Your MGA has moved on. Its policies haven’t.

A transition can be complete for the sales team and unfinished for the policyholder. New submissions reach the right desk. The replacement workflow is live. Management has announced the next phase. Yet a policy written under the previous arrangement can still generate a claim, an endorsement request or a question about its original terms. The current organization chart may not answer who should handle it.

This is not simply a technology migration problem. An MGA can change an administrator without moving every policy, reorganize an underwriting office while retaining its servicing entity, or stop accepting business while continuing to meet existing obligations. These events should not be collapsed into one generic exit narrative. Their different boundaries are precisely what management needs to preserve.

MGA Index’s thesis is that leaders need two operating views: one for the business they want to write next, and another for the policies they remain responsible for servicing. Three public examples illustrate the distinction. They do not establish service failures at the organizations concerned; they show why a single current-state description can be insufficient.

The relevant date may belong to the policy, not the claim

Victor’s Dealer Open Lot FAQ divides claims reporting by policy binding date. For policies bound before April 1, 2026, it directs all claims to Broadspire. For policies bound on or after that date, it directs all claims, regardless of severity or location, to Boulder Claims. The instruction concerns this program, not every Victor product.

The important distinction is the binding date, not when a loss occurs or a claim is reported. An operating system that simply displays the newest administrator could therefore provide the wrong route for an earlier policy. A current contact directory and a policy-specific servicing instruction are different information products.

The management lesson is broader than a contact change. Each transition needs a controlling trigger that can be applied to the individual policy. That trigger might differ across arrangements; it should never be inferred from a convenient system date. Brokers, claims teams and service staff need access to the same rule and a clear escalation path when the underlying record is ambiguous.

An office’s future does not describe every existing obligation

Mosaic’s March 24, 2026 announcement described a restructuring of its Singapore and Dubai underwriting platforms. It said its Singapore service company would remain to administer existing policies while immediately ceasing to accept direct new transactional liability submissions. Mosaic described work with Lloyd’s Asia toward an orderly office closure, with business facilitated through other offices subject to territorial regulatory requirements.

That announcement distinguishes an existing-policy function from a new-business route. It is not evidence that every policy transferred elsewhere or that the announced closure had already been completed. Nor is an office reorganization equivalent to the termination of a policy.

For a leadership team, the implication is organizational: removing a location from the growth plan does not by itself remove the work associated with the book written there. Responsibility needs to survive changes in reporting lines. Otherwise, a task that has become less visible commercially can become harder to allocate operationally.

Stopping underwriting is not the end of operating the book

Volante’s notice about Syndicate 1699 provides a third boundary. It said the syndicate would cease writing new and renewal business at 23:59 GMT on January 31, 2026 and enter solvent run-off. The company said in-force policies remained valid and claims would continue to be handled, with operational capacity and resources maintained for existing obligations.

Those are company statements about the syndicate’s run-off arrangements. The notice does not establish a replacement capacity arrangement or a transfer of its policies. It should not be presented as evidence about the performance of a different Volante business.

The strategic implication is that a production stop and an operating stop are separate decisions. Once new-business activity ends, management still needs to know which services remain, who performs them and what resources support them. A business can be closed to new underwriting while its obligations remain active.

Build a continuity register around policy cohorts

A useful continuity register is not another copy of the sales pipeline. It describes groups of policies that share the same servicing arrangements, while retaining enough detail to resolve exceptions at policy level. The purpose is to make the historical context usable by someone who did not participate in the transition.

MGA Index recommends recording five elements for each material cohort. These are operating questions, not a universal legal checklist; controlling contracts and applicable requirements determine the actual allocation of responsibility.

• The boundary: the product, carrier, policy period and exact event or date that determines which arrangement applies. Preserve the original policy and relevant version history.
• The route: the responsible service team, claims administrator and escalation contact for that cohort, rather than only the contacts used for new submissions.
• The authority: which party can make the requested decision, what requires referral and how an exception is documented.
• The evidence: where the policy record, endorsements, claim correspondence and relevant financial records can be retrieved by authorized staff after systems or offices change.
• The owner and resources: a named accountable person, an agreed service arrangement and a way to fund and review the work that remains.

Avoid false simplicity. Two cohorts may use the same administrator but have different decision rights. Two policies issued through the same brand may not share the same carrier or service arrangement. Conversely, a single administrator can deliberately serve several historical cohorts without requiring a separate platform for each.

The register should identify uncertainty rather than hide it behind a populated field. If the team cannot establish which record controls, the correct response is a named escalation and resolution process, not an assumption that today’s workflow applies. An unresolved exception is manageable when someone owns it; an invisible one is not.

Test continuity before declaring the transition complete

A new-business acceptance test asks whether a suitable submission can be quoted and bound. An existing-policy acceptance test asks whether the organization can still act correctly on business written before the change. Passing the first does not demonstrate the second.

Select representative policies from each material cohort using authorized access and appropriate data controls. Can the service team retrieve the original contract and its amendments? Can it explain which routing rule applies? Can the recipient accept the handoff and identify who has authority to act? Can relevant financial records be reconciled without depending on a departing employee’s memory?

Test the awkward cases as well as the straightforward ones: a later-reported matter on an earlier policy, an endorsement spanning a system change, or a query received through an old broker contact. These are proposed test scenarios, not reported incidents at the companies cited. The point is to test historical context under ordinary operational pressure.

Then ask what happens after the transition team disbands. A procedure that depends on one temporary project manager is not a durable service arrangement. Record ownership, access and escalation should be maintained through ordinary operations, with unresolved exceptions reviewed until they are closed.

The counterpoint: moving everything can make continuity worse

There is no general rule that every old policy should migrate to the newest platform or administrator. An established service team may understand the original wording, records and claims context better than its replacement. A forced move can introduce handoff risk without producing enough benefit. Parallel arrangements may be the considered operating design, not evidence of unfinished integration.

Nor should a smaller MGA build an elaborate governance structure where a concise register and a tested service agreement would suffice. Proportionality matters. The objective is not one platform, an impressive transition dashboard or the fastest possible retirement of a legacy system. It is reliable service under the arrangements that actually govern each policy.

That still requires an economic decision. Management should identify the work and resources that persist when the new-business model changes, rather than treating them as incidental costs that will somehow disappear. A growth plan is more credible when it accounts for both the next book and the obligations inherited from the last one.

The decisive question is simple: if an earlier policy needs attention tomorrow, can the organization explain what happens next, demonstrate why that route is correct and show who is accountable? Until it can, the announcement may be finished. The transition is not.