Insurance Products, Coverage and Claim Triggers

Introduction and Context

Claims management begins long before a claim is notified. It begins with the insurance contract: the product selected, the risks described, the coverage grant, the exclusions, the conditions, and the evidence required to establish a covered loss. A claims professional who cannot interpret policy wording may confuse an event with a covered loss, treat a condition as an exclusion, overlook an endorsement, or approve a payment beyond the policy limit.

Insurance products convert uncertain financial consequences into contractual obligations. The insurer does not generally promise to pay for every adverse event. It promises to respond when specified conditions are satisfied. These conditions vary considerably across property, casualty, motor, health, life, liability, marine and specialty insurance.

For claims and fraud professionals, coverage analysis has two connected dimensions. The first is legal and technical: does the loss fall within the insuring agreement and satisfy policy conditions? The second is evidential and behavioural: are the facts consistent with the policy, the chronology, the claimed amount and the insured’s representations? Fraud diagnostics must never replace fair coverage analysis, but inconsistencies in coverage facts may justify proportionate investigation.

Learning Objectives

After studying this chapter, the reader should be able to:

• interpret insuring agreements, definitions, exclusions, conditions and endorsements;
• distinguish a loss event, a peril, a cause of loss and a claim trigger;
• explain deductibles, excesses, limits, sub-limits, warranties and waiting periods;
• compare coverage structures across major insurance classes;
• apply basic indemnity and benefit calculations;
• identify coverage gaps, evidential issues and potential fraud indicators;
• evaluate the governance, ethical and regulatory implications of claim decisions.

The Architecture of an Insurance Contract

An insurance policy should be read as an integrated document rather than as an isolated schedule or marketing summary. Although terminology varies by jurisdiction and product, most policies contain the following components.

The declarations or schedule

The schedule identifies the parties, insured property or person, policy period, sum insured, premium, deductible, territorial scope and selected extensions. It may also list endorsements that modify the standard wording. A schedule entry can be decisive. For example, a commercial property policy may state that stock is insured for $500,000, while a sub-limit in the wording restricts theft of certain goods to $50,000.

The insuring agreement

The insuring agreement is the insurer’s primary promise. It identifies what is insured, against which risks, during what period, and subject to what limits. Common structures include:

• named-perils coverage, which responds only to listed perils;
• all-risks or open-perils coverage, which responds to accidental physical loss or damage unless excluded;
• specified-benefit coverage, which pays a stated amount when a defined event occurs;
• liability coverage, which responds to legally enforceable liability or specified claims;
• assistance or service coverage, in which the insurer arranges treatment, repair or emergency support.

“All risks” does not mean “all losses.” It means that the initial grant is broad, but exclusions and conditions remain important.

Definitions

Definitions give technical meaning to words that may have an ordinary-language meaning. “Accident,” “pollution,” “flood,” “employee,” “dependent,” “loss,” “vehicle,” “business interruption” and “occurrence” may all be defined. A claims analyst must use the policy definition before relying on everyday language.

Definitions can also create scope limitations. If “flood” is defined as the inundation of normally dry land by rising or overflowing natural water, damage from a burst internal pipe may not qualify as flood, even though the property is wet.

Exclusions

An exclusion removes specified causes, property, persons, circumstances or types of loss from coverage. Exclusions may be:

• peril exclusions, such as war or nuclear contamination;
• loss exclusions, such as consequential loss or depreciation;
• conduct exclusions, such as deliberate acts or use while intoxicated;
• property exclusions, such as wear and tear or defective design;
• temporal or territorial exclusions, such as loss outside the policy period or territory.

The analytical sequence matters. The claimant generally must first show that the loss falls within the insuring agreement. The insurer then relies on an exclusion if applicable. The interpretation and burden of proof depend on applicable law, policy wording and jurisdiction.

Conditions

Conditions regulate the operation of the contract. They may require prompt notice, reasonable precautions, cooperation, proof of loss, preservation of damaged property, disclosure of material facts, or compliance with security requirements. A condition may be a condition precedent to liability, a continuing obligation, or a procedural requirement whose breach has consequences prescribed by law.

A claims professional must distinguish a condition from an exclusion. An exclusion defines what is not covered; a condition describes what must be done for the contract to operate. The distinction can affect causation, remedies and the amount payable.

