Coverholder oversight in the 10 to 20 binder band: where the operating model is breaking, and what fixes it

Managing agents with 10 to 20 coverholders sit in the most operationally complex part of the Lloyd's market. Below 10, oversight runs as a manual exercise. Above 20, the case for systematised oversight has usually already been made. The middle band is where the manual approach is breaking but the procurement case has not yet landed. It is also the band where the audit cycle is starting to bite.

Two regulator moves frame this. Lloyd's Market Oversight Plan 2026, published 13 January 2026, raises the bar on continuous oversight and treats the DA Oversight Manager conversation as a quarterly artefact, sometimes a monthly one. The LMA Coverholder Audit Scope and Associated Guidance refresh, 22 October 2025, widens the audit beyond attestation review into operational testing of the controls underneath. Lloyd's Principles-Based Oversight (2024-25) and the bordereaux v5.2 standard sit alongside both as the supervisory grammar.

This piece walks three areas where the current operating model breaks under that new bar:

1. Evidence assembly is failing the continuous-monitoring expectation.

2. Attestation and RAG colour are standing in for the control.

3. Per-binder workflows fragment the coverholder view.

Area 1: evidence assembly is failing the new expectation

The annual review file in most managing agents is a composite. It draws from the coverholder attestation, the broker confirmation, the third-party audit, the bordereau reconciliation, the complaints log, the claims pattern, and the Head of Delegated Authority's working notes. Each piece lives in a different place. The bordereau team chases late submissions and corrects inaccuracies in the bordereau system. The signal stays inside that system.

The MOP 2026 frames bordereaux as supervisory evidence and expects the DA Oversight Manager to read pattern, not just position. The LMA refresh extends the audit window beyond the annual file into the working evidence. Bordereaux v5.2 sits underneath as the data contract. All building a continuous monitoring expectation rather than the current snapshot approach.

Common shortcomings of the current approach include the 40 percent of the annual review window spent assembling, the mid-year repeat of that assembly when the DA Oversight Manager asks for the operating picture, and the bordereau read as administrative chase rather than control signal. A coverholder that consistently misses bordereau deadlines is signalling something about its operations. The signal is the leading indicator the RAG eventually catches up with, usually four to six months late.

Area 2: attestation and RAG colour are standing in for the control

Three to five attestation-led controls per coverholder typically carry the audit weight. Coverholders attest the controls operated. The MA files the attestation. The attestation becomes the evidence at audit. In parallel, most MAs run a coverholder dashboard with RAG ratings. A coverholder is green at the start of the year, amber at month seven, red at month nine. The Head of DA escalates at amber and intervenes at red.

The LMA refresh of 22 October 2025 directly targets the attestation-as-evidence move. Audit now tests operation, not the document that says operation happened. The FCA's supervisory letters across 2024 and 2025 sit behind that position. The direction of travel is judgement and standing evidence, not a longer checklist.

Where this trips most functions up is the substitution itself. An attestation is evidence someone said the operation happened. A RAG is the receipt of a control failure already in flight. By the time a coverholder is amber, the underlying issue has been live for two reporting cycles. The Head of DA's intervention at amber is intervention at month seven for an issue that started in month three. Neither the attestation nor the colour is the control. Both are summaries that have been promoted to operating signal.

Area 3: per-binder workflows fragment the coverholder view

Most managing agents structure oversight per binder. A coverholder with three binders gets three oversight workflows. Binder A, binder B, binder C are reviewed as separate items. The coverholder, as a single regulated party with one set of underlying controls, gets fragmented across three views.

The MOP 2026 reads oversight at the coverholder level. The DA Oversight Manager conversation is about the coverholder, not the binder. Principles-Based Oversight points the same way. Oversight stops being an event the function prepares for, and starts being an operating layer the function runs from.

The traps in the current shift are subtle. A key-person dependency at the coverholder, for example one underwriter holding three lines of business together, presents as three separate issues in the MA's workflow. The Head of DA solves each in isolation. The single underlying cause is not identified until the third issue prompts the question. The binder is a contractual artefact. The coverholder is the unit of oversight. The current workflow has the relationship inverted.

The solution: continuous oversight as an operating layer

Two mechanisms close the gap.

• Pitfalls in Area 1 and Area 2 (evidence scattered, attestation and RAG standing in for the control) are fixed by the Controls-First Approach. Controls are owned by the operator, decomposed into RCIA so the indicators sit on the controls themselves, and updated in the rhythm of the work. The bordereau, the complaint, the audit finding land on the relevant control as they are produced. The annual review file becomes a formal cut of standing evidence, not a from-scratch build.

• Pitfalls in Area 3 (binder-level fragmentation, key-person dependencies invisible across binders) are fixed by the Parent-Child module. The managing agent sees coverholders as direct children and binders as grandchildren. Controls report on the coverholder. Indicators aggregate upward to the MA view and downward to the binder.

The annual review file still exists. The RAG dashboard still exists. They just stop being the moment the operating picture gets built.

Next step

Pick one coverholder. Identify the three attestation-led controls most relied on at the last audit. Convert one of them to indicator-led evidence drawn from bordereaux v5.2 data, complaints volume, and claims ratio. Map the controls to the coverholder rather than the binder. The exercise takes one week of analyst time and produces a working example the rest of the book can pattern-match against.

If you want to see how this works in a delegated authority setting, the wedge page is at /who-its-for/lloyds-coverholder-oversight.