Gatekeepers
Published 1 October 2026
The commission in this file exists so that you will pay the premium - every rupee of it. Every argument in every sales pitch is that the insurer will be there when it matters. So here is the moment it matters: the claim. The industry's defense of itself begins and ends with one statistic - individual death claims, 97.82% paid. It is a real number, from the regulator's own report. It is also the best number in the entire book, chosen with the same care as everything else the machine presents, and the rest of the claims ledger does not look like it.
Start with the honest crown jewel. In 2024-25, life insurers paid 97.82% of individual death claims by number. When someone dies with an in-force policy, the industry pays, almost every time - and it should be said plainly, because it is the foundation on which the whole edifice of trust is built, and because the widow who receives that cheque is receiving the one thing this entire apparatus exists to deliver. The fine print is in the denominator: claims are counted among policies in force and claimed upon. The policies that lapsed - the third of the book that dies by the thirteenth month - are not in it. The industry's best statistic excludes the industry's worst outcome.
Health insurance, where most families actually meet the claims process, runs a different ledger. Of health claims decided in 2024-25, 87% were settled, 8% were repudiated - refused outright - and 5% sat pending. One in eight decided claims, refused or stuck. On Rs.94,247.60 crore of claims, the refused 8% is not a rounding error; it is thousands of crores of hospital bills that insured families were told, at the worst moment of their lives, they would have to pay themselves. The repudiation reasons live in the fine print: the pre-existing condition not disclosed, the waiting period not run, the treatment classified as something else. Each refusal may be individually defensible. In aggregate they are a business model with a rejection rate, and the rejection rate is priced into the premium you were charged for "cashless, worry-free cover."
The man between you and the cheque
Most health claims are not even decided by your insurer. They pass through a third-party administrator - a TPA - an outsourced claims processor that handled 69% of health claims by number in 2024-25. The TPA's job description, in the honest version, is to apply the policy's exclusions to the hospital's bill, line by line, and approve the smallest defensible payment. It is paid by the insurer, measured on throughput, and invisible to you until the moment it matters. The hospital's billing desk, the insurer's claims team, and the TPA's adjudicator are three professionals negotiating over your emergency; you are the fourth party, sedated.
When the refusal comes, the appeal ladder is long by design: the insurer's grievance cell, then Bima Bharosa, then the ombudsman, then the consumer courts - each step measured in months, each staffed by people who read policy wordings for a living, against a family that reads one for the first time at the worst time. The ombudsman route is free and genuinely independent, and it is also the end of the line for most people: the same annual report that counts 2.58 lakh grievances a year counts how few ever reach a hearing. The machine does not need to win every claim. It needs every claim to be expensive enough to fight that most refusals are simply absorbed - into loans, into savings, into the quiet ledger of the Indian middle class, where they appear not as scandals but as hospital bills, paid.
The three-year shield and its paperwork
Buried in the claims appendix is one of the few laws in this file written clearly for the customer, and it is worth knowing by name: Section 45 of the Insurance Act, 1938. The regulator's own appendix footnote defines repudiated claims as those that cannot be considered under Section 45 - the provision that bars a life insurer from questioning a policy on any ground, including misstatement in the proposal form, once three years have passed from issue. After three years, the promise is meant to be unconditional. Before them, the insurer may dig through the original application for the undisclosed diabetes, the misstated income, the smoking habit the agent himself ticked "no" to close the sale - and repudiate. The agent who filled the form is, by then, statistically long gone from the industry; the widow holds the paperwork he made.
The scale of what survives the process is also in the public record, and it is the industry's best page. LIC alone reports settling 229.31 lakh claims in 2024-25 - death claims, maturities, surrenders, annuities, the entire outflow of the giant, counted one by one. However this file has read the machine, that number is the counterweight: a quarter of a billion promises kept in a single year by a single insurer. The argument has never been that the machine pays nothing. It is that the machine pays exactly what it must, to exactly the customers who survived its filters - the lapses, the surrenders, the repudiations, the attrition of the appeal ladder - and prices the filters into the premium of everyone still inside. A claims record is a defense only when the road to it did not cost the customer the policy first.
The adjudicator's incentive
The TPA's business model completes the picture, and it is visible in the structure rather than any scandal. A third-party administrator is paid by the insurer per policy or per claim processed - a fee for throughput. It is not paid a percentage of what it saves the insurer, which is the industry's standard answer to the obvious conflict; but it competes for insurer contracts, and the metric on which it competes is visible in every tender: turnaround time, network management, and cost containment, a phrase whose operational meaning is the ratio between what hospitals bill and what claims get paid. An adjudicator that pays every bill in full does not win renewals. The incentive is structural, not sinister - no villain is required, which is exactly the problem. The family at the desk experiences the structure as a person: the voice on the helpline reading exclusions from a screen, empowered to say no and required to escalate every yes.
The regulator supervises TPAs as licensed intermediaries, inspects them, fines them - the Royal Sundaram order in this year's penalty annexure touches the claims-closure practices that sit at the heart of the model - and yet the basic arrangement endures, because it is genuinely useful to the insurer: claims processing is expensive, specialized, and reputationally radioactive, and outsourcing it buys scale plus distance. When the denial goes wrong, the insurer can point at the TPA, the TPA can point at the policy wording, the policy wording can point at the disclosure the agent never made, and the agent - statistically - has already left the industry. Four parties, four doors, and the family's claim dying of old age in the corridor between them. That corridor is not a flaw in the machine's design. Measured by outcomes, it is one of the design's most reliable components.