Stage-Gate Governance Model & Gate Review Board Charter — Version 2.1

Lighthouse Financial Services Company — This document is the governing instrument for the Beacon Index Advantage product development program. It establishes the Gate Review Board, defines what the Board may decide and what it may not, sets out the five gates and the five outcomes available at each, and specifies how program funding is released, withheld and returned. It is not a project management plan. A project management plan answers how will this be delivered. This document answers a prior question: should this continue to exist — and it answers it five times, on the record, with a vote.

1. Governance Philosophy & Principles

New product development at an insurance carrier consumes capital and specialist capacity long before it produces a single policy. The pricing actuary who models a rider cannot simultaneously support in-force experience studies; the configuration capacity spent extending the annuity administration platform is capacity not spent on servicing releases. Every month a weak product concept survives is a month of scarce actuarial, legal and platform capacity that a stronger concept does not get. The purpose of this governance model is to make the decision to stop as easy to take, and as well evidenced, as the decision to continue.

Five principles govern every gate in this program.

1. Continuation is earned, not assumed. The default at every gate is that funding stops. The program team carries the burden of showing why it should not.
2. Money follows the decision, not the plan. No stage is funded before the gate that authorises it. The program has an approved cost of $27,904,000; it does not have $27,904,000.
3. The people who bear the risk hold the votes. Actuarial owns pricing risk, Compliance owns regulatory risk, Distribution owns the revenue assumption, Finance owns the capital. Each holds a vote. The Program Manager holds none.
4. Evidence quality is separable from the recommendation. The Board may reject a decision package on the quality of its evidence without forming a view on the product. This is what a recycle is for, and it is not a soft rejection.
5. Every decision is recorded, including the ones that hurt. Recycles, conditions and cancellations are entered in the register with the same formality as approvals. A governance record that only contains approvals is a record of nothing.

2. What Stage-Gate Governs That Project Management Does Not

This program also has a project management layer — a schedule, a resource plan, a risk register, a change process. That layer answers questions of execution. It does not, and cannot, answer the question this document exists for. A schedule can tell you a product will launch four months late. It cannot tell you the product should not launch at all.

The distinction matters because the two layers fail differently. Project management fails by delivering late or over budget. Stage-gate governance fails by delivering exactly what was asked for into a market that no longer wants it, or at a price the capital cannot support. The second failure is more expensive and considerably harder to see coming, because every execution metric is green right up until launch.

3. Why This Program Is Documented Mid-Flight

This program is documented at Stage 2 of 4, with three gates still ahead of it, and that is a deliberate choice rather than an incomplete record.

A completed stage-gate program is, in hindsight, indistinguishable from a waterfall program. Look backward at a launched product and every gate reads GO, the conditions have all closed, the recycle has vanished into the rearview, and four separate funding releases look like a single budget. The governance is invisible precisely because it worked. It is only observable while the decisions are still open — while conditions are outstanding, while money is unreleased, while a gate has not yet voted.

4. Board Mandate & Standing Authority

The Gate Review Board is a standing cross-functional body constituted by the Chief Product Officer to hold decision rights over the Beacon Index Advantage program at each defined gate. Its authority is specific and bounded.

4.1 The Board may

• Grant, withhold or condition the release of any stage funding tranche.
• Cancel the program at any gate, and outside a gate on a Chair-convened extraordinary session where a material adverse finding cannot wait.
• Return a decision package to the program for remediation without deciding on the product.
• Impose binding conditions on a continuation decision, with named owners and due dates.
• Authorise draws against the contingency reserve.
• Reduce program scope, including removing product features and target jurisdictions.

4.2 The Board may not

• Direct the program team's day-to-day work, sequencing or staffing. Execution sits with the Program Manager and the functional leads.
• Approve a filing submission to any state insurance department. That authority rests with Legal & Compliance and the appointed illustration actuary and is not delegable to this Board.
• Waive an actuarial sign-off, a compliance sign-off, or any statutory requirement. Where a required sign-off is withheld, the Board's only available outcomes are RECYCLE, HOLD or CANCEL.
• Release funds for a stage whose entry gate has not been passed.

5. Composition, Seats & Voting Rights

Seats are held by role, not by individual. A member may send a delegate holding written authority from the seat-holder; a delegate may vote. No individual may hold two seats, and no member may vote on a gate at which they are also the accountable owner of an open condition — see §6.3.

6. Quorum, Voting Rules & Conflict of Interest

6.1 Quorum

Quorum is four of the five voting seats, and must include the Chief Actuary and the General Counsel & Chief Compliance Officer. A gate cannot be held without both, because both hold blocking sign-offs. If quorum fails the gate is adjourned, not defaulted; the stage tranche remains unreleased and the schedule absorbs the delay.

6.2 Carrying a decision

• GO or GO WITH CONDITIONS requires a simple majority of votes cast, with no blocking sign-off withheld.
• CANCEL requires a simple majority. It does not require unanimity, and the Board is explicitly instructed not to treat cancellation as requiring a higher bar than continuation. An asymmetric bar is how organisations acquire projects they cannot stop.
• RECYCLE may be carried by any two voting members, without majority. Evidence quality is deliberately the cheapest objection to raise.
• Abstention is recorded with a stated reason and does not count toward the majority.

