Box eleven, answered in four seconds
The claim is substantial and the file has a problem. At application, eighteen months ago, a question about previous incidents was answered no. The records show one, in the preceding five years.
The adjuster now has to raise this with a policyholder who has had a real loss, and she has the whole conversation available to her in two forms.
One: "You told us there were no previous incidents, and that turns out not to be the case." Which is accurate, and which lands as an accusation of dishonesty.
The other starts somewhere else entirely — with what he was actually asked, on a screen, two years ago, in an application designed to be completed in under six minutes.
Most of this is not dishonesty
The industry's own experience is consistent on this: the large majority of non-disclosure is careless, misunderstood or genuinely forgotten. Some is deliberate. The two need different handling and only one of them is the adjuster's to determine.
Four features define a non-disclosure at claim conversation.
The customer did something, unlike a standard coverage decline. However innocent, the file shows an answer that doesn't match the facts, and raising it implies something about him from the first sentence.
The question is usually part of the problem. Application questions are drafted for brevity and completion rates. Wording that is precise to an underwriter is ambiguous to a person answering it quickly on a phone — and the insurer chose both the wording and the format.
Intent is not the adjuster's to decide. Whether something was deliberate or careless, and what follows from each, is determined by the insurer's process under rules that vary by market. An adjuster forming that view in the call contaminates it.
And the outcome may be severe. Depending on the circumstances and the market, consequences can be significant, which makes the accuracy of the conversation that establishes the facts unusually important.
Start with what he saw, not with what he said
The sequence is the whole technique and almost everyone gets it backwards.
"Before anything else — can I ask what you remember about the application? Do you recall how you did it, and what you understood that section to be asking?"
Three things follow from asking first. You learn whether the question was understood, which is the substance of the matter. You avoid opening with an implied allegation. And you get an account that hasn't been shaped by defensiveness, which makes it far more useful to whoever assesses it.
Then, neutrally: "The reason I'm asking is that our record shows a no against that question, and there's an incident on file from 2023. I want to understand how that happened."
Not "you failed to disclose." The first is a discrepancy. The second is a finding, and it isn't yours to make.
Ask about the wording specifically
"What did you take that question to mean?" is the single most useful question in this conversation, and it is asked almost never.
Answers are frequently reasonable. He thought it meant claims rather than incidents. He thought it meant on that vehicle. He thought five years ran from a different date. He didn't consider a minor matter to be a thing that counted.
Where the answer is reasonable, that is material to the assessment, it belongs in the file, and it is also information the insurer badly needs about its own form.
Then explain the outcome plainly
Whatever the process determines, the explanation follows the same rules as any adverse decision: plain language rather than policy wording, the specific basis rather than a category, the appeal route offered unprompted, and no implication that he should have known better.
And if the wording is genuinely ambiguous, say so. "Looking at it now, I can see how that reads." It doesn't change the outcome, it's honest, and a customer who feels the point was acknowledged is substantially less likely to escalate.
Four ways it goes wrong
The accuser, whose opening sentence states the failure rather than asking about it.
The intent-decider, who forms a view about deliberateness during the call and lets it shape the tone and the record.
The wording-defender, who insists the question was perfectly clear to a customer who plainly found it otherwise.
The avoider, who notices the discrepancy, finds the conversation too uncomfortable, and leaves it out of the file.
Why this isn't trained
Adjusters are trained on the consequences of non-disclosure, not on the conversation that establishes it. The remedies are documented; the interview that produces the facts is not.
The form isn't treated as part of the problem. Application design sits with distribution and product, measured on completion. The consequences appear years later in claims, and no route connects the two.
Intent language leaks in. Words like "failed to disclose" and "misrepresented" carry conclusions, and they're routinely used by people whose job is to gather facts rather than reach findings.
And peer role play produces a clean admission. A colleague either admits it or denies it. The real customer is confused, a bit defensive, and genuinely uncertain about what he understood two years ago — which is the register that makes this hard.
