← Stage-Gate NPD Suite Gate Assessment Standard

Gate Decision Framework — Version 2.0

Download Word
Programme timeline · status 16 Oct 2026Read the full story →
Harborline
Aug 2025
Cancelled
Gate 0
Feb 2026
Go
Stage 1
Business case
Gate 1
Apr 2026
Recycled
Gate 1
Jun 2026
Go w/ conditions
Stage 2
Development
You are here
Gate 2
Apr 2027
Gate 3
Oct 2027
Gate 4
Feb 2028
Launch
Mar 2028
Gate 5
Sep 2028

Lighthouse Financial Services Company — The Stage-Gate Governance Model establishes who decides and how the process runs. This document establishes what is assessed and to what standard. It defines the two-tier test applied at every gate, the evidence that counts, the score bands and their relationship to the five outcomes, the specific criteria for each of the six gates, and the scored record of the gates already held.

The framework structures the decision. It does not make it. A weighted score is an input to a vote, never a substitute for one. Nothing in this document obliges the Gate Review Board to approve a program that scores well or to cancel one that scores badly. Its purpose is to make the Board argue about the right things, in the same terms, at every gate — and to make a departure from the score visible and attributable rather than quiet.

Table of Contents

Part I — The Assessment Method
  1. Purpose & Relationship to the Governance Model
  2. The Two-Tier Test — Must-Meet and Should-Meet
  3. The Scoring Scale
  4. The Evidence Standard
  5. Score Bands & Their Relationship to Outcomes
  6. Criteria Are Fixed at the Prior Gate
Part II — Criteria by Gate
  1. Gate 0 — Concept Screening
  2. Gate 1 — Business Case
  3. Gate 2 — Development Complete & Filing Readiness
  4. Gate 3 — Validation & Filing Approval
  5. Gate 4 — Launch Readiness
  6. Gate 5 — Post-Launch Review
Part III — The Record
  1. Scored Record — Gates Held to Date
  2. Why the Gate 1 Recycle Was Not a Low Score
  3. Document Control
Part I — The Assessment Method

1. Purpose & Relationship to the Governance Model

Two documents govern gate decisions on this program and they do different work. The Governance Model defines the Board, its authority and its limits, the outcomes available, quorum and voting, and the protocols for recycling and cancelling. This framework defines the questions the Board answers and the standard of proof it requires before answering them.

The separation matters in practice. Authority questions change rarely; assessment criteria change at every gate, because what you need to know about a product at concept screening is not what you need to know about it eight weeks before it goes on sale. Holding them in one document would mean revising the constitution every time the criteria moved.

2. The Two-Tier Test — Must-Meet and Should-Meet

Every gate applies two tiers in sequence. The first is binary and the second is scored. The order is not negotiable: a package that fails the first tier is not scored at all, because a weighted average is an excellent way to bury a disqualifying fact under five satisfactory ones.

Tier 1 — Must-Meet (knockout, binary, unweighted)

Each must-meet criterion is answered yes or no. A single no removes GO and GO WITH CONDITIONS from the outcomes available at that gate, whatever the Tier 2 score would have been. The Board's remaining choices are RECYCLE, HOLD or CANCEL.

Must-meet criteria are deliberately few and deliberately absolute. They exist to capture the conditions under which continuing would be indefensible rather than merely unattractive — a product that cannot lawfully be sold, a price the capital cannot support, a guarantee that cannot be hedged.

Tier 2 — Should-Meet (weighted, scored 1–5)

Should-meet criteria capture attractiveness rather than admissibility. Each is scored 1 to 5 and weighted; the weights total 100% and are set for each gate in advance under §6. Weightings shift across the program's life: early gates weight market and strategic questions most heavily, later gates weight execution and operational readiness, because by Gate 4 the market question has already been answered and the open risk is whether the organisation can actually deliver.

3. The Scoring Scale

Terminology: this programme labels the gate outcomes GO, CANCEL, HOLD, RECYCLE and GO WITH CONDITIONS. Robert G. Cooper's Stage-Gate framework uses a blunter word where this says CANCEL — same decision, same authority. See the Governance Model §9.

ScoreMeaningWhat it implies for the decision
5Strong — materially better than the comparable alternative, evidencedSupports GO
4Good — meets the standard set at the prior gate, evidencedSupports GO
3Adequate — acceptable but with a known weakness that is being managedSupports continuation, usually with a condition attached
2Weak — below the standard set at the prior gate, with no credible remediationArgues against continuation
1UnacceptableArgues for cancellation
UUnscoreable — the evidence to score this criterion is absent or not credibleAutomatic recycle recommendation. See §4 and §14
The U is the most important value on this scale. Without it, an assessor faced with missing evidence has to invent a number, and the number that gets invented is almost always a 3 — which reads as "adequate" and passes quietly into the weighted average. Recording U forces the gap to surface as a gap. The Gate 1 recycle on this program was driven by two U ratings, not by a low score.