6.3 Conflict of interest

A voting member who is the accountable owner of an open gate condition abstains from the vote on the gate at which that condition is verified. At Gate 2 this affects N. Adeyemi (owner of the Chief Actuary sign-off chain behind GC-03) and R. Castellanos would have been affected by GC-04 had it not closed on 14 August 2026. Where abstention would break quorum, the condition is verified by an independent reviewer appointed by the Chair before the gate convenes.

7. The Chair — Facilitation Without a Vote

The Program Manager chairs the Gate Review Board and does not hold a vote. The Chair is seated from the Programme Management Office and is not a member of the Board.

This is the single most consequential structural choice in this model, and it is worth stating why. The Program Manager assembles the decision package, commissions the analysis inside it, and has spent months building the thing being judged. That is precisely the profile of someone who should not vote on it. Separating the two roles means the Chair's incentive is aligned to the quality of the decision rather than to a particular outcome, and it means a CANCEL is not a personal defeat for the person running the process.

Chair responsibilities

• Issue the decision package no fewer than ten business days before the gate. A package issued late is grounds for automatic adjournment at the request of any voting member.
• Confirm gate entry criteria are met before convening, and decline to convene where they are not.
• Table dissenting analysis alongside the recommendation. Suppressing a dissent is a governance breach, not an editorial judgement.
• Record the decision, the vote, all conditions with owners and dates, and the reasons, within two business days.
• Track conditions to closure and verify evidence before the next gate.
• Convene an extraordinary session where a material adverse finding cannot wait for the next gate.

8. The Five Gates & the Stage Model

Stages are periods of funded work. Gates are decision events between them. Gates are held on a Thursday; stages begin on a Monday and end on a Friday. The calendar is planned backward from the launch date of 06 March 2028, which is set by the distribution cycle and is the fixed point from which every preceding date derives.

Gate 5 differs from the others in kind. It releases no funding and cannot stop a product that is already in the market. Its purpose is accountability for the business case: it reviews the post-launch measurement window of 06 Mar 2028 to 01 Sep 2028 against what Gate 1 was told, and transfers benefit ownership to the operating line. A program that never holds Gate 5 has no mechanism by which its forecasts are ever checked, which is how optimistic business cases become a habit rather than a mistake.

9. Gate Outcomes Defined

Five outcomes are available at every gate. They are defined here because a governance model in which "go" and "not yet" are the only real options is not a governance model.

10. Gate Entry Criteria & the Decision Package

A gate may not be convened until entry criteria are met. This is a Chair responsibility and is not subject to a vote — a Board cannot decide well on a package that should not have reached it. Every gate package contains, at minimum:

1. The recommendation, stated in one sentence, with the outcome being sought.
2. The updated business case — volumes, pricing, capital, hurdle rate, and every assumption that has changed since the last gate, with the change flagged rather than absorbed.
3. Actuarial position on pricing and reserve adequacy, signed or explicitly withheld.
4. Compliance position on filing, forms and suitability, signed or explicitly withheld.
5. Status of every open condition from prior gates, with evidence.
6. The risk position — movements on R-01 through R-12, and any newly realised issue.
7. Financials — spend against the released tranche, the tranche requested, and any contingency drawn since the last gate.
8. The dissent section. Any material analysis that does not support the recommendation is tabled with it, attributed, and not summarised away.

11. Gate Conditions — Issue, Track, Verify, Close

A condition is a binding obligation attached to a continuation decision. It is not a recommendation and it is not an action item. Conditions carry an owner who is a named individual, a due date that precedes the next gate, and a verification method stated at the time of issue rather than negotiated at closure.

Five conditions were issued at Gate 1. Three are closed and two are open.

GC-03 and GC-05 both fall due before Gate 2 on 01 April 2027. Neither is discretionary: GC-03 speaks to R-06 (pricing assumptions failing external review) and GC-05 to R-05 (hedging readiness lagging launch). If either is unverified when Gate 2 convenes, the Chair will table a recycle recommendation.

12. The Recycle Protocol

A recycle returns the package without deciding the product. It is the mechanism that lets a Board say we cannot responsibly decide on this evidence without being forced into a false choice between approving something unproven and canceling something that may be sound.

On a recycle the Board must state, in the minute, the specific deficiencies to be remedied and the date by which the package returns. The stage tranche is not released; the program continues to draw only on its current stage funding, and any overrun caused by the loop is a contingency draw requiring Board authorisation under §15.

The loop is recorded here rather than absorbed into the Gate 1 approval because the cost of the recycle is a real cost of this program and because the alternative — a Gate 1 that reads simply as approved — would misrepresent how the decision was actually reached.

13. The Cancellation Protocol & Orderly Wind-Down

Cancellation is an available outcome at every gate and, on a Chair-convened extraordinary session, between gates. The following applies on a CANCEL.

