Delegation Without Control is Hidden Exposure
Measuring, pricing and managing risk is at the heart of every insurance business. And yet some critical parts of the insurance ecosystem have not received the attention they deserve.
Delegated business now accounts for close to 40% of all business written at Lloyd's1, and it is growing. Yet the data and reporting infrastructure beneath it runs weeks behind the rest of a typical insurer's book. The SMF24 holder is accountable for an exposure that the managing agent would not tolerate managing on this lag for any other line. That asymmetry is no longer a resourcing issue. It is a material governance gap.
This is not new ground. As long ago as 2019, the Mazars Delegated Authority Management Survey found that 53% of Lloyd's managing agents reported struggling with resources in their delegated authority teams, a proportion that had risen since 20172. In the years since, the volume of delegated business has continued to grow. The resource problem has not gone away.
The asymmetry exists because the data infrastructure under delegated business has, until recently, been too slow and too expensive to support a near real-time operating model. The managing agent built its measurement discipline around those parts of the book where data was available in timely manner. Delegated business was not one of them. AI-assisted matching has changed that calculation. The bordereaux already in flight can now be reconciled, translated and surfaced in days rather than weeks, without replacing the systems the managing agent already runs. The historical reason for the asymmetry is no longer the reason it persists.
The barriers that remain are organisational and commercial, not technical. Coverholders are reluctant to invest in new systems unless capacity providers demand it. Brokers have built intricate processes in Microsoft Excel which have done the job for years. Managing agents need better data but are often unwilling to fund the coverholder's upgrade. With budget allocations consumed by regulatory compliance and business development, the bordereaux pain has been tolerated. The result is a market where pockets of automation coexist with large volumes of manual processing, and where coordination across parties with different systems, budgets and incentives is the real constraint. Three things have now shifted that balance: the market, regulatory scrutiny, and AI adoption.
What changed in March
The market's fastest-growing and most data-intensive segment had been quietly waiting for a centralised solution that was not coming in the form or timeframe expected.
On 19 March 2026, Lloyd's confirmed in its full-year results that its council had approved the decision to sunset the name and original vision of Blueprint Two. The chief executive's statement was direct: "despite the efforts of many skilled and committed people, the project has not yielded the benefits that were originally envisioned"3. Lloyd's remains committed to a phased re-platforming of the market's operational infrastructure, but the comprehensive market-wide solution that was meant to lift the whole ecosystem at once is no longer the plan.
For delegated business this development matters more than for any other part of the market. The Blueprint Two provisions for delegated authority were, in practice, always its weakest component4. The market's fastest-growing and most data-intensive segment had been quietly waiting for a centralised solution that was not coming in the form or timeframe expected. Insurers and managing agents that postponed their own investment in delegated data infrastructure did so based upon a logic that was rational at the time but is no longer.
The infrastructure is already there
Lloyd's own infrastructure supports the move. The Delegated Data Manager, launched in 2018, was always intended as the centralised repository for delegated reporting5. The Delegated Contract and Oversight Manager holds the binder data. The Coverholder Reporting Standard defines the field set. The pieces of the data architecture are in place. What was missing was a way to translate the messy reality of monthly bordereaux submissions into the clean form those systems expect, at speed, without committing to a multi-year platform build. The current generation of AI-assisted matching closes that gap. The matching layer sits over the bordereaux already in flight and over the systems the managing agent already runs. No new platform; no data standard imposed on coverholders from above; no two-year programme. Production deployments of these new AI-enabled systems are set to increase rapidly.
The regulator has not been silent
Lloyd's has named the gap publicly for several years. The 2025 Market Oversight Plan stated that the Corporation's focus would include "delegated claims data (timeliness and accuracy) and how that data is utilised to drive performance"6. The Chief Underwriting Officer's Q2 2025 market message was blunter: a laser focus on coverholder oversight, with restrictions on the ability to delegate threatened where coverholders are not managed well1. From the start of this year, each managing agent has a dedicated Delegated Authorities Oversight Manager assigned to it, with structured engagement and lifecycle reviews designed to build a complete view of the agent's delegated capability.
