Chief Operating Officer / India
Decisions that cost money first and earned trust afterwards
- Gross direct premium, year to March 2026
- ₹9,420 Cr
- Motor 38 per cent, health 34 per cent, crop 14 per cent, commercial lines 14 per cent
- Median cashless decision, FY2025-26
- 41 min
- Down from 3 hours 10 minutes in FY2021-22, after authorisation moved in-house
- Grievances per 10,000 policies, FY2025-26
- 14
- Down from 31 in FY2021-22, after the grievance panel could revise decisions
Decisions & consequences
01Taking the cashless decision back from three administrators
In 2021–22 every health claim Nilakhya received was processed by one of three third-party administrators, and a patient waiting for cashless approval waited a median of 3 hours 10 minutes. One discharge in nine took more than six hours to approve. Health produced 58 per cent of our grievances, retail renewals had slipped from 81 to 77 per cent, and our data sat in three systems that could not be compared, so overbilling at a hospital used by two administrators was invisible to us. The regulator was also signalling tighter expectations on cashless timelines, which arrived in its 2024 health master circular.
I proposed bringing retail health claims in-house to a hub in Chennai and reducing the administrators for group health from three to one. Our CFO opposed the plan on cost: ₹62 crore to build over two years and about ₹38 crore a year to run, against ₹29 crore of administrator fees saved. The head of our group health business wanted to keep all three, because large corporate clients valued their administrators' help desks on site. One contract required nine months' notice. The board approved the in-house hub for retail and a single group administrator, accepting a net rise in operating cost.
We hired 46 doctors and 380 claims staff, including 140 people from the administrators we were leaving, and moved hospitals across in four waves, starting with the 600 that sent us the most claims. Every rejection or partial approval now needs a medical officer's signature and a reason a patient can follow. In August 2023, when the second and third waves overlapped, 1,900 authorisations backed up in a week, and we returned two states to an administrator for three weeks.
By 2025–26 the median cashless decision took 41 minutes and the median discharge approval 2 hours 5 minutes, against 5 hours 40 minutes before. Retail health retention reached 84 per cent. Medical audit found about ₹41 crore a year of billing outside agreed packages that the earlier arrangement had paid unchallenged. Our policyholder protection committee now sees turnaround by hospital tier every quarter.
I would migrate by region rather than by volume, because volume-based waves left almost every city running two systems at once and hospital desks did not know whom to call. I would also bring the administrators' staff across before the first wave, not during it.
Paying Chennai's flood claims before every survey was finished
In the first ten days of December 2023, after the cyclone rains that flooded Chennai, Nilakhya received 11,900 motor claims from the city, more than nine months of its ordinary volume. Another 1,450 claims came from shops and small businesses. Many cars had stood in water above the dashboard. Garages were flooded themselves, and our surveyors in the city could inspect about 250 vehicles a day. At that rate inspection alone would take seven weeks, while families went without vehicles and storage charges mounted.
I decided that licensed surveyors would assess vehicles with an estimated loss below ₹3 lakh from video and photographs instead of waiting for a visit; that cars submerged above the dashboard would be assessed early for constructive total loss rather than sent for repair; and that owners of likely total losses would receive half the expected settlement on account. Our head of fraud control argued that video assessment would invite inflated claims and old damage. Our chief underwriting officer warned that our catastrophe reinsurance recovery depended on proving that each loss came from the event. Both concerns were legitimate. I accepted a higher risk of overpayment and redeployed 160 staff from Pune and Lucknow, knowing that ordinary claims there would slow.
We engaged 48 additional surveyors from other states, opened three temporary salvage yards on the city's edge and required a geotagged photograph of standing water with every claim to support the event record. One paid claim in eight was audited afterwards.
The median motor settlement was nine days, and 92 per cent of claims were paid within thirty; after the 2015 floods, my previous employer's median had been thirty-one days. The audit of 1,490 paid claims found overpayment on 2.6 per cent, about ₹3.4 crore when extrapolated across the event, and 41 claims went to investigation. Our reinsurers agreed the recovery without dispute. In Pune, ordinary motor settlement slowed from eleven days to sixteen for December.
