hear from carriers, consistently, is some version of this: they reduced friction, conversions climbed, results looked great for about 18 months — and then loss ratios started moving. The mandate from leadership flipped from growth to controls. And the growth team got tasked with putting back friction they had spent 18 months removing.
That cycle is running on repeat across the industry right now.
The Faceless Application Problem
There’s a human dimension to this that the data alone doesn’t fully capture.
When someone applies for car insurance through a live agent, they’re making their misrepresentation to a person. There’s eye contact (or at least a voice on the phone). Social norms create at least some friction around lying directly to another human being.
A digital application has none of that. You’re alone with a form. The premium counter updates in real time as you adjust your answers. You can review your quote and see what removing a driver does to your monthly payment. There’s no one watching, no one who might ask a follow-up question. The psychological cost of telling a small lie on a form is much lower than telling that same lie to a person.
The research backs this up. A 2021 survey found that 14% of Americans admitted to lying to their car insurer — nearly double the rate from the year before. The generational gap in that number is striking too. Baby Boomers admitted to it at 8%. Gen Z admitted to it at 18%. The generation that grew up interacting with forms rather than agents is also the generation most comfortable adjusting what they submit.
This isn’t a Gen Z character problem. It’s a design problem. When you make application misrepresentation easy, consequence-free, and instantaneously rewarded with a lower premium, more people do it. The incentive structure was accidentally optimized for exactly the behavior carriers are trying to prevent.
What’s Slipping Through
The categories of misrepresentation that digital applications accelerated all share the same profile: they’re hard to catch with third-party data, they’re easy to adjust mid-session, and they’re financially meaningful.
Garaging address is the highest-frequency example. An applicant lists a suburban address rather than where the car actually sleeps at night — a higher-rated territory. It’s difficult to verify, easy to rationalize (“I spend weekends there”), and the premium difference can be significant. Verisk estimates garaging misrepresentation costs carriers $32.5 million annually in Miami alone.
Mileage is another. A 2012 survey found that 16% of consumers admitted it was acceptable to lie about their mileage to an insurer. The industry estimates that figure translates to over $5 billion annually in lost premiums. With a digital form, adjusting the mileage field and watching the premium drop is a two-second transaction.
Hidden drivers are more intentional. Someone with a teenage driver — or a driver with a recent DUI, recent accidents, or a suspended license — simply doesn’t list them. Twelve percent of standard policies and 15% of nonstandard policies have hidden driver issues. Loss ratios on those policies run more than double the average because the hidden driver is almost always the higher-risk one.
Agent manipulation adds another layer. Some agents, motivated by conversion, guide applicants toward answers that improve quote outcomes. It doesn’t require explicit fraud — a subtle suggestion, an omitted question, a convenient interpretation of where the car is “primarily” garaged. The end result looks identical to applicant misrepresentation but originates from the distribution channel.
Why Third-Party Data Doesn’t Solve It
The natural response is to throw more data at the problem. MVRs, CLUE reports, LexisNexis, Experian — all valuable, all widely used. But they’re built to verify information against existing records. They catch misrepresentation that has already been documented somewhere.
They can’t catch application misrepresentation that’s happening in real time, for the first time, in the session itself.
A false garaging address has no record trail until something happens. A hidden driver has no record of being hidden. The applicant who just removed a high-risk driver from their application, watched their premium drop $800, and then submitted — that behavior happened entirely within the session. No database has it. No MVR captures it.
This is the gap that behavioral data was designed to close. Not as a replacement for third-party verification — those data sources do what they do well — but as a layer that captures what happens during the session, before submission, before a policy is issued.
And it turns out that how someone fills out a form — where they hesitate, what they edit, how long they spend on specific questions, whether they removed a field and retyped it — is one of the most reliable signals of intent that exists.
The Implication
Going digital didn’t create premium leakage or application misrepresentation. But it changed the conditions under which it happens. The old tools — designed for a slower, agent-mediated world — weren’t built to catch risk that materializes in a four-minute self-service session.
The carriers that are getting ahead of this aren’t adding friction back to their applications. They’re adding intelligence. There’s a difference. Friction makes every customer’s experience worse to catch the 10–15% who might be misrepresenting. Intelligence looks at every session — like a bag scanner that screens every applicant without pulling anyone out of line — and flags the specific ones that warrant a closer look.
The growth teams don’t have to unwind their work. And underwriting gets the signal it needs without waiting for loss ratios to tell the story 18 months too late.
→ Back to the Premium Leakage Guide
→ How Much Does Premium Leakage Actually Cost? A Loss Ratio Breakdown
→ How Behavioral Analytics Stops Premium Leakage Before It Starts
Curious how top carriers are addressing the digital application integrity gap? Let’s talk.
