Shared Evidence: A Continuous Interpretive Layer for Lloyd’s Principles Based Oversight

15 July 2026

— Executive Summary

Lloyd’s continuous performance oversight (CPO) and resulting view of London Market businesses are informed by those same businesses’ annual self assessment of their principle alignment. This article describes two views of business activities: the Lloyd’s view, maintained through CPO, and the managing agent’s, expressed in its annual submission. Each is authored separately and delivered in different cadences. If these are in eventual dispute, they resolve to the Lloyd’s view.

The article asks whether misalignment between Lloyd’s and managing agents’ views contributes to the “shadow costs” borne by market businesses. It asserts that there are essential and accidental causes of misalignment, and that an evolution of the framework should address the accidental causes.

The proposition is to add a shared interpretive layer to the PBO framework. The layer’s actions and events would result in a standing record supporting oversight of business outcomes. The layer would:

  1. aim to reduce the level of disagreement at end-of-year reconciliation
  2. support the agent’s annual narrative attestation, which may now draw on the year’s standing record of submissions to the interpretive layer
  3. support pro-active behaviours by managing agents throughout the year

Managing agents would submit interpretations, mapped to the principle definitions, of directly sourced evidence of business outcomes. The regulator would optionally assess submissions based on their internal consistency and sufficiency, responding with acceptance/rejection of the compilation into the interpretive layer. Responses would not be mandatory for each submission, decoupling the resourcing of the regulator and businesses and allowing for flexibility in resource allocation, risk management and materiality judgements. Adjudication of categories and other decisions remain as they are in current framework structures but enhanced by an outcome focused shared interpretive layer.

To help visualise the proposition, two interactive simulations are provided. They model views of the layer from the perspective of the managing agent. The ‘10000 ft’ view shows the layer’s development over 12 months of contributions to its submission API. As a primary technical surface for the layer, the submission API is an example of a ‘flexible open architecture, and enablement of individual strategic choice’.

The second simulation demonstrates that choice enacted — strategic building on an open system. It presents a close up of the work involved in generating a compilation of evidence, submitting it and receiving a response or, as is the default in this zoomed in simulation, not receiving one.

The article leaves open questions as an invitation to shape the conversation and examine the foundations of the proposal. The ‘10000 ft’ simulation also invites comment to help its development towards a quantitative model supporting the case for future investment.

— Introduction

In its dual role as promoter and regulator, Lloyd’s has published extensive material on its Principles Based Oversight (PBO) processes and contributed, along with London Market participants, to the 2022 Inquiry into Commercial insurance and reinsurance regulation. Its written submission to the Inquiry indicated a direction of travel for the nascent PBO framework: systematised interaction between Lloyd’s and Market Participants would serve as an example of streamlined engagement. The subsequent success of that direction enabled the PRA to increasingly rely on PBO for their own oversight of the London market.
In 2026, PBO is described as an embedded yet evolving framework.

“Whilst now is not the time to be adding complexity or making fundamental changes to PBO it must continue to evolve.”

“Our Principles-Based Oversight Regime (PBO) has been in operation for three full years, and it is now well embedded.”

This article looks to contribute to that evolution of the interaction between market participants and the regulator. It recommends a new interpretive layer, supporting conversations about shared evidence of business outcomes.

1 Principles Based Oversight

PBO is comprised of activities and interactions between managing agents and the Lloyd’s Corporation Market Oversight function. PBO mandates that managing agents assess their alignment with the Principles from their first-hand knowledge of business outcomes. PBO activities include BAU meetings, timetabled and ad-hoc submissions, audits and the set of end-of-year self assessment narratives.

The Corporation runs Continuous Performance Oversight (CPO), a year-round programme in which technical analysts review submissions from managing agent counterparts. Analysis and data is exchanged across actuarial, claims, underwriting performance and oversight functions, and perhaps others.

“Lloyd’s operates a Continuous Performance Oversight approach. Core oversight for all syndicates will include Lloyd’s challenging syndicates in light of market conditions, plans and actual writings throughout the year. Targeted oversight will be enacted to look at syndicates and classes where this is expected to be particularly relevant.”