Endorsements and clauses

An endorsement changes, adds to or deletes wording. It may broaden cover, restrict cover, alter a deductible, introduce a warranty, or replace a standard clause. Endorsements must be read together with the base policy, and precedence clauses should be checked where wording conflicts.

Examples include replacement-cost endorsements, flood extensions, cyber exclusions, named-driver endorsements, business-interruption extensions, agreed-value clauses and machinery-breakdown endorsements.

Claim Triggers and Causation

A claim trigger is the event, condition or legal circumstance that activates potential coverage. Triggers differ by policy type.

Event-based triggers

Property policies often respond to physical loss or damage occurring during the policy period. The trigger may be fire, storm, theft or accidental damage. The critical questions are when the damage occurred, what caused it, and whether that cause is insured.

Occurrence-based and claims-made triggers

Liability policies commonly use either an occurrence trigger or a claims-made trigger.

Under an occurrence policy, coverage is generally connected to when the injury or damage occurs, subject to the wording. Under a claims-made policy, coverage is generally connected to when the claim is first made against the insured and reported in accordance with the policy, often subject to a retroactive date.

A claims-made policy may therefore require analysis of:

1. when the act, injury or damage occurred;
2. when the insured first became aware of circumstances that might lead to a claim;
3. when the third party made a claim;
4. when the insured notified the insurer;
5. whether the event falls after the retroactive date and within the policy period.

The distinction is particularly important for professional indemnity, directors’ and officers’ liability, cyber and medical malpractice insurance.

Loss-discovered and reporting triggers

Crime policies may use discovery wording, responding when the insured discovers a loss, while other policies use a reporting trigger. Discovery wording can create complex questions about the date on which a reasonable insured should have identified the loss.

Benefit triggers

Life and health policies frequently use defined benefit triggers. Death, survival to a specified date, diagnosis of a listed critical illness, admission to hospital, disability or incurrence of eligible medical expenses may activate payment. Unlike indemnity insurance, the amount may not depend directly on the claimant’s actual financial loss.

Causation

Coverage analysis often requires identifying the dominant, effective or proximate cause, depending on applicable law and policy wording. A sequence may include a storm, roof damage, water entry, mould and business interruption. The analyst should separate:

• the initiating event;
• the physical mechanism of damage;
• immediate and remote consequences;
• concurrent causes;
• excluded causes;
• avoidable or pre-existing conditions.

Where insured and excluded causes operate together, the policy may contain a concurrent-causation clause. Courts and regulators may interpret such clauses differently, so legal advice may be necessary for high-value or disputed claims.

Financial Mechanisms in Coverage

Deductibles and excesses

A deductible or excess is the portion retained by the insured. A fixed deductible is subtracted from an otherwise covered loss. A percentage deductible is calculated by reference to a stated base, such as the sum insured or the loss amount. The policy wording must be checked because these mechanisms produce different outcomes.

For a simple fixed deductible, the covered payment before limits is:

where P is the payment before policy limits, L is the covered loss, and D is the applicable deductible. The formula assumes that the loss is covered, valuation is agreed, and no sub-limit or co-insurance adjustment applies.

If a loss is $80,000 and the deductible is $5,000, the payment before other adjustments is $75,000. If the loss is $3,000, the payment is zero under a $5,000 deductible.

Limits and sub-limits

The policy limit is the maximum payable for a covered category during a defined period or for a defined event. A sub-limit is a lower limit applying to a particular item or extension. Limits may be per occurrence, per claim, per person, per vehicle, per location or in the aggregate.

A limit is not necessarily a valuation. A sum insured may be inadequate even when the policy limit is technically respected.

Average, underinsurance and co-insurance

Some property policies apply an average or co-insurance condition when the declared value is below the actual value at risk. A common form is:

where S is the sum insured, V is the value at risk, L is the covered loss, and D is the deductible. This simplified expression assumes a proportional average clause, a single property category, and no separate sub-limit. The payment cannot exceed the loss, hence the min function.

For example, if the value at risk is $1,000,000, the sum insured is $600,000, and the covered loss is $200,000, the proportional payment before deductible is:

600,000 ÷ 1,000,000 × 200,000 = 120,000

With a $10,000 deductible, the payment is $110,000. Some modern policies contain margin clauses, agreed values or designated categories that modify this calculation; the standard formula must not be applied automatically.