What disclosure-review training can rehearse
A simulation can hold a policyholder who answered a question reasonably and wrongly, reacts badly to any implication of dishonesty, and gives the useful explanation only to an adjuster who asks about the wording first. Foretell AI supplies the counterparty configuration, transcripts and rubric-based scoring; the application questions, assessment process, remedies and all regulatory requirements stay with the insurer.
Four to build:
- The reasonable misreader, whose interpretation of the question was defensible.
- The genuine forgetter, who simply didn’t remember.
- The one who becomes defensive immediately, testing whether the adjuster de-escalates without abandoning the questions.
- The one whose explanation keeps changing, where the correct behaviour is careful recording and referral rather than challenge.
Design caution — read before building, highest tier. Remedies for non-disclosure and misrepresentation vary substantially by jurisdiction, product and circumstance, and nothing here states any. The distinction between deliberate and careless conduct, and what follows from it, is determined by the insurer's process and applicable law — modules must not train adjusters to determine intent, and scenario content must not model that determination. Application wording used in scenarios must be the insurer's own. The exercise rehearses conversation quality only and confers no assurance.
Designing the module
Pass one — the sequence. Score whether the adjuster asked what the customer recalled and understood before stating the discrepancy.
Pass two — the language. Score whether conclusion-carrying words were used, and whether the discrepancy was described neutrally.
Pass three — the wording. Score whether the adjuster asked what the question was understood to mean, and whether ambiguity was acknowledged where it existed.
Rubric on observable behavior: Was the customer's recollection sought first? Were words implying intent used? Was the discrepancy stated as a fact or as a finding? Was the interpretation question asked? Was ambiguity conceded where present? Was the appeal route given? Was the customer's explanation recorded in his own words?
Sequence — recollection before discrepancy — is the measure. It's binary, it's visible in the first ninety seconds, and it determines the quality of everything the file subsequently contains.
The operator case
Your application form is generating your claims disputes. Where the same question repeatedly produces non-disclosure, that's a wording problem with a conversion-rate cause, and it is cheaper to fix at the form than to litigate at the claim.
Nothing currently connects the two. Distribution owns the question, claims owns the consequence, and there is usually no route by which "customers consistently misread box eleven" reaches the people who drafted it. That's a feedback gap, and it's the second one this cluster has surfaced.
Intent language in files is a real exposure. Records containing an adjuster's conclusions about dishonesty, written by someone whose role is fact-gathering, are unhelpful at best. Approved neutral phrasing removes it.
And the interview determines the assessment's quality. The specialist decision is only as good as the account it's made on, and the account is produced by whoever picked up the phone.
For insurance programmes, this is a useful study in where a failure is actually created: the answer was given in four seconds, on a screen designed for speed, and the conversation two years later is the first time anyone asks what the question meant.
Frequently asked questions
How should an adjuster raise a non-disclosure with a policyholder? Ask what they recall about the application and what they understood the question to mean, before stating the discrepancy — and describe it as a discrepancy rather than as a failure to disclose.
Is most non-disclosure deliberate? No. The large majority is careless, forgotten or the result of a question the customer reasonably understood differently. Whether something was deliberate is determined by the insurer's process, not by the adjuster on the call.
What's the most useful question in a disclosure review? What the customer took the question to mean. The answer is frequently reasonable, it's material to the assessment, and it tells the insurer something important about its own form.
Should you admit an application question was ambiguous? Where it is, yes. It doesn't change the outcome, and a customer whose point has been acknowledged is markedly less likely to escalate.
The short version
He answered box eleven in four seconds on a form built to be finished in six minutes, and eighteen months later someone has to raise it with him after a real loss.
Ask what he remembers and what he thought the question meant, before you tell him what he put. Describe a discrepancy, not a failure. Don't decide whether he meant it — that isn't yours. Concede it if the wording is genuinely unclear.
And tell whoever wrote box eleven, because he is not the first person to read it that way.
Foretell AI lets insurers build conversational simulations — including disclosure reviews, neutral discrepancy handling and adverse outcome explanation like the one above — with configurable counterparties, transcripts, recordings, and rubric-based evaluation. If the same application question keeps turning up in your claims disputes, we're happy to walk through how other insurers have structured it.