The annual cycle of framework interactions — including points of contact such as Account Managers, regulator internal committees like the Market Oversight Group (MOG), or the Syndicate Business Discussions (SBD) and Capital Planning activities — has a tempo and calendar running parallel to the customer-facing business that makes up the market.

A June 2026 description of the Market Governance and Risk Manager role, for Principle 10 only, may add further detail to our understanding of CPO.

Through building trusted partnerships and having an open dialogue with your portfolio of managing agents you will ensure business is conducted in line with the defined Governance and Risk Oversight Objectives and Principles. You will have oversight of :

  • Board/ Risk/ Audit Committee Packs
  • Risk Management Framework
  • Outputs from external party reviews (e.g. Board effectiveness Reviews) ORSAs, Internal
  • Audits, PRA PSM Letters etc; and
  • Access to the C-Suite, Senior Management and iNED’s.

From this will make recommendations to the Senior Manager on whether additional oversight activity is required and to ensure any relevant concerns arising from the document review and meetings/interview

Original source backed up for reference

2 Two patterns of change and one substrate

PBO has a foundational document, The Principles for doing business at Lloyd’s, which categorises outcomes of business activities that are aligned with the declared principles of the London Market. Principles, Sub-principles, Maturity Matrix and guidance therein are intended to change minimally year on year. This control of change can be contrasted with an annually released ‘Market Oversight Plan’ which ‘sets out Lloyd’s priorities and planned areas of focus for the coming year’.

Together these form two patterns of change and one trusted partnership layer: annual versioned release of the Principles with documented differences between versions aiming for predictability, an annual release of a new Market Oversight Plan, and a CPO substrate running year-round across both.

3 Evidentiary work performed and exchanged

In the PBO framework the regulator works to understand market participants. Managing agents must work to understand themselves and then supply reports and data to the regulator as part of CPO. The outcome focus of PBO requires the agent to compile evidence of their self assessed material level of principle alignment. The inputs here are business decisions and behaviours, mapped with the aid of guidance or other interpretation.

Between engagements and submissions, both sides maintain separate evidentiary work. For analytical teams, engagements are dense and continuous, contrasting with the annual cadence of board level narrative submissions. We can see the likelihood of having two views of the businesses, especially with the long-cadence processes. Where two views exist there will eventually be some form of reconciliation.

The existence of a Lloyd’s view is mentioned in a role description

The principal accountabilities of this role include:

…

Responsible for maintaining the ‘Lloyd’s view’ on the Delegated Authority capability of allocated MAs, feeding into governance forums.

and the managing agent view being described as a feature of the Oversight Framework

The Oversight Framework:

Allows managing agent senior management to interpret and apply Lloyd’s expectations in the way most appropriate to their business;   

Across both sides, those interpreting the principles form an ‘interpretive community’. Such communities are subject to what Julia Black called the Interpretive Paradox. Here divergent readings of principles may occur given the open nature of the principles themselves.

“Principles … are formally characterized by general, imprecise terms. This is meant to give flexibility. However, in practice principles can receive very specific interpretations. There can be benefits: certainty is produced through the development of an interpretive community which gives particular content and meaning to the principles. However, interpretive communities can fracture, and the regulatory regime may contain several interpretive communities, each with a different interpretation.”

CPO, with its open dialogue with managing agents and trusted access to decision makers is a form of ‘interpretive community’. In CPO, interpretations are authored and reported regulator-side under that trust relationship.

We may think of misalignment in interpretations in PBO as falling into one of two types: Essential and Accidental, a concept adapted from Fred Brooks. Disagreement and its resolution are an essential aspect of the framework given principles are open to interpretation. Accidental disagreement may arise due to different work cadences and exposure to differing sources. Addressing the causes of accidental disagreement makes those issues dissolve.

4 Expressions of business needs

A. This oral evidence from 2022 reveals a business need: driving the interaction.

“One of the important elements for us is that some of that interaction is driven by us because we want to make sure that we get ahead of any requests we receive and that we are doing all the right things across the quite broad pantheon of different businesses we have in the UK.”