Indemnity and benefit principles

Indemnity aims to place the insured, financially, in approximately the position occupied immediately before the insured loss, subject to policy terms. It is not intended to create a profit. Actual cash value may reflect depreciation, while replacement-cost coverage may pay the cost of replacement if specified conditions are met.

Life insurance and many personal accident policies are usually benefit contracts. The policy pays a stated amount when the defined trigger occurs, although medical expense and income-protection products may contain indemnity features.

Coverage Across Major Insurance Classes

Property insurance

Property claim analysis begins with the insured location, interest and physical damage. The analyst should establish the date of loss, cause, ownership or insurable interest, policy period, valuation basis and applicable deductibles. Fire, storm, escape of water, theft and machinery breakdown each involve different evidence.

Business interruption coverage usually requires physical damage by an insured peril and may cover lost gross profit, revenue reduction, continuing expenses and reasonable extra expense. The indemnity period, trends, savings and the policy’s calculation clause are central. A loss of sales without covered physical damage may not trigger a traditional business-interruption policy, although contingent business interruption or non-damage extensions may apply.

Motor insurance

Motor policies may include comprehensive damage, third-party property damage, bodily injury liability, personal accident and roadside assistance. Coverage depends on the vehicle, driver, permitted use, territorial limits and policy conditions. Material issues include unauthorized use, unlicensed driving, intoxication, undeclared modifications, prior damage, salvage value and whether repairs restore pre-loss condition.

Telematics, repair-network data, police records and photographs may support both fair settlement and fraud detection. However, data must be collected lawfully and used proportionately.

Health insurance

Health coverage may be reimbursement-based or benefit-based. Key terms include waiting periods, pre-existing conditions, exclusions, eligible providers, medical necessity, room-rent limits, co-payments, deductibles and annual limits. A claim trigger may be treatment by an eligible provider rather than the diagnosis alone.

Claims teams must protect medical confidentiality. Automated fraud indicators should not treat unusual treatment, chronic illness or demographic characteristics as evidence of fraud without clinical and contextual review.

Life insurance

Life claims require confirmation of the insured event, policy status, identity, beneficiary entitlement and exclusions. Contestability provisions, suicide clauses, non-disclosure and misrepresentation issues may be relevant, but they must be handled according to applicable law and fair-treatment standards.

The date of death is normally central, but policies may also cover terminal illness, critical illness or disability. Documentary evidence may include death certificates, medical records, policy applications and premium history.

Liability insurance

Liability coverage requires more than proof that an accident occurred. The claimant may need to establish duty, breach, causation and legally recoverable damage. The policy may cover defence costs, settlements, judgments or investigation expenses. “Claims-made,” “occurrence,” “retroactive date,” “circumstance,” “consent to settlement” and “allocation” clauses deserve careful attention.

The insurer’s duty to defend may be broader than its ultimate duty to indemnify, depending on jurisdiction and wording. Reservation-of-rights communications should be precise and timely.

Marine insurance

Marine insurance may cover hull, cargo, freight, liability or war risks. Transit claims require analysis of the voyage, attachment and termination of risk, packaging, seaworthiness, bills of lading, survey evidence, salvage and mitigation. Incoterms may allocate commercial risk between buyer and seller, but they do not by themselves determine the insurance contract’s response.

General average involves extraordinary expenditure or sacrifice intentionally and reasonably made for the common safety of a maritime adventure. It should not be confused with an ordinary cargo loss.

Specialty insurance

Cyber, political-risk, trade-credit, agriculture, aviation, energy, construction and parametric products require product-specific interpretation. Cyber policies may distinguish security failure, privacy breach, business interruption, dependent business interruption, ransomware and regulatory response. Parametric insurance pays when an objectively measured parameter reaches a stated threshold, such as wind speed or rainfall, rather than indemnifying the exact physical loss. Basis risk—the difference between the payment and the actual loss—is therefore a central issue.

Warranties, Conditions and Duties

A warranty may require the insured to maintain a particular state of affairs or perform a specified act, such as maintaining an alarm system, using approved locks, storing hazardous materials safely or complying with a vessel’s trading limits. The legal effect of breach varies significantly across jurisdictions. Claims professionals should avoid treating every breach as an automatic reason to reject a claim.