The Senior Managers and Certification Regime makes this personal. The Chief Operations function (SMF24) carries named accountability for the operational arrangements of a regulated firm. At a managing agent, that includes the oversight of delegated business. The Oversight Manager is the regulator's response to a gap that it has been describing in plain English for at least two years. The expectation is no longer being announced. It is being acted upon.
What we see going wrong
We see the same issues recur where delegated authority oversight is treated as a compliance exercise rather than an operating model question. These observations are consistent with the patterns identified in the Mazars survey2, the TIN Delegated Authority Strategy Day pre-event survey7 and industry analysis of bordereaux processing cycles8.
The bordereaux backlog as an accepted norm. The team spends the first two weeks of every month processing last month's bordereaux8. By the time the data is in the aggregation view, the next month's submissions have begun arriving. The lag is built into each managing agent's operating model and constitutes a risk in and of itself.
Binder standards that exist on paper but not as rigorously in practice. The binding authority agreement specifies data quality and timeliness requirements. Nobody has audited whether any coverholder actually meets them1. The binder is a legal document, not an operating document.
Oversight that depends on individuals, not systems. A very small number of experienced analysts know which coverholders are problematic. When analysts are on vacation, or leave the business, the knowledge goes with them2. The operating model is transient rather than embedded in the institution.
Pilot paralysis. The managing agent ran a proof of concept eighteen months ago. The results were promising. Nobody signed off the next step. The pilot sits on a shared drive, the vendor has moved on, and the bordereaux backlog continues.
What good looks like, in the order that works
The managing agents that are closing the gap are not running parallel transformation programmes. They are doing four things in sequence, and they are doing the first three before they commit to infrastructure spend.
First, they audit their binding authority portfolio against the operating model their own oversight regime actually requires. The standards in the binder, the cadence of review, the thresholds that should trigger escalation. Where these are vague or inconsistent across coverholders, they are tightening them now.
Second, they map the data flows from each coverholder, identifying the small number of fields that actually drive the managing agent's portfolio and aggregation decisions. The Coverholder Reporting Standard field set is the technical starting point; what matters is which of those fields carry decision weight for this agent's particular book.
Third, they run an AI-assisted matching layer over the existing bordereaux flow as a fixed-scope pilot, typically on the top portion of the coverholder population by premium. The output is a rapid view of the existing portfolio rather than a new platform, with exception cases surfaced for the team that previously spent the bulk of each month on manual reconciliation.
Fourth, and only fourth, they decide what to invest in for the long term. By that point the operating model is clear, the data flows are understood, the AI tooling has been validated in production on the managing agent's actual book, and the infrastructure question is sized against demonstrated need rather than vendor proposition.
The pattern is familiar
The London Market has been through cycles like this before. A regulator describes a gap. The market commits to a centralised solution. The centralised solution arrives late, partially delivered, or, in this case, redirected. The managing agents that postponed their own investment in anticipation find themselves still holding the original problem, now with less time and a sharper regulatory posture.
For the Head of Delegated Authority, the Chief Underwriting Officer, and the SMF24 holder. None of these requires a project to answer. All should be answerable from the data the managing agent already holds.
The lag. What is the elapsed time between a coverholder writing a policy and the managing agent having that risk represented in its aggregation view? This should be answered in days, not as a process description.
The variety. How many distinct bordereau layouts does the managing agent currently receive each month? How many staff days are spent reconciling them?
The binder. How many of the managing agent's coverholder relationships have a binding authority agreement that specifies the data quality and timeliness standards the agent requires for its own oversight, and how many do not?
The visit. If the managing agent's Delegated Authorities Oversight Manager asked today for evidence that the agent has reviewed performance against each binder in the last quarter, on what data could the agent answer?
