Why More Indians Are Leaving Insurance Apps for Human Advisors

App-first insurance solved buying and left claiming untouched. Rising mis-selling complaints, a stagnant 3.7 per cent penetration rate and a consistent pattern in public reviews explain why a growing number of Indian policyholders now want a person, not an interface.

For a decade the assumption in Indian insurance was that the intermediary was the problem. Agents mis-sold, took commission, disappeared after the sale. Remove them, put the whole thing in an app, and the customer wins. A great deal of that thesis was correct, and the buying experience in India improved beyond recognition because of it.

What has become clear more recently is that digitisation solved the easy half. Buying was the easy half. Claiming was always the hard half, and it is the half where an app has the least to offer — because a contested claim is not a transaction to be processed, it is an argument to be made.

The numbers that frame the problem

IRDAI’s annual report for 2024-25 makes uncomfortable reading for the "technology alone fixes this" thesis. Insurance penetration in India was static at 3.7 per cent of GDP, against a world average nearer 7.3 per cent — with life penetration slipping to 2.7 per cent and non-life flat at 1 per cent. Fifteen years of digital distribution has not moved the number.

The protection gap — the share of economic loss that is uninsured — remains in the region of 70 to 80 per cent across most segments. And mis-selling, the specific problem digital distribution was supposed to eliminate, was named by IRDAI in that same report as a significant ongoing concern. Grievances categorised under unfair business practices rose to 26,667 in FY25 from 23,335 in FY24, reaching 22.14 per cent of all complaints against life insurers, up from 19.33 per cent.

Penetration is flat, the protection gap is 70 to 80 per cent, and mis-selling complaints are rising. Whatever the app-first decade fixed, it was not the thing that keeps Indians underinsured.

What people actually say when they leave

Read enough reviews on Trustpilot, MouthShut and the app stores, or enough threads on Indian personal finance forums, and the complaints converge on a small number of themes. They are strikingly consistent across companies, which is the point — this is a pattern in the model, not an indictment of any single brand.

• Buying took four minutes; claiming took four weeks and a great deal of chasing.
• There is no one to escalate to — a ticket number, a chatbot, and a queue, but no named human who owns the outcome.
• Status updates have to be pulled rather than pushed; nobody proactively says what is happening.
• The exclusion that mattered was in the wording all along, and nobody flagged it at purchase because nobody read it aloud.
• Grievance responses arrive as templates, and follow-ups go unanswered.

Public reviews skew negative by construction — satisfied customers rarely write them — and every insurer and platform in India accumulates angry ones. What makes this pattern worth attention is not its volume but the fact that it clusters so tightly around a single missing element rather than around price, product or speed.

Why the app model struggles precisely here

This is not incompetence, and it is not indifference. It is what the model is shaped to do. A self-serve funnel is optimised for the high-volume, standard path: a healthy buyer, a standard product, a clean claim. That path is genuinely well served, and for most customers most of the time it works.

A contested claim is the opposite kind of problem. It is low-volume, non-standard, and it turns on the specific interaction between a clause and a set of medical or accident facts. Someone has to read the rejection letter against the policy wording, decide whether the insurer’s ground actually holds, assemble evidence, and put a written case. That work does not automate well, and — critically — it generates no revenue in a model funded by the next sale.

There is a second structural point. Most of what determines whether a claim succeeds is decided years earlier, at proposal stage: whether pre-existing conditions were disclosed properly, whether the sum insured and room rent category were sensible, whether the waiting periods were understood. A frictionless purchase flow is optimised to reduce the number of questions asked at exactly the moment when asking more questions is what protects you.

What is actually shifting

It would overstate the case to say Indians are abandoning insurance apps — they are not, and app-based purchase continues to grow. The more accurate description is that the market is separating two things it had bundled together: the transaction, and the advice.

• People are buying simple, standard cover on apps and keeping it there, because for that job the app is genuinely better.
• They are seeking human help for the complicated cases — pre-existing conditions, elderly parents, family floaters, business exposures.
• They are paying for independent reviews of policies bought elsewhere, which the seller has no incentive to provide.
• They are looking for someone specific when a claim is rejected, having discovered that a support queue is not a substitute.
• Advisory-first propositions have grown quickly on exactly this demand, including brands built on an explicit no-spam-calls promise.

That last point matters. The rise of advisory-first digital brands in India is itself evidence for the argument — companies competing on the quality and honesty of a human conversation are winning share, which is not what you would expect if the intermediary had really been the problem all along.

What to do if this describes your situation

None of this is an argument for cancelling policies or switching insurers, both of which usually cost you continuity benefits and accrued waiting periods for no gain. The useful responses are smaller and mostly free.

• Read your own wording before you need it — room rent limit, co-payment, sub-limits, waiting periods still running, and the exclusions most likely to apply to you.
• Find your completed proposal form and keep it with the policy. If disclosure is ever disputed, that document is the evidence, and most people cannot produce it.
• Get a policy review from someone who did not sell it to you, particularly if you hold several policies from different sources that nobody has looked at together.
• Change the servicing intermediary if yours has been unreachable. This costs nothing, does not touch your policy, and does not reset waiting periods or no-claim bonus.
• If a claim has been rejected, get the rejection in writing with the exact clause cited, and act on it — most people simply give up at this point, which is what the process quietly relies on.

The honest version of the argument

We are a fee-based advisory, so the conclusion that people need advisors is convenient for us. The version worth defending is narrower and more defensible: technology fixed distribution, and distribution was never the binding constraint on whether Indians end up properly covered and properly paid.

The binding constraints are that people do not understand what they bought, and that nobody is contractually on their side when the insurer says no. An app can fix the first with better design. Only a person who is paid to be on your side can fix the second — and whether that person is paid by you or by the insurer determines how reliably they are on your side when it costs something.

Related: what the different intermediary categories actually mean, a balanced look at the no-agent digital insurer model, and how to read insurer complaint data properly.

If you are at the claim stage already: RiskPe claim recovery, rejected health claim guide, or what to do when cashless is denied at the hospital desk.