The appropriate analysis asks:

1. What exactly does the clause require?
2. Is it a warranty, condition, representation or exclusion?
3. Was it breached, and when?
4. Did the breach relate to the loss?
5. What remedy does applicable law and the policy provide?
6. Was the requirement clearly communicated to the policyholder?

Fair claims practice requires giving the insured a meaningful opportunity to provide information, explaining the decision, and distinguishing lack of evidence from evidence of non-coverage.

A Structured Coverage-Analysis Framework

A practical framework is the T-CLEAR sequence:

• T — Trigger: What event, condition or claim activates potential coverage?
• C — Contract: Which policy, period, schedule and endorsement apply?
• L — Loss: What was physically, financially or legally lost?
• E — Exclusions and exceptions: Does an exclusion apply, and is there an exception or buy-back?
• A — Amount: What valuation, deductible, limit, sub-limit or average clause applies?
• R — Requirements: Were notice, cooperation, mitigation and proof obligations satisfied?

This sequence is not a substitute for legal interpretation, but it prevents common analytical errors. It also creates an auditable claim file.

Applied Case: Commercial Fire and Business Interruption

The following is a fictional illustrative case designed for educational purposes.

Background

Northstar Foods operates a refrigerated distribution warehouse. Its property policy covers fire and specified accidental damage. The schedule lists building and contents cover of $2 million, a $25,000 deductible, and a business-interruption extension with a 12-month indemnity period. A separate endorsement applies an average clause to stock.

A fire begins in an electrical control panel on 10 March. The building is damaged, stock is destroyed, and operations stop for four months. Northstar claims $900,000 for stock, $400,000 for building repairs and $650,000 for lost gross profit and additional expenses.

Problem

The adjuster must determine whether the fire is covered, whether the stock is underinsured, whether business interruption is triggered, and whether a maintenance warranty was breached. The insured’s stock records also contain unexplained increases shortly before the fire.

Analysis

The trigger is physical fire damage during the policy period. The policy covers fire, and no initial fact suggests that the fire was deliberately caused by the insured. The electrical-panel warranty requires documented annual inspection. Records show that an inspection was six months overdue, but the wording and applicable law must be examined before deciding the consequence. The overdue inspection is not automatically proof that the fire was excluded.

The stock’s actual value immediately before the fire is established at $3 million, while the sum insured is $2 million. If the average clause applies, the proportional factor is:

2,000,000 ÷ 3,000,000 = 2 ÷ 3

The covered stock loss before deductible is therefore approximately:

900,000 × 2 ÷ 3 = 600,000

The deductible and any applicable stock sub-limit must then be applied. The calculation depends on whether the deductible is shared across building and contents or applied separately.

Business interruption is potentially triggered because insured physical damage caused the interruption. The adjuster must reconstruct expected sales, variable costs, trends, saved expenses, mitigation, seasonality and the indemnity period. The claimed $650,000 cannot be accepted merely because it appears in the claim form.

The unexplained stock increase is a fraud-diagnostic lead, not a conclusion. The claims team should reconcile purchase invoices, warehouse records, inventory systems, production data, sales records and valuation methods. It should interview relevant staff, preserve digital evidence, and give Northstar an opportunity to explain discrepancies. A forensic accountant may distinguish genuine stock accumulation from inflated records.

Approach and outcome

The insurer appoints a fire investigator, quantity surveyor and forensic accountant. The fire investigator identifies an accidental electrical failure. The policy wording permits a proportionate reduction for underinsurance but does not make the inspection clause a basis for total forfeiture absent a material connection established under applicable law. The stock valuation is adjusted using verified records. Business-interruption figures are recalculated after removing saved costs and unsupported projected growth.

The final settlement consists of the covered building and stock loss after the applicable average clause and deductible, plus verified business-interruption loss within the policy limit. The suspicious inventory entries are referred for further investigation, but payment is not delayed unreasonably in respect of independently verified amounts.

Learning points

• A claim trigger does not determine the final amount payable.
• Underinsurance and deductibles must be calculated from the wording.
• A warranty issue requires legal and factual analysis.
• Fraud indicators require investigation, corroboration and procedural fairness.
• Partial, undisputed amounts may be paid while disputed components are examined.

Fraud Diagnostics and Evidence

Fraud diagnostics should be integrated with ordinary claims governance. Useful indicators include late notification without explanation, inconsistent timelines, altered documents, duplicate invoices, unusual claims immediately after policy inception, implausible repair patterns, financial distress, inflated inventories, staged accidents and links among claimants, providers or repairers.