The drift. If a coverholder's book deteriorated by 15% in a single quarter, how long would it take the managing agent to know, and at what point would the Active Underwriter be notified by the operating model rather than by an individual analyst?
The asymmetry. Compared with how the managing agent measures and manages every other exposure on its books, where does the measurement of its delegated book genuinely meet the same standard, and where does it not?
What is different in 2026 is not the pattern but the toolset. The data infrastructure question for delegated business no longer needs to be answered with a multi-year platform programme. It can be answered with the infrastructure the market already has, plus a matching layer that closes the asymmetry between how delegated business is measured and how the rest of the managing agent's book is measured.
Delegation without control is hidden exposure. The reason that exposure has remained hidden is that the data has, historically, arrived too late to do anything about it. That reason no longer holds. The managing agents that act on this in 2026 will close a gap that the rest of the business has been treating as someone else's problem. The ones that do not will find the Oversight Manager closing it for them.
For principals on both the underwriting and claims sides of delegated business, the most useful first step is a focused, time-boxed diagnostic of the firm's own delegated book, typically scoped to the top portion of the coverholder population by premium volume. This is an exercise a well-run managing agent can largely structure and run for itself, and we would encourage principals to do exactly that.
A good diagnostic audits the binding authority oversight model against what the firm's own regime actually requires, confirms where the real pain points sit, and clarifies their root causes. Where it points toward remediation, the tooling to act now exists: AI-enhanced platforms (such as Hercules.ai and others) can sit over the existing bordereaux flow without a multi-year platform build. The value of the diagnostic is that it establishes what, if anything, is worth doing, in what order, and at what scale, before any commitment is made.
A thorough diagnostic applies to both the premium and the claims side of delegated business. On the underwriting side, it assesses the firm's maturity against the capabilities a modern delegated authority operating model requires. The same logic applies to delegated claims: the quality and timeliness of claims bordereaux; the oversight of TPAs and delegated claims administrators; and the firm's visibility of leakage, reserving accuracy and settlement authority against the terms of the binder.
None of this requires external help to begin, and much of it is best owned internally. But if the diagnostic surfaces more than your internal team has the capacity to address, or if it would simply be useful to talk it through with someone who has looked closely at these problems, we would be glad to have that conversation. Contact Richard Spencer at new-business@new-linkconsulting.com.
- 1.Rachel Turk, Chief Underwriting Officer, Lloyd's of London. Q2 2025 market message, reported in Insurance Journal, "Lloyd's Has 'Laser Focus' on Oversight of Coverholders to Proactively Avoid Market Risk," 16 May 2025.
- 2.Mazars (now Forvis Mazars), Delegated Authority Management Survey 2019, published 15 July 2019; based on responses from 30 managing agents representing approximately 55% of the Lloyd's market.
- 3.Patrick Tiernan, Chief Executive of Lloyd's of London. Statement on the conclusion of Blueprint Two, Lloyd's full-year 2025 financial results report, 19 March 2026.
- 4.Industry analysis of Blueprint Two's delegated authority provisions following its conclusion; see Hercules.ai, "The Blueprint Is Dead. Long Live the Firm," 23 March 2026.
- 5.Lloyd's of London, Delegated Data Manager, lloyds.com; Delegated Data Manager went live in 2018 as the central bordereaux processing platform.
- 6.Lloyd's of London, 2025 Market Oversight Plan; section on delegated business and claims data focus.
- 7.The Insurance Network, Delegated Authority Strategy Day 2025 Pre-event Survey Results, published 25 April 2025; based on responses from DA ecosystem participants including syndicates, MGAs and coverholders.
- 8.Vitesse, "Insurance Claims Reconciliation: A Guide to Automation for Carriers," 30 March 2026. States: "For carriers with multiple delegated authority partners, manual reconciliation cycles routinely take two to four weeks per month."