— Richard Dudley, Aon, HoL Industry and Regulators Committee, oral evidence

B. Here, in 2022, Lloyd’s described the opportunity to shape the governance policies they were subject to through early and active co-development of policy.

“There is a real opportunity to engage more actively and earlier in the process so you are not simply sitting there receiving what the regulator is going to do as an information-sharing exercise, but you are able to really help them develop the policies.”

— David Sansom, Chief Risk Officer, Lloyd’s, HoL Industry and Regulators Committee, oral evidence

5 PBO Framework responsiveness to business needs

A. New Entrant documentation describes pathways that waive the full Principles review and shorten timelines from 6–8 months to 3. These have the effect of reducing the resourcing burden on new entrants to the London Market.

“the Captive Syndicate will not be required to complete a full Principles review with the Market Development New Entrants team.”

“SIAB will also not be required to complete a full Principles review with their syndicate application.”

Developing a new Lloyd’s underwriting business, June 2022

B. Outlook Signalling is a new framework feature that tells businesses where Lloyd’s opinion on them is heading. It addresses some of the temporal issues involved in transferring data, narrative and understanding between parties.

“… clear and consistent signposting of potential changes to a syndicate category, which we will describe as being on either a positive or negative “outlook”, alongside more transparency around the key milestones and deliverables expected to be achieved to address areas which do not “meet expectations” under PBO”

C. The Chief of Market Oversight has made a commitment to giving thirteen months’ notice, absent exceptional circumstances, for changes to individual Principles, their sub-principles, or the reporting requirements about them. This commitment creates a stable core for PBO.

6 What this reading leaves open

A. Are the needs of Brokers and Lloyd’s, as documented in 2022, mirrored today in the relationship between managing agents and oversight functions in the PBO framework?

B. The 2022 written submission to the Inquiry suggested BAU costs were driven by supervisory processes. The Lloyd’s CEO has spoken of the ‘shadow costs’ borne by the Market. Do anticipation costs and cadence differences between parties in the framework contribute to these? Is PBO under-invested because ‘shadow costs’ are hidden?

C. Would managing agents choose to share evidence of outcomes for a more effective oversight experience?

D. Would minimising accidental causes of disagreement better support Lloyd’s own oversight objectives by reducing disagreement resolution by authority?

“There may be cases where Lloyd’s and the managing agent do not agree on a Principle or sub-principle rating. Both views will be available to Lloyd’s teams. It is always the Lloyd’s view which is used in the calculation of the syndicate category and in any other oversight decisions.”

— 2025 Principles Board Attestation Guidance, p.7

Lloyd’s places primary responsibility for oversight on managing agents’ boards and management

Lloyd’s oversight is holistic and joined up

7 Where this reading leads

If the previous section’s questions were resolved along the lines suggested by the 2022 evidence, how might the PBO framework evolve in response? How would its development align with the 2026 strategy of reducing “the friction, measured by cost and time that the Corporation imposes on the market”? The reading so far suggests it would do so by addressing accidental causes of misalignment.

Julia Black’s approach to fractured interpretations of principles was to support shared understanding within an “interpretive community”. A new communication layer intended to share understanding of evidence of outcomes could foster and support the equivalent structures in the PBO framework.

In CPO, the regulator writes reports, underpinned by trust between the parties, into regulatory systems. This layer instead provides for optional, ongoing and incremental submission of evidence. Each submission carries the business’s own interpretation of the relevant principle. Submissions are attributable to their authors, a feature that contrasts with the reporting mechanisms in CPO. Optional submission is matched by optional responses from the regulator. The rest of the framework would not be targeted by this addition; essential disagreement would continue to be settled by current framework structures.

Proposal features #

The programmable submission interface enables agent applications to interact with the layer. It will define the range of acceptable submission structures for client applications to submit

The layer would present a programmable interface (API) to the market. Submissions to the API, made under the control and wishes of the business, would be comprised of compilations of evidence accompanied by metadata describing the submission. The evidence should describe principle aligned business outcomes or on occasion, unaligned outcomes.

