Checked before
it's paid.
Chasing money back after it's paid is slow and rarely works. So every claim is checked on the way in — and anything that looks wrong is held for a person to decide, not declined by a machine.
Held, not declined. Both go to a reviewer today, and both members are told their claim is being looked at.
6
checks on every claim before payment
0
claims declined by a score alone
2
reviewers before any decline
100%
of flags written to a case file
Claims on this page are illustrative. The checks are the product; what they catch on your book is yours to measure.
Each one has a check, and a place.
Out-patient fraud is rarely dramatic. It's small amounts, repeated — a bill used twice, a strip of vitamins on a medicine bill. These are the patterns, and the point in the flow where each is met.
Someone who isn't on the policy is treated on a member's benefit.
The member scans the desk's code from their own phone; dependants have their own profiles, so nobody signs in as someone else.
At the deskOne bill submitted again a month later — or to a second employer's policy.
Every bill is fingerprinted from its image, number, clinic and date, and matched against every claim already paid.
On the billA ₹1,800 becomes ₹8,100, or a date is moved inside the policy year.
The photo is checked for signs of editing, and the amount is read against what that clinic normally charges for that line.
On the billCosmetics, supplements or a second strip of something billed as medicine.
Each medicine and test on the bill has to be on the prescription it came with, dated before it.
Against the policyA letterhead and a registration number that belong to no one.
The registration on the bill is looked up against the medical council register before the claim moves.
Against the registerTests per visit creep up, or bills cluster just under a cap.
Each network partner is compared with its peers over months. A drift opens a review of the clinic, not of its patients.
Across the networkThe first defence isn't a check at all. Inside the network the member scans in and the wallet pays the clinic directly — there is no paper bill to copy, edit or submit twice.
Every flag becomes a case.
The checks don't decide anything. They fill a queue that people work through — with the signal, the evidence and the clock on every line.
Held claims
4 openNetwork watch
tests per visitA diagnostics partner, Indore
- Peer clinics, same city1.4
- This clinic, April1.5
- This clinic, October3.6
Opens a review of the clinic. Its patients' claims are still decided one by one.
Signals, not scores
A reviewer sees why a claim was held in words — 'duplicate of CLM-19907' — not a number to trust.
The clock is visible
Held money is the member's money. Every case shows how long it has waited and whose desk it's on.
Exportable as it stands
The case file goes to the insurer's fraud unit in one piece, not rebuilt from emails.
Most claims are honest.
A fraud system that treats every member as a suspect costs more in trust than it saves in rupees. These are the rules we run it by.
Four rules
A flag holds money. It never declines it.
No score, model or rule can turn a claim down. Every 'no' has a named reviewer behind it.
The member is told, plainly, and not accused.
“We're checking this bill against one we've already paid” — the reason, and what would clear it.
A decline is read twice.
Before a held claim is declined, a second reviewer reads the case file. Reopened claims go to someone new.
Wrong flags are counted too.
Held claims that turn out clean are tracked against the checks that raised them, and noisy checks are retuned.
What fraud reduction should we expect?
We don't publish one. It depends on your cohort, your plan design and how much of your spend already runs cashless through the network — where most of these patterns can't happen at all. What we commit to is the checks, the case file behind each flag, and a monthly report you can hold against your own baseline.
Won't honest employees get caught up in this?
Some clean claims will be held — that's what a check is. What matters is what happens next: the member is told in plain words why, and what would clear it; nothing is declined without a person reading it; and every wrongly held claim counts against the check that raised it.
How does this fit with the insurer's fraud team?
IRDAI's 2025 fraud-monitoring guidelines ask insurers to run a dedicated fraud monitoring unit. The case file here — the signal, the evidence, who reviewed it and what was decided — is built to be handed to that unit as it stands, rather than re-assembled from emails.
What happens to a clinic that drifts?
The network team reviews the clinic: its billing against its peers, a sample of its cases, a conversation. Members who were treated there are not penalised for it, and their claims are decided on their own merits.
What data do the checks use?
Only what's needed to decide the claim — the bill, the prescription, the member's cover and the history of paid claims. Diagnoses stay with the reviewing doctor. Nothing here is shared with the employer beyond counts.
Stop paying for the same bill twice.
Book a walkthrough and we'll run the checks against a sample of your own claims file.