1. Funding stops immediately. Unreleased tranches are cancelled. No further commitments may be entered from the date of the decision.
2. Final accounting within twenty business days — released funds, funds spent, and unspent funds returned to the capital pool, reported to the CFO and the Executive Committee.
3. Committed external obligations are settled, not abandoned. Vendor termination-for- convenience provisions are exercised; the Vendor Management Plan governs.
4. The team is released on a stated date, with reassignment handled through the functional leads rather than left to attrition.
5. Work product is preserved and indexed — pricing models, filing drafts, market research. A cancelled concept frequently contains the components of a viable one, and the value of that material is lost entirely if nobody is made responsible for keeping it.
6. A cancellation record is filed stating the decision, the vote, the rationale and the final accounting.
7. Post-cancellation re-entry is permitted only at Gate 0, with a new business case and new funding. See §18.

13.1 Published cancellation thresholds for the gates ahead

The Board has recorded, in advance, conditions under which it would expect to stop the programme rather than continue. Publishing these before the gate is deliberate: a threshold agreed in advance is considerably harder to rationalise away than one debated in the room.

13.2 Cancellation applied to scope

Cancellation authority is exercised at scope level more often than at program level, and two instances are on the record. At Gate 1 the Board cancelled two of the five proposed crediting strategies (D-03, condition GC-01), returning their build allocation to the reserve; the decision was informed by I-02, the illustration engine's inability to support the hypothetical-performance presentation those two structures required. A proposed direct-to-consumer application channel is tabled for a cancellation decision at Gate 2 on the same basis — it cannot be supported inside the released tranche and does not carry its own case.

14. Tranche-Based Funding Release

The program has an authorised cost of $27,904,000, comprising a base of $25,600,000 and a gate contingency reserve of $2,304,000 (9% of base). The base is divided into four stage tranches, each released only by the gate that authorises the stage it funds.

Stage 1 closed over its tranche, at $2,412,000 against $2,180,000. The overrun is exactly the cost of the Gate 1 recycle loop and was met by an authorised contingency draw rather than by silently absorbing it into the Stage 2 release — which would have concealed the cost of the recycle inside development spend.

15. Contingency Reserve Authority

This is a deliberate departure from the conventional arrangement in which a program manager holds a reserve and reports its consumption. In a stage-gate program the reserve is not a buffer against execution variance — it is the mechanism by which the Board funds the consequences of its own decisions. The recycle it paid for at Gate 1 was the Board's decision, not the program's.

16. Performance Reporting Within the Released Tranche

Performance is therefore reported within the released tranche only. The Program Dashboard, the Weekly Status Report and every gate package state cost and schedule performance against the current stage, name the stage, and show unreleased tranches as unreleased rather than as future baseline. Cumulative program spend is reported as a fact — $6,950,000 to date — not as performance against a baseline that does not exist.

Readers accustomed to program-level earned value should read this as a feature of the funding model rather than a gap in the reporting. The two are not compatible: earned value assumes a scope baseline authorised end to end, and this program is authorised one stage at a time.

17. Gate Decision Log to Date

18. Precedent — Why This Is a New Program and Not a Restart

This Board has cancelled a product development program before. The Harborline Buffer Series, a registered index-linked annuity concept, passed Gate 0 in March 2025 and was cancelled at Gate 1 on 7 August 2025. The recorded rationale was that registration, broker-dealer distribution licensing and the associated compliance obligations loaded cost and elapsed time the business case could not carry, and that the buffered structure required hedging capability beyond the in-house desk. The cancellation protocol at §13 was followed; the final accounting and the full decision record are in the Prior Concept Cancellation Record.

The Beacon Index Advantage concept is a successor to that decision, and the relationship between the two requires stating precisely, because the failure mode here is well known.

The distinction is procedural and it is auditable. Harborline was closed: funding cancelled, final accounting filed, unspent funds returned, team released. The successor concept was re-screened from Gate 0 on 05 February 2026 with a new business case, a new roster and new funding, and had to earn its Stage 1 tranche on its own evidence. Nothing carried across but the learning — specifically, that the registration burden was the binding constraint, which is why this product is a non-registered fixed indexed annuity filed through the state insurance lane rather than a registered product (D-01).

Recorded plainly: the reason this program exists in the form it does is that the Board cancelled the previous version of it.

19. Decision Record, Related Documents & Control

19.1 Governing decisions carried by this document

19.2 Related documents

• Gate Decision Framework — scoring criteria and the evidence standard applied at each gate.
• Gate 1 Business Case Package and the Gate 1 Recycle Memorandum — the record behind §12.
• Gate Conditions Register — the live GC-## record maintained between gates.
• Program Budget — the tranche release schedule and the labour baseline behind §14.
• Prior Concept Cancellation Record — the Harborline decision and final accounting (§18).
• RAIDD Log — R-01 to R-12, A-01 to A-08, I-01 to I-06, DEP-01 to DEP-07, D-01 to D-11.

19.3 Document control

Owner: C. Tyrrell, NPD Program Manager and Chair of the Gate Review Board. Approving authority: G. Marchetti, Chief Product Officer. Review cadence: at each gate, and on any change to Board composition or funding authority.