Submission metadata would carry an interpretation attribute recording how the evidence maps to and aligns with the principles. That interpretation could reference guidance or describe the managing agent’s understanding and actions. The metadata would allow the regulator to weight or route messages without imposing new roles or permissions on the sending organisation. An agent’s evidence compilation exercise that previously solely supported narrative attestation may now be spread throughout the business year, providing interpretive layer context for engagement with the regulator.

The API would decouple the technology and behaviours supporting evidence compilation and submission from corresponding responses generated by the oversight functions. Decoupling here enables individual agents to choose different software and approaches in meeting their needs. The variation in evidence compilation approaches would be constrained by the API definition, in effect, a contract on the submission format.
The defined submission interface decouples the interpretive layer from agent application. This allows agents to choose between a spectrum of applications when compiling their evidence

For each agent, year-long evidence compilations would be created with assistance from a software application. Different application types would have varying shadow cost implications, though those cost and efficiency measurements may be elusive.

Managing agents’ accrued costs may correlate with their compilation submission patterns, which they control. One cost is common to all of these applications, arising from material held in line of business systems that offer no point of integration. Such evidence would need to be captured on a case by case basis, and only where an agent elects to surface it. But this shadow cost increase is not a given, as its inclusion as evidence of outcomes would likely result from presentation to management, and therefore be an absorbed or partially absorbed cost.

In Appendix 1, three application types are presented in order of assumed decreasing shadow cost.

In lower shadow cost applications, evidence of behaviours, e.g. timestamped interlinked conversations, is a native feature of these platforms and readily available for inclusion in a compilation. In higher cost applications the same material is reported on, i.e. copied or re-entered. In lower shadow cost applications, the endorsement signal would be captured from organic BAU interaction rather than the deliberate addition of an endorsement step that the web-app and DEC require. Thus we have a gradient of costs required to show engagement at senior management and board level. When these behaviours surface in channel conversations, the evidence of engagement with business outcomes could be efficiently sent to the interpretive layer.

This business communication platform visualisation shows a single compilation up close: an outcome surfacing in a channel through to its submission to the interpretive layer and the regulator’s optional response.

Addressing the challenge of the variety of the businesses in the London Market: Submissions would have an optional response. This supports high touch to low touch businesses and recognises the need for resource decoupling between the regulator and market

There would be a defined set of response and event types for evidence compilation submissions. These would indicate the regulator’s acceptance or rejection or indeed the absence of a response entirely. Acceptance indicates that the interpretation and its evidence are internally consistent and sufficiently documented to support the referenced maturity level. Acceptance wouldn’t grant safe harbour from future assessment under the framework. Rejection may be accompanied by a note of explanation.

Businesses and the regulator would not be coupled by an expectation of mandatory response from sending an evidence compilation to the submission API. Availability of resources, risk assessments, sampling policy and other risk and materiality ratings all factor into the regulator’s response, or lack thereof, to submissions.

Businesses can choose to escalate out of band i.e. using the current framework structures and relationships. The layer provides a vector, i.e. direction and visibility, for the 2022 expressed business need of anticipatory interaction.

An interpretive layer allowing optional responses does not necessarily lead to an inert system. For high touch businesses where the expectation is for close oversight, a standing record accumulating without responses would not be expected and would be available for both sides to escalate. For lower touch businesses, a lack of response may be expected and could lead to businesses tuning their submission rate. Low touch businesses may also pre-emptively request higher regulator engagement.

The design therefore supports differential scaling either side of the API i.e. rate of processing of submissions and the rate of submissions are not tied to each other. Submission could then stand without endorsement from the regulator but available to inform both sides, aligning with framework objective 6. One would expect higher touch businesses, i.e. those perceived to present a greater risk, to have resources allocated in a way that resulted in acceptance or rejection rather than silence.


An interactive simulation helps visualise a conceptual managing agent’s perspective on working with the interpretive layer. The full set of controls is best explored in the simulation itself. Two are highlighted here: the MA transmission rate and the Regulator Response pattern.

The first describes the managing agent’s conceptual transparency with respect to the fraction of evidence that is transmitted to the layer. Turning it down to zero effectively simulates no interpretive layer at all, the current state. The second control plays out preset evidence response patterns in the simulation. Each preset groups regulator resourcing levels, response latencies and accuracy of response analysis.