No indicator proves fraud. Indicators should generate hypotheses that are tested against independent evidence. A strong investigation uses triangulation: policy data, external records, physical inspection, financial records, digital metadata, witness evidence and expert opinion. The investigation must comply with privacy, evidence, anti-discrimination, consumer-protection and reporting requirements.

Artificial intelligence can assist with document classification, anomaly detection, image comparison, network analysis and triage. Its limitations include biased training data, false positives, opaque scoring, adversarial manipulation and difficulty interpreting context. Human oversight, explainability, model validation, access controls and appeal mechanisms are essential. Automated denial based solely on an unexplained score is poor governance and may breach regulatory expectations.

Challenges, Regulation and Professional Ethics

Coverage disputes commonly arise from ambiguous wording, conflicting endorsements, changing technology, concurrent causes, uncertain valuation, late notice and cross-border events. Regulatory regimes increasingly require clear communication, timely claims handling, fair treatment of vulnerable customers, data protection and documented reasons for decisions.

Claims professionals should:

• use the current policy version and verify endorsements;
• separate facts, assumptions, expert opinions and conclusions;
• record the reasoning behind coverage and quantum decisions;
• communicate reservations and requests for information clearly;
• protect confidential personal and commercial information;
• manage conflicts of interest among adjusters, experts and service providers;
• escalate legal uncertainty and high-severity claims;
• review declined claims through quality assurance and complaint processes.

The central ethical principle is proportionality. Investigation should be sufficiently rigorous to protect the insurance pool, but not so intrusive or adversarial that it unfairly burdens legitimate claimants.

Contemporary Developments and Best Practice

Insurance products increasingly combine traditional indemnity with real-time data, embedded insurance, parametric triggers, usage-based motor cover, telemedicine, connected-home devices and cyber services. These developments improve speed and risk segmentation but create new questions about data ownership, algorithmic fairness, trigger reliability and responsibility for sensor failure.

Best practice is to maintain a product-specific coverage matrix containing the trigger, insured interest, exclusions, conditions, limits, deductibles, evidence requirements and escalation points. Claims teams should test the matrix against realistic scenarios before a product is launched. Periodic wording reviews should consider litigation trends, emerging perils, climate-related losses, supply-chain dependencies and changes in regulation.

Chapter Summary

Insurance coverage is a structured contractual response, not a promise to pay every loss. Sound claims management requires integrated reading of the schedule, insuring agreement, definitions, exclusions, conditions and endorsements. Claim triggers vary: property may depend on physical damage, liability on occurrence or claims-made wording, life on a defined event, health on eligible treatment, and specialty products on highly specific contractual parameters.

Deductibles, limits, sub-limits, average clauses and valuation rules determine the amount payable after coverage is established. The T-CLEAR framework—Trigger, Contract, Loss, Exclusions, Amount and Requirements—provides a disciplined method for analysis. Fraud diagnostics should identify and test inconsistencies rather than predetermine dishonesty. Technology and AI can improve claims operations, but human oversight, privacy, transparency and procedural fairness remain essential.

Key Terms

Claim trigger: The event, condition or circumstance that activates potential policy coverage.

Insuring agreement: The part of the policy stating the insurer’s primary promise to pay or perform.

Exclusion: Wording that removes specified risks, property, conduct or losses from coverage.

Condition: A contractual requirement governing how the policy operates.

Warranty: A contractual promise concerning a specified state of affairs or action.

Endorsement: A document or clause that modifies the base policy wording.

Deductible: The amount retained by the insured before the insurer pays.

Excess: A policy amount that the insured must bear; often used synonymously with deductible.

Policy limit: The maximum amount payable under a stated coverage.

Sub-limit: A lower limit applying to a particular category or extension.

Indemnity: Financial restoration, subject to policy terms, without allowing profit from loss.

Benefit contract: A contract paying a stated amount when a defined event occurs.

Occurrence trigger: Coverage linked principally to when injury or damage occurs.

Claims-made trigger: Coverage linked principally to when a claim is first made and reported.

Retroactive date: The earliest date of acts or events that may be covered under certain claims-made policies.

Average clause: A proportional underinsurance provision reducing payment when the sum insured is inadequate.

