Charging a Claims-Servicing Fee on TPA-Heavy Accounts

Every employee benefits book has one. A group health account, somewhere between eight hundred and four thousand members, placed at a perfectly ordinary rate, that consumes more of the firm's week than three larger property programmes combined.

The account is not badly placed. The premium is right, the cover is right, the insurer is fine. What makes it expensive is a party that was never in the commission conversation: the Third Party Administrator running the claims. A TPA-administered account generates work with no equivalent anywhere else in a commercial book:

• A cash deposit account that runs dry on a Friday and strands three admissions.
• Pre-authorisation queries that arrive from the hospital, go to the TPA, get partially declined, and land on the broker's account manager because the member's spouse is standing at a billing counter in Indore.
• A network hospital that has quietly stopped honouring cashless for this TPA and told nobody.
• Enrolment files the HR team sends in a format the TPA portal rejects, monthly, for four months.

None of that is claims advocacy as the firm's fee schedule understands it. It is administration of somebody else's operating problem, performed by the broker because the broker is the only party in the chain that answers the phone. The question here is narrow: whether that workload justifies a separate invoice, and how to construct one that survives an engagement-letter review without brushing against Section 41 of the Insurance Act, 1938.

The most useful thing a broker can hold in their head is that the TPA is a separate licensed entity, engaged by the insurer, and not a department of it. That separation is the source of nearly every hour the broker loses.

The contract the broker can enforce is with the insurer. The service standards negotiated at placement sit in the insurer's proposal. The party performing the service is a third entity with its own contract, volumes, staffing and commercial pressures, none of which the broker signed or can see. When the TPA underperforms, the broker's only real lever is to escalate to the insurer and ask it to lean on its own vendor. Two hops on every issue, and the second hop is invisible.

The TPA's incentives are not the insurer's, and neither are the client's. A TPA is typically remunerated as a function of the business it administers, not of how satisfied any member is. Its cost base is its claims processors and its call centre, so every marginal query answered is a marginal cost. An account manager who calls four times about one pre-authorisation is, on the TPA's ledger, a cost centre. From the client's perspective that broker is doing their job. Both are correct, which is why the friction never resolves.

The cash deposit account, universally the CD account, is where TPA administration becomes a broker problem most visibly and most avoidably.

The mechanism is simple. On many group health arrangements, particularly self-funded or partially self-funded ones, an amount is held against which the TPA settles cashless claims as it authorises them. Hospital bills draw down the balance. The balance is replenished. When it hits zero, cashless stops.

The failure is a monitoring failure, and the monitoring is nobody's job. The TPA knows the balance in real time but has no obligation to forecast it, and its notification, where one exists, is automated, addressed to a shared mailbox, and read by nobody. The HR team has never been told the account exists, or was told once at inception by someone who has since left. The client's finance team treats replenishment as an ordinary payable and processes it on its ordinary cycle, which is not the cycle a stranded admission runs on. And the insurer is not operationally close enough to notice.

So the broker notices, at the point of failure, which is a call from a member's family, and then spends a day arranging an emergency transfer and getting the TPA to lift the block.

A firm that does this properly reads the balance on a fixed cadence, computes a burn rate from the account's recent draw-down, projects the exhaustion date, and raises replenishment with finance on a lead time matching their payables cycle. It reconciles the TPA's draw-down statement against the claims actually authorised, because the two do not always agree and the difference is the client's money.

That is a recurring, skilled, finance-adjacent task performed monthly for the client's benefit. It is not placement and it is not claims advocacy. It looks exactly like a service a firm would invoice for if it were performed by anyone other than a broker already paid a commission for something else.

The second cost centre is pre-authorisation, and the honest description is not flattering to the industry.

A cashless admission runs a sequence: the hospital's insurance desk raises a request against the member's identity, the TPA's medical team assesses it against the policy terms, and an approval, a partial approval, a query or a denial comes back. When it works the broker never hears about it. The traffic that reaches the broker is the residue, and the residue is not random. It clusters:

1. Identity and eligibility mismatches. The member was added in a mid-month endorsement that has not propagated to the TPA's master. The dependant's date of birth does not match the hospital's record. The member left and was never deleted. Each is an enrolment-data problem surfacing at the worst possible moment, and unpicking it means reconciling the client's HR master against the TPA's.
2. Partial approvals the family reads as a denial. The TPA approves against the room category the member is entitled to, the hospital admitted them a category higher, and the difference is a proportionate deduction discovered at discharge. The broker's job is translation, under emotional pressure, of a policy term the member never read.
3. Queries that stall. The TPA asks the hospital for a document, the hospital's desk is understaffed, nobody follows up, the request sits. The broker becomes the follow-up function for a conversation it is not a party to.
4. Genuine coverage disputes. The smallest category, and the only one resembling what the firm's claims practice was built for.