4. The Evidence Standard

A criterion is scoreable only where it rests on evidence of the following kinds. This list is restrictive by design.

Counts as evidenceDoes not count
A written commitment from an accountable party (a signed channel commitment, a vendor statement of work, an executed term sheet)A verbal indication, a meeting note, or a relationship manager's confidence
An actuarial or compliance opinion, signed, with its assumptions statedAn opinion in preparation, or one whose assumptions are not disclosed
A model output whose inputs are current and whose version is identifiedA model output whose inputs have been superseded — see §14
Observed data from a comparable in-force product or a completed pilotAn analogue drawn from a materially different product or market
A regulator's written determination or published standardAn expectation of how a regulator will respond

Where evidence is of the right kind but its currency is in doubt, the assessor records the date of the underlying input alongside the score. A score supported by a nine-month-old capital charge is not the same score as one supported by this quarter's.

5. Score Bands & Their Relationship to Outcomes

Bands indicate which outcomes the evidence supports. They do not select one.

≥ 3.50All must-meet passed. Package supports GO.
3.00 – 3.49Continuation supportable, but with identified weaknesses. Supports GO WITH CONDITIONS; each weakness scoring 3 or below should carry a condition.
2.50 – 2.99Continuation not supportable on this evidence. Supports RECYCLE, or HOLD where the cause is external.
< 2.50Supports a CANCEL recommendation from the Chair.

Overriding rules

  1. Any must-meet failed → GO and GO WITH CONDITIONS are unavailable regardless of the weighted score.
  2. Any criterion scored U → the Chair tables a RECYCLE recommendation. The Board may override, but the override and its reasons are recorded.
  3. A Board decision that departs from the band is permitted and is minuted with the reasons. Departures are reviewed at Gate 5 — a framework whose bands are routinely overridden is not describing how the organisation actually decides, and should be changed rather than ignored.

6. Criteria Are Fixed at the Prior Gate

The criteria and weights for Gate N are locked at Gate N−1, recorded in that gate's minute, and may not be altered afterwards except by a Board decision that is itself minuted.

This is an anti-gaming provision and it cuts both ways. A program team cannot be judged at Gate 2 against criteria invented after Stage 2 was already spent; equally, the team cannot negotiate the bar downward once it becomes clear the product will struggle against it. The Gate 2 criteria at §9 were fixed at Gate 1 on 11 June 2026 and have not moved since.

Part II — Criteria by Gate

7. Gate 0 — Concept Screening

Gate 0 asks one question: is this concept worth the cost of a business case? It is a cheap gate by design and its must-meet list is short, because the purpose of Stage 1 is precisely to establish the facts that later gates will test.

Must-meet

Should-meet — scored 05 February 2026

CriterionWeightScoreWeighted
Strategic fit with the annuity portfolio25%4 / 51.00
Market attractiveness25%4 / 51.00
Feasibility — can it be built and filed20%4 / 50.80
Distribution appetite15%4 / 50.60
Preliminary economics15%3 / 50.45
Weighted total100%3.85

Outcome GO, carried 5–0. Stage 1 released at $2,180,000. No conditions attached. The single score of 3 — preliminary economics — was expected at this gate and is the reason Stage 1 exists.

8. Gate 1 — Business Case

Gate 1 is the first gate with real money behind it: it releases $11,640,000, the largest single tranche in the program. Its must-meet list is correspondingly demanding.

Must-meet

Should-meet — scored 11 June 2026 (second convening)

CriterionWeightScoreWeighted
Market attractiveness20%4 / 50.80
Competitive position — illustrated cap against peer set15%3 / 50.45
Distribution commitment20%3 / 50.60
Financial return against hurdle20%3 / 50.60
Technical & operational feasibility15%3 / 50.45
Risk profile10%3 / 50.30
Weighted total100%3.20

Outcome GO WITH CONDITIONS, carried 4–0–1. The score of 3.20 sits in the conditions band, and the four criteria scoring 3 map directly onto the conditions issued: competitive position and financial return to GC-03 (external pricing peer review), distribution commitment to GC-04, feasibility to GC-01 and GC-05. This is the intended behaviour of the framework — every weakness that supports a continuation decision should leave the room with an owner attached to it.

9. Gate 2 — Development Complete & Filing Readiness

Criteria fixed at Gate 1 on 11 June 2026. Gate convenes 01 April 2027; releases $7,450,000.

Must-meet

Should-meet

CriterionWeight
Filing readiness — completeness of the Compact submission package25%
Pricing robustness under the external review and current option budget25%
Platform & illustration build completeness against specification20%
Competitive position of the illustrated product at current rates20%
Distribution readiness — channel and appointment pipeline10%

10. Gate 3 — Validation & Filing Approval

Convenes 28 October 2027; releases $4,330,000. This is the gate at which an external party — the regulator — holds a decisive input the program cannot influence, which is why the must-meet list here is almost entirely outside the team's control.