I would sign salvage yards and surveyor surge pools before the monsoon rather than during the flood, which cost us the first three days. And I would have told customers in Pune and Lucknow why their claims were slower, instead of letting them discover it.
Suspending cashless treatment at 37 hospitals without stranding their patients
In early 2024 our claims analytics showed that 37 network hospitals in four districts of eastern Uttar Pradesh were admitting our policyholders for short stays at more than three times the state's rate. The admissions were mostly for fever and gastroenteritis, the case papers were strikingly similar, and many policies had been sold by the same small group of agents shortly before the first claim. Claims from the cluster had reached ₹96 crore a year at a loss ratio of 142 per cent, and about 41,000 of our policyholders lived there.
I recommended suspending cashless treatment at all 37 hospitals at once while we investigated, rather than one at a time. Our head of retail health sales and the head of our Lucknow hub disagreed. They argued that honest policyholders would lose cashless care close to home, that agents would leave us and that some hospitals would be suspended unfairly. They were right on each point. I judged that investigating hospital by hospital would warn the rest and leave the pattern in place for another year.
We honoured every admission already under way, sent each hospital a notice setting out our findings with a hearing within thirty days, and committed to decide every reimbursement claim from the cluster within seven days. Investigators reviewed 1,100 claim files and visited every hospital. Where the evidence showed fabricated records, at four hospitals, we filed police complaints.
After the hearings, 11 hospitals were reinstated and 26 stayed outside the network. Claims from the cluster fell to ₹51 crore in 2024–25, and we estimate that about ₹38 crore a year of payments for unnecessary admissions has stopped. Grievances from the cluster rose from about 40 a quarter to 310 in the first quarter after the suspension, then fell back by the year's end. Some agents did leave, and retail health sales in those districts were 22 per cent lower in 2024–25.
I would have written to every policyholder in the cluster before the suspension took effect, with the nearest network hospitals and a number to call, not a week afterwards. And I would have suspended in two tranches by strength of evidence. Most of the 11 hospitals we reinstated would have been in the second, and they bore a cost they had not earned.
Letting the grievance team overrule the claims team
When I became COO in 2021, a policyholder who complained about a claim decision was answered by the department that had made it, often by a colleague of the person who had made it. I had run that department for five years without seeing the problem clearly. In 2021–22 we received 31 grievances for every 10,000 policies, and the Insurance Ombudsman made 212 awards against us. Nearly two-thirds of those awards concerned partial health settlements: deductions for items the policy treated as non-payable, or proportionate cuts to a whole bill because the patient had chosen a room above the policy's limit.
I proposed moving the grievance team out of claims to report to me, with a dotted line to the chair of the board's policyholder protection committee, and creating a review panel of senior doctors and claims managers who had no part in the original decision and could revise it. I also proposed that any deduction above ₹10,000 must cite the policy clause in plain words in the settlement letter. Our chief claims officer, who had succeeded me, argued that the panel would second-guess her teams and raise claims cost. Our CFO estimated that cost at ₹22 crore a year. Both were close to right.
In its first year the panel reviewed about 6,800 grievances and revised 19 per cent of the decisions, adding ₹17 crore of payments. The claims team found that hard, and for some months relations between the two teams were poor. We then began returning every revision to the original assessor with its reasoning, and the claims team changed its practice on proportionate deductions without being instructed to.
By 2025–26 grievances had fallen to 14 per 10,000 policies and Ombudsman awards against us to 74. The revision rate fell to 8 per cent, because fewer weak decisions reached the panel. The policyholder protection committee now sees the revision rate every quarter and treats a rise as an early sign of trouble in claims practice.
I would have sent the panel's reasons back to the claims team from the first month instead of the eighth, which would have shortened the resentment considerably. And I would have designed the panel with the chief claims officer rather than presenting it to her.