Categories one through three are administration. They scale with member count and with the client's HR data hygiene, and they arrive without warning at 8pm. Category four is the work the commission arguably bought.

Now the arithmetic.

Group health placements carry commission in the vicinity of the commercial liability band, roughly 10 to 15 percent of premium, with realised yields on retail health and miscellaneous retail running as high as 15 to 20 percent where variable components are included. Healthy-looking next to the 8 to 10 percent realised yield on a large property programme.

They do not look healthy once the servicing hours are attributed. The property programme's yield funds a renewal submission, a placement negotiation, a policy check and a handful of endorsements. The group health account's yield funds all of that plus the CD account, plus the pre-authorisation residue, plus enrolment reconciliation, plus a monthly report the TPA's export will not produce cleanly, plus the network hospital problem.

The right way to test it is not by benchmark. It is by the firm's own cost:

• Take one representative TPA-heavy account. Log every touch for a full quarter, by category, with the person and the minutes.
• Apply the firm's loaded cost per hour for each grade involved.
• Subtract the total from the account's commission.
• Do the same for a comparable-premium account on a line without a TPA in the chain.

Most firms that run this honestly discover two things. The TPA-heavy account's contribution is a fraction of what its yield implied, and on the worst accounts it is negative. And the cost is concentrated in a few accounts, so the book-level average hides it completely.

The placement commission is consideration for effecting the placement. On a TPA-administered account the broker also performs an ongoing vendor-management and member-support function for a third party it did not choose. It is not that the commission is too low. It was priced for a different job.

Network hospital management. Members care about one thing: whether the hospital nearest them will admit them cashless. That depends on the TPA's empanelment, which changes without notice, for commercial reasons between the TPA and the hospital that neither will explain to a broker. A hospital can be listed on the published network and, in practice, be refusing cashless for that TPA over a payment dispute. The work is maintaining a view of which hospitals in the client's actual geographies honour cashless for this TPA today, flagging gaps before a member finds them, and negotiating case-by-case admissions where the gap has already bitten.

Enrolment churn. A scheme with meaningful attrition generates a continuous stream of additions, deletions, dependant changes and mid-term corrections. HR produces these in whatever format its systems emit; the TPA accepts the format its portal demands. The broker sits between the two, and on a high-churn account that is a recurring reformatting-and-reconciliation job with a hard deadline, because an unpropagated addition becomes a pre-authorisation denial two weeks later.

What unites all of it is the test that also makes a fee defensible: would this work exist if the account were placed and then left alone? For the property programme, largely no. Here, all of it would exist, all of it would fail, and the client would feel every failure. It is the account's operating requirement, and somebody pays for it whether or not anybody invoices it.

The IRDAI (Insurance Brokers) Regulations, 2018 permit a broker to charge a client fees for risk management services and claims consultancy under a written agreement, provided the fee is for work distinct from what the placement brokerage already remunerates. That proviso is the whole design constraint, and on a TPA-heavy account it is satisfiable in a way it often is not elsewhere, because the distinctness is factual rather than argued: the work exists because a third-party administrator sits in the chain.

1. Scope it to named workstreams, not to a relationship. Name the deliverables: cash deposit monitoring and replenishment forecasting to a stated cadence, reconciliation of the TPA draw-down statement against authorised claims, enrolment file reconciliation between the HR master and the TPA master, an escalation function for pre-authorisation queries, network-availability verification for stated locations, and a monthly reconciled utilisation report. A scope of "claims support" is not a scope; it is what the client believes commission already bought.
2. Price from the firm's cost, not from a benchmark. Log every servicing touch on a representative account for a quarter, categorised, with the grade of person and the minutes. That gives hours by grade; loaded cost per hour gives the base; margin is a decision. A fee derived this way survives procurement's only real question, some version of "why this number." A fee copied from another firm's schedule does not.
3. Invoice the client, with GST, against a separate engagement letter and ledger code, and disclose the commission alongside it. Not netted against premium, not adjusted through the insurer. A client asked to pay a fee will ask what the firm earns on the placement; the firm that answers cleanly keeps the fee, and the firm that deflects loses both.

The Section 41 boundary. Section 41 prohibits offering any rebate of commission or of the premium shown on the policy as an inducement to take out or renew a policy, with a fine that may extend to INR 10 lakh, and it reaches the policyholder who knowingly accepts as well as the intermediary. A servicing fee charged to a client is money flowing the other way and is not a rebate. But the near-miss is common: a firm that offers to waive the fee on renewal, set it against premium, or credit it back has stopped charging for a service and started discounting the placement. The fee must be a real charge for real work, invoiced and collected.

When not to charge. Where no TPA sits in the chain, the argument does not exist and should not be manufactured. And where the client is small enough that the fee conversation costs more relationship than the work costs the firm, reprice the placement or decline the renewal rather than invoice.