Concurrent causation: A situation in which more than one cause contributes to a loss.

Basis risk: The difference between an index-based insurance payment and the insured’s actual loss.

Fraud indicator: A fact or pattern that justifies further inquiry but does not itself prove fraud.

Insurable interest: A legally recognized interest in the subject matter of insurance.

Review Questions

5 Short Answer Questions

1. What is the difference between an exclusion and a condition?
2. Why does an all-risks policy not cover every possible loss?
3. Define a claims-made liability trigger.
4. What is the purpose of a deductible?
5. Why must endorsements be read with the base policy?

5 Long Answer Questions

1. Explain the architecture of an insurance policy and show how its components interact during claims assessment.
2. Compare occurrence-based and claims-made liability coverage, including the importance of notification and retroactive dates.
3. Discuss how limits, sub-limits, deductibles and average clauses affect claim settlement.
4. Compare claim triggers and evidence requirements in property, health, life, motor and marine insurance.
5. Explain the ethical and governance requirements for using AI in fraud diagnostics.

3 Analytical / Application Questions

1. A warehouse is insured for $600,000, its value at risk is $1,000,000, and it suffers a covered loss of $200,000. Ignoring the deductible, calculate the payment under a proportional average clause and explain your assumptions.
2. A professional indemnity claim is made in June, but the insured first received a threatening letter in February and the policy has a retroactive date of January. Identify the coverage questions that must be investigated.
3. A motor claimant reports theft two days after purchasing a policy, provides inconsistent keys, and submits an invoice from a repairer linked to a family member. Design a proportionate investigation plan that protects both the insurer and the claimant.

Multiple Choice Questions with Answers

Question 1

Which policy component states the insurer’s primary promise to pay or perform?

A. The insuring agreement

B. The deductible clause

C. The claims diary

D. The salvage report

Correct Answer: A. The insuring agreement

Question 2

A fixed deductible of $5,000 applies to a covered loss of $80,000. Ignoring other adjustments, what is payable?

A. $75,000

B. $80,000

C. $85,000

D. $5,000

Correct Answer: A. $75,000

Question 3

Under a claims-made liability policy, which date is generally central to activation of coverage?

A. The date the claim is first made and properly reported

B. The date the original building was constructed

C. The date the premium was advertised

D. The date the insurer was incorporated

Correct Answer: A. The date the claim is first made and properly reported

Question 4

What is the principal effect of a sub-limit?

A. It sets a lower maximum for a specified category or extension

B. It eliminates every policy exclusion

C. It guarantees replacement cost in all cases

D. It transfers ownership of the insured property

Correct Answer: A. It sets a lower maximum for a specified category or extension

Question 5

A property worth $1,000,000 is insured for $600,000 and suffers a covered loss of $200,000. Under the simplified proportional average formula, what is payable before deductible?

A. $120,000

B. $200,000

C. $300,000

D. $600,000

Correct Answer: A. $120,000

Question 6

Which statement best describes an insurance fraud indicator?

A. It proves fraud conclusively

B. It justifies further proportionate inquiry

C. It automatically voids the policy

D. It permits denial without evidence

Correct Answer: B. It justifies further proportionate inquiry

Question 7

What is a typical trigger for traditional business-interruption coverage?

A. Any decline in sales, regardless of cause

B. Covered physical damage causing interruption, subject to wording

C. A competitor opening nearby

D. A change in the company logo

Correct Answer: B. Covered physical damage causing interruption, subject to wording

Question 8

What does basis risk mean in parametric insurance?

A. The difference between the index payment and actual loss

B. The deductible for a liability claim

C. The cost of marine salvage

D. The insurer’s annual licensing fee

Correct Answer: A. The difference between the index payment and actual loss

Question 9

Which document commonly identifies the policy period, sum insured and selected endorsements?

A. The policy schedule or declarations

B. A witness statement only

C. A repair invoice only

D. A weather forecast

Correct Answer: A. The policy schedule or declarations

Question 10

Why should an overdue warranty requirement not automatically be treated as proof that a claim is excluded?

A. Its wording, legal effect and connection to the loss must be examined

B. Warranties never have contractual consequences

C. All warranties apply only after settlement

D. An insurer may ignore every policy condition

Correct Answer: A. Its wording, legal effect and connection to the loss must be examined