The rules in someone’s head: referral discipline and key-person risk in delegated authority

Every delegated authority operation has a written referral schedule. Most also have a second, unwritten one, which is the version an experienced underwriter actually applies.

The two are rarely in conflict. The unwritten version is usually more conservative, more nuanced and better calibrated to the book than anything a document could capture. It reflects twenty years of knowing which trade lanes go wrong, which brokers submit optimistically and which commodity descriptions deserve a second look.

This is a strength right up until the moment it becomes the single largest operational risk in the business.

Two ways it fails

The first is departure. When the person carrying the unwritten rules leaves, retires or is off for a month, what remains is the written schedule, which everyone has been quietly treating as a starting point. Decisions get made to the letter of a document that was never the real control. Nobody notices until something binds that should not have.

The second is audit. A capacity provider reviewing your binder will test whether the authority was exercised as granted. The honest answer in many operations is that it was exercised more carefully than granted, which is commercially reassuring and evidentially useless, because none of the extra caution was recorded as a decision.

An underwriter who applied judgement and got it right leaves no trace. An underwriter who applied judgement and got it wrong leaves a claim. The record of good judgement is systematically absent.

What recording it actually means

Not replacing the judgement. Capturing it.

A referral rule that lives in the workflow is applied every time, by everyone, including the person covering for a colleague. When an underwriter overrides it, the override is a recorded decision with an author and a reason rather than an absence of a record.

Over time this does something more useful than satisfying an auditor. It turns tacit knowledge into something the business owns. The rules that experienced underwriters carry become inspectable, which means they can be discussed, refined and taught, rather than rediscovered by the next person the hard way.

The related question: when does screening happen

The same principle applies to sanctions and counterparty checks. A check that runs after binding is a report. A check that can prevent a bind is a control. In delegated authority the distinction matters more than elsewhere, because you are exercising someone else’s pen and they are entitled to know it was exercised within the rules they set.

Where we sit

tigerlab is one platform with two parts. The brokerage side handles placement, policy administration and settlement across four ledgers. tigersuite adds the delegated authority side: referral rules applied in the workflow with overrides recorded, sanctions and KYC screening as a gate on binding, and a shipping portal so client declarations arrive as structured data rather than as attachments.

Firms that only place need the first. Firms holding binding authority for cargo need both, and they were built to sit together rather than to be integrated afterwards.