There is a moment in every insurance relationship when the abstraction becomes concrete: the claim. Everything before it — the comparison, the premium, the rider you declined — was preparation for a single decision made by someone else about your money, at a time you did not choose.
Fiscora can help you buy the right policy. It cannot help you at the claim, because the claim is not an arithmetic problem. It is a question of standing: whether you are a party to a decision or merely its subject. This article is about acquiring that standing.
Why "unappealable" is a category, not a complaint
The instrument governing this publication takes an unusually strong position here, and it is worth quoting because it reframes the problem: proportionality and the right to be heard are constitutive of justice, not additions to it. A decision that cannot be contested is not a just decision that happens to lack an appeal; it is not a just decision.
That is not rhetoric. It is a claim about what kind of thing a decision is. If a determination affects your livelihood, healthcare or housing, and it was made automatically, and no specific reasons were given, and there is no appeal to a human with authority to reverse it — then four conditions have stacked, and the register that governs this publication treats that stack as an absolute prohibition, not a bad practice.
The four elements matter because they are cumulative. Remove any one and the absolute is not engaged:
| Element | What removes it |
|---|---|
| Material consequence | The decision does not touch money, health, housing, work or a service you depend on. Rare in insurance. |
| Automated determination | A named human actually made the call, exercising judgement. |
| No specific reasons | You were told why — not a category code, an actual reason referencing your facts. |
| No appeal to a human with reversal authority | Someone exists who can say yes after someone said no, and you can reach them. |
This table is your checklist. When a claim is denied, work down it. You are establishing which element is present, because the element that is missing is the one you demand.
The four things to ask for, by name
1. Specific reasons
Not "your claim does not meet the policy criteria." That is a category, not a reason. A specific reason references your facts: which criterion, which document, which date, which clause.
The most effective sentence we know in this domain is close to this: "Please state the specific provision relied on, the specific evidence in my file that you consider engages it, and the name and role of the person who made this determination." It is effective because each element is separately answerable, and an organisation that answers two of three has told you something about the third.
2. A human with authority to reverse
The critical phrase is authority to reverse. Many appeal routes lead to a person who can only re-run the process. Ask directly: "Does the person reviewing this have authority to overturn the original determination, or only to confirm that the process was followed?" The answer tells you whether you are appealing or being processed.
The telephone trapA phone line that routes to the same automated system is not a human alternative. The governing instrument names this explicitly, because it is the most common way an appeal route is nominally provided and practically absent. If the call ends in the same answer with no new reasoning, you have not appealed yet.
3. A stated time limit
Ask when you will have a decision, and get it in writing. An indefinite process is a denial delivered slowly, and it is particularly effective against a claimant who is ill or out of income. A stated limit converts silence into a fact you can cite.
4. Published aggregate error rates
Rarely available, always worth asking for. "What proportion of determinations of this type are overturned on appeal?" A high overturn rate is the single most useful fact you can hold, because it converts your individual dispute into an instance of a known pattern. Where the figure is published by a regulator rather than the insurer, use the regulator's.
The inference problem: priced on what you never told them
Underwriting has moved a long way from the medical questionnaire. Pricing increasingly draws on behavioural proxies — purchase patterns, device data, location history, credit-adjacent signals — from which conclusions about your health, stability and risk are inferred. The register names this the human as product: the person is the raw material, not the customer, and cannot know what is taken.
The governing instrument's response is a duty it calls the inference inventory: an operator profiling persons must maintain a list of the inference classes it draws, expose it to the subject in readable terms, permit correction, and permit severable refusal — refusing one inference class without losing unrelated function.
Most insurers do not do this. Asking anyway is not futile, for three reasons: the request creates a record, the answer or refusal is itself informative, and in several jurisdictions a subject-access right gives you a partial version of the same thing with legal force behind it.
- Ask what categories of data beyond your application were used in pricing your policy.
- Ask whether any inference about your health was derived rather than disclosed.
- Ask whether you can decline a data category without losing coverage. The answer defines whether your consent was real.
- Where a subject-access right exists in your jurisdiction, use it. It is the same question with an enforcement mechanism.
What you can compel varies enormously by jurisdiction. Some regimes give a right to an explanation of automated decision-making and a right to human review; others give almost nothing. The questions above are universal; the leverage behind them is not. Check what your own jurisdiction provides before assuming either that you are powerless or that you are protected.
Gharar: why this category is structurally opaque
Insurance is the clearest commercial instance of what the governing instrument calls Gharar — excessive, structurally concealed uncertainty in a dealing. This is not an accusation of bad faith. Uncertainty is the product: you are buying a transfer of risk whose terms are necessarily complex because the events they describe are.
But complexity that is necessary and complexity that is convenient are different things, and the second hides inside the first. The practical test: can you state, in one sentence, what would have to happen for this policy not to pay? If you cannot, the concealment is doing work. That question is worth asking before you buy, when you have leverage, rather than at the claim, when you have none.
The document to demand at purchaseNot the summary, not the brochure — the full benefit schedule and the definitions section. The definitions are where the product actually lives. A policy paying 70% under an "any occupation" definition is weaker than one paying 60% under "own occupation," and no headline figure will tell you which you hold.
Preparing the apparatus while you are well
The asymmetry at the heart of this category is timing. The insurer makes the decision in the ordinary course of business, with staff and systems and no urgency. You respond while ill, unemployed or bereaved. Nothing equalises that, but preparation shifts it measurably.
- Keep the policy documents you will need, not the ones you were sentThe full schedule and definitions, dated, in a place someone else could find. A named person who knows where they are.
- Record the claim route before you need itWho to contact, in what form, with what deadline. Deadlines are frequently short and frequently the reason a valid claim fails.
- Start a dated log at the first contactEvery call: date, time, name, what was said, what was promised. This single habit wins more disputes than any argument, because it converts your account from memory into evidence.
- Never accept a decision by phoneAsk for it in writing. A verbal denial is not appealable because it did not, in an evidentiary sense, occur.
- Escalate on a schedule, not on emotionInternal appeal, then the regulator or ombudsman. Each step has a time limit; missing one usually ends the matter regardless of merit.
Where this skill generalises
Everything above transfers. The same four elements — material consequence, automation, absent reasons, absent human appeal — describe a credit refusal, a benefits sanction, a platform deactivation, a tenant-screening rejection and an algorithmic hiring rejection. The vocabulary is the same because the structure is the same.
That is the reason this article exists at MetaX rather than inside an insurance guide. Learning to contest a claim denial teaches a general competence in a specific, high-stakes instance where you have documents, a contract and often a regulator. It is the best available training ground for a skill you will need repeatedly in contexts with far less protection.
The honest limits of this advice
Contestability is not a guarantee. Many denials are correct. The apparatus described here helps you find out which yours is, and ensures a wrong one is not upheld by default. It does not turn an uncovered event into a covered one.
It costs energy you may not have. The realistic advice — log everything, escalate on schedule, never accept a verbal answer — assumes capacity that a seriously ill claimant does not have. This is a real limitation and the answer is to name a person in advance who can act for you.
Some of this is jurisdiction-dependent. The right to human review of an automated decision exists in some regimes and not others. We describe the questions and the structure; the enforcement varies, and we would rather say so than imply a universal entitlement.
And the burden should not be yours. Every technique here is a workaround for a system that should give reasons unprompted. Teaching individuals to contest a defective apparatus is necessary and is not a fix — a point worth holding onto, because the existence of a good workaround is often the reason a structural repair never happens.
Revision log. Sep 2026: first publication. Written as the mastery track for the insurance desk's structural issues, referenced from the "More Details" section of Fiscora's insurance guides. Reviewed annually.