Starting the conversation about the interpretive layer #

The interpretive layer would leave all other aspects of the PBO framework intact. The two views of business principle alignment still meet for end of year resolution. Now those views would have a year of written interpreted evidence of principle outcomes to draw on. Through that standing record, interpretation differences would have had the opportunity to be contained and responded to throughout the year.

The 10000 ft simulation of the agent’s view of the interpretive layer is intended to stimulate conversation on modelling new behaviours and resourcing around the layer. In order for it to support investment cases, the model will need further development and refinement. The current simulation supports varying parameters to compare scenarios and reveal how the outputs move in response. Absolute values have not been validated and would require calibration against empirical data.

This article aims to foster conversation around its open questions. It also welcomes requests for clarification, and hopes to receive correction or refinement of the arguments made throughout. The simulations need your input. Your guidance may support the development of other views onto the interpretive layer, such as a regulator’s aggregate view of managing agents.

Reach me at / linkedin.com/in/john-d-harnett.

— Appendix 1 — Options for compiling evidence

Compiling evidence of principle alignment and its supporting discussions will be assisted by software applications. These will vary on how the evidence is captured and also how the resulting endorsements are documented and compiled. Choosing between options will be affected by current software and communication practices. An organisation’s shadow cost due to evidence compilation and submission will be dependent on multiple factors, more than are shown here, but the submission API provides agents with a choice and ability to move along the application spectrum. The three options are shown in order of assumed decreasing shadow cost, based on the axes of capture of evidence and endorsement. But the initial cost of adoption, for organisations that are not prepared to extend their default email and document systems or are not on the path to discussing business outcomes on business communication platforms, makes the total cost of ownership a more complex calculation.

Web app, a bespoke data-entry tool. Requires uploading of evidence from source applications and reporting or recreation of the endorsements

Conceptually separate from the agent’s current systems. Compiling evidence and its submission are at a relatively higher cost than the subsequent options.

The web app would support dragging in evidence from other surfaces where business outcomes are discussed and stored. It would enable reference to the principles, guidance and entering interpretation and endorsements to complete the structured message. It would require a model of review and endorsement of compiled evidence before submission.

Distributed Evidence Compilation (DEC) extends applications where evidence is recorded to support the compilation process. Endorsements would have to be recreated in DEC in a similar manner to the Web app i.e. re-keying and reporting of endorsement that had previously been made

These applications, likely to be Outlook and SharePoint or variants on these, would tend not to have native support of endorsement and discussions associated with source documents. DEC is intended to have lower shadow cost implications than the browser based web app. DEC may also extend browser based applications, where these expose APIs accessible to third parties. Add-ins and APIs lower the context switching costs between the source of evidence of business outcomes and its compilation for submission to the API. Endorsement and conversation around compiled artifacts would be similar to the experience of using the web-app briefly described above. DEC would require coordination of the compilation and submission processes. Much of this can be supplied as a service to managing agents. New shadow costs may arise from managing the state of evidence compilations via a new user interface which would be separate from the source applications. The new user interface would have a similar user experience to the web app previously discussed.

Business communication platforms most readily support capture of evidence and endorsements. Timestamped messages, user identities and reactions mean evidence and endorsement both arise natively in channels, so only minimal extension is needed to submit ‘complete’ evidence packages

Increasing the investment in building out DEC, perhaps to reduce the coordination costs, starts to make it look more like the out-of-the-box experience of the third option, a business communication platform.

These platforms support endorsement through both messages and the lighter weight reactions that may once have been thought trivialisations but in the aggregate are significant.

Where email generally flows directly between people, conversations on business communication platforms happen in channels, which support focused discussion and responses to uploaded documentation. In channels where the focus would be on outcomes of doing business at Lloyd’s, generators of accidental cost seen in DEC and the web-app would be comparatively reduced. The reduction follows from evidence of behaviours and the endorsement signal both arising in the channel itself rather than in a separate compilation step.

Compilation and submission would be assisted by a custom chatbot: a ‘principles bot’. A simulation of a principles bot being used within its host business communication platform shows a low profile and efficient addition to base platform behaviour.