IRDAI Inspections of Broker Commission: Findings, Penalties and Remediation

Until 5 February 2026, a broking firm that picked up an adverse observation on commission had a natural place to settle it: the renewal file. Registrations ran in cycles, so an unresolved issue got cleaned up before renewal or dealt with at it. The renewal cycle was both a deadline and a safety valve.

The Sabka Bima Sabki Raksha (Amendment of Insurance Laws) Act, 2025 removed it. Intermediary registrations have been perpetual since 5 February 2026. There is no scheduled moment at which the regulator reconsiders whether your firm should hold its licence, and none at which you get to demonstrate that you fixed something. The licence became a durable asset on that date, and an asset that can be taken away only one way.

The thesis of this post follows, and is worth stating without decoration: enforcement is now the only route by which a broking licence is lost. Every consequence that used to be spread across the renewal decision has been concentrated into the enforcement process.

So the question that used to be "will this survive renewal scrutiny" has become "what happens if an inspection team pulls this file." This post is about the second: what an inspection is, how a finding becomes a notice, how a notice becomes an order, and what closure requires. The returns side of the same problem is covered separately in building commission MIS that reconciles to IRDAI returns; this post picks up where that discipline fails.

The distinction is basic. A return is something you compile, sign and file; you control the framing and the aggregation. An inspection is an examination on your premises, in your systems, against records you did not curate first, by people whose job is to find the gap between what your return said and what your ledger shows. The practical difference shows up in three places.

Sampling, not totals. Your returns present aggregates. An inspection team works from samples chosen by criteria it need not share in advance: large accounts, hard-won accounts, accounts where remuneration is unusual for the line, accounts where a service fee sits alongside commission, connected-party splits. Aggregates hide the outlier. Sampling is designed to find it.

Working papers, not conclusions. Inspection teams build a record as they go. What your staff say in the room in week one becomes part of the file weighed in month four.

Scope that can widen. An inspection may open on one theme and follow what it finds. A commission examination that turns up a service agreement with no priced deliverable becomes a conduct examination, and conduct examinations reach parts of the file that commission examinations do not.

What gets pulled is predictable: the placement file and the insurer's commission schedule for each sampled policy, the firm's ledger at policy level, credit notes and reversals, any client fee agreement alongside a commissioned placement, co-broking splits with the counterparty's registration evidence, and the reconciliation between commission booked and received. A firm that cannot produce those inside a working day has already told the team something about its controls.

IRDAI has issued an Exposure Draft of the IRDAI (Manner and Procedure for Imposing Penalties) Regulations, 2026, pursuant to the Sabka Bima Sabki Raksha (Amendment of Insurance Laws) Act, 2025, with stakeholder comments invited by 9 July 2026. As at the date of this post, it is a draft. It is not notified, it is not in force, and nothing in it binds anyone yet.

What it sets out is a procedure where the framework previously left room. It contemplates:

1. Show-cause notices as the formal opening of a penalty proceeding, rather than an adverse finding travelling straight to a decision.
2. Service of notices, defined rules on how and to whom a notice is delivered, which decides when your clock starts.
3. Replies from the noticee, on the record, before determination.
4. Personal hearings, physical or virtual, giving the noticee a chance to be heard rather than only read.
5. Reasoned orders: the determination must explain itself rather than assert itself.
6. Proportional determination of the penalty, weighing the gravity of the contravention, its duration, the impact on policyholders, and any unfair gains made.

Duration is the factor that catches firms out. An arrangement that ran eleven months across forty renewals is not a single lapse, and a framework weighing duration explicitly is one where the length of time you did not notice becomes part of the finding. Unfair gains changes remediation: a determination that considers what you gained implies that giving it up is part of the answer, not a gesture.

The inspection report is the pivot: the document in which working papers become findings, and the last point at which a firm's own explanation can prevent an issue from hardening. Firms mishandle two things here.

The first is the response to the draft findings. When an observation comes back for comment, the instinct is either to argue everything or to concede everything. Arguing every observation, including the plainly correct ones, spends credibility you need on the contestable ones. Conceding everything writes the finding for the other side. What works is boring: sort the observations into ones that are factually wrong (prove it with the document, not an assertion), ones characterised more heavily than the facts support (show the limiting facts), and ones that are simply right (say so, with what you have already done about it).

The second is who answers. The principal officer's name is on the registration. Delegating the response to whoever is least busy is a decision the firm will re-read later in a very different light.