Must-meet

Should-meet

CriterionWeight
Breadth of approved jurisdictions against the launch plan25%
Validation results — end-to-end issue, illustration and servicing25%
Retained economics at approved rates25%
Operational readiness of new business and policyholder services15%
Residual risk position against the register10%

11. Gate 4 — Launch Readiness

Convenes 24 February 2028; authorises launch on 06 March 2028. Gate 4 releases no new tranche. It answers whether the organisation can sell and service what it has built, and the weighting shifts decisively toward execution.

Must-meet

Should-meet

CriterionWeight
Distribution readiness — wholesaler coverage and pipeline30%
Operational readiness — capacity against forecast volumes30%
Marketing and advisor enablement completeness20%
Competitive position at launch rates20%

12. Gate 5 — Post-Launch Review

Convenes 07 September 2028, reviewing the window 06 Mar 2028 to 01 Sep 2028. Gate 5 is not an authorisation gate and does not use the score bands. It cannot stop a product already in the market. It exists to hold the business case to account and to transfer benefit ownership to the operating line.

What Gate 5 assesses

  1. Actual against forecast — sales volume, average case size, crediting strategy mix, persistency and rider election, each against what Gate 1 was told.
  2. Where the business case was wrong, and in which direction. Recorded specifically, not as an aggregate variance.
  3. Whether the conditions issued at Gate 1 achieved what they were issued to achieve. A condition that closed on schedule and changed nothing is a finding.
  4. Whether Board decisions departed from the score bands, and whether those departures proved sound (§5).
  5. Benefit ownership transfer to the product line, with named owners and a measurement cadence that outlives the program.
A program that never holds Gate 5 has no mechanism by which its own forecasts are ever checked against what happened. That is how optimistic business cases stop being a mistake and become a habit: nobody is ever present at the moment the forecast is disproved.
Part III — The Record

13. Scored Record — Gates Held to Date

GateDateMust-meetWeighted scoreBandDecision
Gate 005 Feb 20263 of 3 passed3.85GOGO 5–0
Gate 1 (first)30 Apr 2026Not reached — 2 criteria scored UNot scoredRECYCLE
Gate 1 (returned)11 Jun 20265 of 5 passed3.20CONDITIONSGO WITH CONDITIONS 4–0–1
Gate 201 Apr 2027Not yet convened — criteria fixed at §9; GC-03 and GC-05 must close first

14. Why the Gate 1 Recycle Was Not a Low Score

The distinction this section draws is the reason the framework has a U value at all, and it is the most frequently misread event in the program's record.

At the first convening of Gate 1 on 30 April 2026, the package was not scored. Two criteria could not be assessed on the evidence presented:

  1. Distribution commitment — U. The volume assumptions had been built from internal channel modelling. No written commitment from any partner was in the pack, and the evidence standard at §4 does not admit a relationship manager's confidence as a substitute.
  2. Financial return against hurdle — U. The return calculation used a capital charge that had been superseded. The output was of the right kind but its input was not current, which §4 places outside the standard.

Neither finding said the product was weak. Both said the Board could not responsibly form a view either way — and a Board that votes anyway in that position is not governing, it is guessing with a quorum. The package returned on 11 June 2026 with written commitments from three partners and a recomputed return on the current charge, and scored 3.20.

The counterfactual is the point. Had the two unscoreable criteria been assigned a default 3 instead of a U, the package would have scored in the conditions band at the first convening and Stage 2 would have been released on 30 April 2026 — six weeks earlier, and on a distribution assumption that no partner had confirmed and a return computed on a superseded charge. The recycle cost $232,000 and six weeks. Releasing $11,640,000 against unverified volume assumptions would have cost considerably more, and the program would not have found out until Gate 3 at the earliest.

Recorded as issue I-01. The remediation is traceable through conditions GC-04 (re-validated volumes) and the corrected capital charge carried into the revised case.

15. Document Control

VersionDateChange
1.005 Feb 2026Issued at Gate 0. Two-tier test, scale and bands established; Gate 0 and Gate 1 criteria fixed.
1.130 Apr 2026Evidence standard (§4) tightened following the Gate 1 recycle; the U value formalised on the scale after being applied ad hoc at that gate.
2.011 Jun 2026Reissued at Gate 1. Criteria and weights fixed for Gates 2, 3 and 4 (§§9–11). Gate 5 defined as a non-authorising accountability gate. Criteria-locking provision added at §6.

Owner: C. Tyrrell, NPD Program Manager and Chair of the Gate Review Board. Approving authority: the Gate Review Board. Related: Stage-Gate Governance Model & Gate Review Board Charter, Gate Conditions Register, Gate 1 Business Case Package, Gate 1 Recycle Memorandum.