Who are the parties in an insurance distribution chain, and how would you calculate and settle their commission?

The parties, and whose side they are on

The single distinction that organises everything else is agency. A broker acts on behalf of the customer: they source cover from a panel of insurers, they owe the customer a duty, and the insurer is their supplier. An agent acts on behalf of the insurer, whether tied to one insurer or representing several. The labels are used loosely in conversation and their legal meaning varies by jurisdiction, but the direction of the duty is what drives requirements: it determines who is responsible for suitability advice, who holds the customer data, who the customer complains to, and who the customer's premium legally belongs to while it is in transit.

A managing general agent, called a coverholder in the Lloyd's market, goes further. It holds delegated authority to underwrite and bind on the insurer's behalf within an agreed appetite, and often to handle claims. The insurer is on risk for policies it did not price and did not see at the point of sale, which is why delegated authority comes with audit rights, bordereau reporting and appetite constraints expressed in the binding agreement.

An aggregator, or price comparison website, is not a distributor of cover so much as a generator of quote requests. The customer enters their details once and the aggregator fans them out to a panel; you return a price and, if chosen, the customer arrives on your journey to complete the purchase. Payment is usually per acquisition rather than as commission on premium.

Embedded insurance is cover sold inside another transaction: travel cover at checkout, device protection with a phone, cancellation cover with a ticket. Structurally there is almost always an MGA or an insurer behind the brand the customer sees, integrating over an API, and the distinguishing feature is that the customer's relationship is with the retailer and not with you.

What each channel does to the technology

Broker and agent channels need a quoting interface built for a professional user rather than a consumer, and a servicing model in which the intermediary rather than the customer initiates mid-term changes, which changes your authorisation model.

Aggregators impose two hard constraints. You are quoted on every request whether or not you win, so quote volume dwarfs bind volume and every per-quote cost, particularly paid external data enrichment, multiplies accordingly. And you must be able to reproduce a price you gave against data you did not collect yourself, then re-price cleanly when the customer amends their details after arriving on your own journey.

Embedded channels are API-first and high-volume with a very short decision window inside someone else's checkout. The interesting engineering problem is usually not the quote but the aftercare: the customer often does not remember who underwrote the cover, so claims notification, servicing and renewal all have to work for someone who did not knowingly buy a policy from you.

Commission as a consequence of premium

The mistake to avoid is treating commission as an attribute of a policy. It is not a number; it is a stream of transactions, each of which is caused by a premium transaction. Model it that way and the awkward cases resolve themselves.

Every premium transaction, new business, endorsement additional premium, endorsement return premium, cancellation return premium, instalment adjustment, produces a corresponding commission transaction at the rate applicable to that transaction, with the same sign. A return premium therefore produces negative commission automatically, which is what clawback is. No special case required.

The base is the commissionable premium, which is not the amount the customer pays. Insurance premium tax and statutory levies are collected on top of premium and are not commissionable; nor are fees the intermediary charges the customer directly, which are the intermediary's own income and never enter your ledger. Getting the base wrong is the single most common cause of a broker statement dispute.

Transaction	Premium	Commission at 15%
New business, 12-month term	+600.00	+90.00
Endorsement, added cover, 6 months remaining	+80.00	+12.00
Cancellation at month 8, pro-rata return	−226.67	−34.00
Net position	453.33	68.00

The rate itself is not one number either. It varies by product, by channel, by intermediary, and by transaction type, and it is effective-dated, because renegotiating a broker's terms must not restate commission already accrued on business written under the old terms. So the rate lookup carries the same versioning discipline as a rate table.

Beyond flat commission there is contingent income: overrides, profit commission and sliding-scale arrangements that pay more when the account's loss ratio comes in better than a target. These cannot be calculated at the point of sale because they depend on how the year develops, so they are accrued as an estimate and trued up when the account is reviewed. Keep them in a separate accrual, clearly distinguished from transactional commission, or you will never be able to explain either.

Settlement, where cash and ledger diverge

Two arrangements are common and they behave very differently.

Under gross settlement, the intermediary remits the full premium and you pay commission back separately. Cash flows are clean and the reconciliation is simple, and it is the usual arrangement for agents.

Under net settlement, which is standard for brokers, the broker deducts their commission and remits the balance. Nothing arrives that you can match against a gross premium figure, so reconciliation is against a statement of account, and the account is a running balance across many policies and many transactions, in both directions, over a period. Cash arrives as one payment covering hundreds of movements. Allocating that payment to the right transactions is the actual work, and the residual differences that never quite clear are what credit control spends its time on.

Two more things belong in an answer at senior level. Premium in the hands of an intermediary is subject to client-money rules in most regulated markets, with requirements about segregated or trust accounts and about whose money it is if the intermediary fails; the detail differs enough by jurisdiction that you should say which regime you have worked under rather than generalise. And commission is earned over the policy term rather than at the point of sale, so an insurer's accounts distinguish written from earned, and the unearned portion of commission is a balance-sheet item, not something you can quietly ignore because the cash has already moved.

The way to sanity-check any commission design is to run a cancellation through it. If a mid-term cancellation with a pro-rata return premium does not produce a negative commission entry, correctly dated, at the rate that applied when the premium was written, the model is wrong regardless of how well it handles new business.