A show-cause notice is not a penalty. It states what the regulator considers you to have done, identifies the provision it considers contravened, and asks you to show cause why action should not follow. Two features matter. It fixes the case: what is in the notice is what you answer, and what is not is not, so the specificity of the allegation is worth reading three times. And it starts a clock, which is why the service provisions in the exposure draft are more consequential than they look. A reply filed after the period has run is an absence, and an absence is decided against. The reply itself is constrained by what the firm said and did at the inspection stage: there is no version of this process in which a good reply rescues a bad inspection response.

The personal hearing the draft contemplates can be physical or virtual. It is not a trial and not a negotiation, but the chance to put the firm's case to the decision-maker directly and answer what the written reply did not resolve.

The firms that use it well send the person who knows the file, not only the most senior title, because a hearing where nobody can answer a specific question about a specific placement is worse than no hearing. They lead with what is conceded, clearing the disputed ground. And they bring remediation already done rather than planned, because a firm describing what it intends is asking for time, while a firm describing what it has done is describing a fact.

A reasoned order is the determination, and the requirement that it be reasoned is a protection rather than a formality: an order that explains itself can be read, tested and challenged, and its reasoning constrains the next one.

On quantum, two anchors exist on the public record. The first is the statutory ceiling on the contravention. Section 41 of the Insurance Act, 1938 carries a fine which may extend to INR 10 lakh, following the Insurance Laws (Amendment) Act, 2015. Read alone, that invites a firm to treat rebating as a line item. That is a mistake for the reason above: a proportionality framework weighing duration does not see one contravention where an arrangement ran across a book.

The second is what actually gets ordered. IRDAI is enforcing Section 41 actively, and the visible enforcement lands where responsibility for distributor conduct sits. A penalty of INR 1 crore on Reliance General Insurance involved findings of payouts to entities and to an individual agent. That is an insurer-side order, and the pattern is the one broking firms should internalise: the exposure attaches to whoever carries responsibility for the conduct. IRDAI's warnings and penalties page is the public record, and a better guide than any commentary on it, including this one.

Paying a penalty does not close a finding. This is the most common misconception in the sequence, and it produces the second, worse proceeding eighteen months later. The penalty addresses the contravention. The finding says something about the firm's controls, and controls are addressed by remediation, which the firm proposes, executes and evidences. What a credible undertaking contains:

• A cause statement, not an apology. Which control was absent, or present and bypassed, or present and not operating. "Human error" is not a cause; it is the symptom of a missing control.
• Scoping. The sample is not the population. If the finding rests on four sampled policies, say how many in the full book carry the same pattern. A firm that does not scope its own issue will have it scoped for it, on less favourable assumptions.
• Disgorgement where there was gain. If the arrangement produced a benefit the firm was not entitled to, the undertaking should say what happens to it. Where unfair gains weigh in determination, retaining the gain is an aggravating fact.
• The control change, with an owner and a date. Not "training will be strengthened." A named approval step, a named register, a named person, a date it went live.
• Independent verification. Internal audit or an external reviewer testing the changed control after a quarter of operation, reported to the board. This is the step firms skip, and the one that converts a promise into evidence. Keep all of it in one closure file, so that when the question is asked again two years later, the answer is a document rather than a recollection.

1. Self-sample your own book on the criteria an inspection would use. Pull ten placements: the three largest by remuneration, the two where the commission percentage most departs from the firm's pattern for that line, the two hardest-won accounts of the last year, one with a client fee alongside commission, one co-broking split, and one connected-party arrangement. For each, ask whether the file answers why the economics are what they are. If not, that is a finding waiting for an author.
2. Fix the response capability, not just the records. Name who leads a response, who assembles documents, who may speak to the team, and who reviews anything the firm puts in writing. Do it before the letter, because the roles get allocated badly under pressure.
3. Test service and clock discipline. Where does a notice physically arrive, who opens it, and what happens in the first twenty-four hours, which is trivial until a notice sits in a mailbox for three weeks.
4. Run the four-factor self-assessment. For any arrangement the firm is uneasy about, write down honestly: how grave, how long has it run, did any policyholder pay for it, did the firm gain. Firms find their real exposure on the duration axis almost every time.
5. Comment on the draft, and read the published orders. The comment period closed on 9 July 2026, but the instrument is not finished with, and industry bodies remain the practical route for a mid-market firm's views on procedure to land anywhere. Read the published orders for the patterns: which arrangements attract attention, how conduct is characterised, and where the line is currently drawn.

The firms that come through an inspection intact are not the ones with the cleanest history. They are the ones whose file explains itself, whose people know their roles, and whose remediation is already running before anyone asks. Since 5 February 2026 that has stopped being good hygiene and started being the only defence there is.