Lighthouse Financial Services Company — Gate 1 convened on 30 April 2026 and did not carry. This session was held on 04 May 2026, two business days later, to put the findings, their mathematics and the available options in front of the Executive Committee before any money was committed to a second attempt.
1. Why This Session Exists
A recycle is a normal gate outcome and does not by itself require an executive session. Two things about this one did.
- The reported return changed. One of the findings was an error in the capital charge that moved the program's projected return by 120 basis points. The Committee approved a business case; the case it approved was not the case that exists.
- CANCEL was live. $11,640,000 had not yet been committed, and this Board had cancelled a predecessor concept at its own Gate 1 the previous year. Presenting the options without CANCEL among them would have made the session decorative.
2. Gate 1 Was Not Scored
A low score records disagreement about evidence that exists. A U records that no defensible evidence exists yet. Had the scale lacked a U, an assessor would have entered a 3 against each — a value that reads as “adequate” and disappears into a weighted average. The program would have proceeded, and $11,640,000 would have been committed against a capital charge that was wrong.
3. Finding 1 — The Capital Charge
| Basis | Capital ratio | Projected IRR | Headroom vs 11.0% hurdle |
|---|---|---|---|
| As presented at the first convening | 3.33% | 14.6% | 360 bp |
| Corrected | 4.2% | 13.4% | 240 bp |
The charge was understated by 87 basis points. Because required capital is held against account value throughout the life of the block, an error in that ratio compounds across the whole projection rather than affecting a single year. The corrected case still clears the hurdle — by 240 basis points instead of 360.
4. Finding 2 — Distribution Evidence
The second U was distribution commitment. The Year 1 volume forecast rested on internal modeling with no written evidence from any channel partner. Nothing about the forecast was shown to be wrong — it was shown to be unevidenced, which is a different finding and the one the U rating exists to record.
How much volume can the corrected case absorb?
| Volume vs plan | IRR | NPV at hurdle | Verdict |
|---|---|---|---|
| +0% | 13.4% | $11,009,118 | clears |
| -5% | 12.9% | $8,618,608 | clears |
| -10% | 12.4% | $6,228,098 | clears |
| -15% | 11.9% | $3,837,588 | clears |
| -20% | 11.4% | $1,447,077 | clears |
| -25% | 10.8% | $-943,433 | FAILS |
| -30% | 10.1% | $-3,333,943 | FAILS |
5. What the Loop Costs
| Item | Amount |
|---|---|
| Retained core team - pricing, product, distribution analytics (6 weeks) | $148,000 |
| External capital-charge refresh and recomputation support | $34,000 |
| Channel commitment negotiation and legal review of indications | $28,000 |
| Extended program management and gate re-preparation | $22,000 |
| Total, drawn from the gate contingency reserve | $232,000 |
Six weeks of elapsed time, and the launch date does not move — so Stage 2 compresses from a planned 46 weeks to 40. The loop is paid for in schedule float, not in launch date.
6. Root Cause
7. Five Options, With Pros and Cons
Option A — Return to Gate 1 on the corrected basis Recommended
Close both findings, re-table the package at the second convening, absorb the loop.
For
- The corrected case still clears the hurdle by 240 basis points (13.4% vs 11.0%).
- No repricing, so no competitive exposure on the illustrated cap.
- The launch date is preserved; the loop is absorbed by compressing Stage 2.
- Distribution evidence is obtainable inside the six-week window.
Against
- Costs $232,000 and six weeks.
- Stage 2 compresses from 46 weeks to 40, removing the float that would have absorbed a later problem.
- Headroom is materially thinner than the Board was originally shown (360bp → 240bp).
Option B — Re-price to restore the original return
Lower the illustrated cap to widen margin and recover the 120 basis points.
For
- Arithmetically effective: 10bp of additional margin returns 14.9%, above the originally presented 14.6%.
- Requires no schedule change.
Against
- The gain evaporates at a volume cost the program is already exposed to. If the lower cap costs 10% of volume, the return falls back to 13.9% — a net gain of only 49bp over doing nothing.
- The cap is 9.25% against a peer 9.00% and a competitive floor of 8.50% fixed at this gate. There is little room to move.
- It trades a return problem for a volume problem, and volume is the variable the model says actually moves the answer.
Option C — Reduce program cost further
Cut build scope to lower the authorized program cost.
For
- Arithmetically effective, and more so than expected: removing $2,000,000 of build cost returns 13.91%, 51bp above the corrected case.
Against
- The scope reduction has already happened. Two crediting strategies were cut at this same gate under GC-01; there is no comparable slack left.
- What remains is people, filing fees, platform configuration and distribution enablement. Cutting the filing footprint or the enablement spend is cutting volume — which routes straight back to Option B's objection.
- It treats a governance failure as a budget problem.
Option D — HOLD pending the annual assumption refresh
Pause the gate until Corporate Actuarial's refresh (DEP-07) lands.
For
- Decides once, on better data, rather than twice.
- Removes the risk of re-pricing again at Gate 2.
Against
- The refresh lands after the launch window this product is aimed at.
- HOLD is defined for causes outside the program's control. This cause is inside it — the Chair's own sequencing. Using HOLD here would misuse the outcome and set a precedent for relabeling internal failures as external ones.
- Competitor launches are expected in the same quarter (R-07).
Option E — CANCEL and return the unspent Stage 1 tranche
End the program now, before the Stage 2 tranche is committed.
For
- Avoids committing $11,640,000, the largest single tranche.
- This Board has done it before and knows how — the predecessor concept was cancelled at its own Gate 1 with the unspent tranche returned to capital.
- The error was real and the Committee is entitled to ask whether it is the only one.
Against
- The corrected case still clears the hurdle by 240 basis points. Cancelling a case that clears would be an over-correction — punishing the program for the quality of its evidence rather than the quality of its economics.
- The shelf-refresh need does not go away; it returns as a later, more expensive program with the same distribution problem.
- Stage 1 capability — pricing basis, filing route, platform assessment — is lost and would be rebuilt.
8. Every Lever Routes Back to Volume
The options were analyzed independently and converged on one finding, which is the deck's central slide.
Volume is the one variable that moves the answer, because it is the only one fixed cost does not scale with. Written commitments stand at $168,000,000 against a Year 1 target of $185,000,000 — the program is already spending its volume tolerance. Any option that spends more of it to buy back return is selling the thing the case depends on to fix the thing the case can survive.
9. Recommendation
The recommendation carries two conditions the Chair proposes attaching at the second convening: an independent external review of the rider pricing basis, and written re-validation of Year 1 volume with the three largest channel partners. Both later became GC-03 and GC-04.
10. Basis of the Mathematics
Every figure in this session is produced by the same cash-flow model that backs the Cost-Benefit Analysis, which reproduces the program's locked return figures to within half a basis point or refuses to load. The capital ratio presented at the first convening (3.33%) is recovered by solving the model backward from the 14.6% that was tabled — it is derived from the record, not reconstructed from memory.
Prepared by C. Tyrrell, NPD Program Manager and Chair of the Gate Review Board. Related: Gate 1 Recycle Memorandum · Gate 1 Business Case Package · Cost-Benefit Analysis · Gate Conditions Register.