A stage-gate new product development program at a US life & annuity carrier: the design, filing and launch of Beacon Index Advantage, a fixed indexed annuity with an optional guaranteed lifetime withdrawal benefit rider. Five gates, a cross-functional Gate Review Board voting go / cancel / recycle / hold, funding released one stage at a time, and a state insurance department filing route through the Interstate Insurance Product Regulation Compact. Fictional company, product and people; the methodology, regulatory route and governance mechanics are real.
The Program Story
Six documents record decisions taken over three years, and each is written to stand alone — so none of them tells you the sequence. This does: a predecessor cancelled, a successor re-screened from the first gate, a gate that was recycled, a gate that carried with strings, and three gates still ahead. Seven chapters, plus who decides, how to read the numbers, and three reading routes depending on how long you have.
Governance & Decision Authority
Stage-Gate Governance Model & Gate Review Board Charter
The governing instrument. Board mandate and its explicit limits, five voting seats plus a non-voting chair and CRO observer, quorum and conflict rules, the five gate outcomes defined, the recycle and cancellation protocols, published cancellation thresholds for the gates ahead, tranche-based funding release, and why this program publishes no full-program earned value baseline. 19 sections, 5 parts.
Gate Decision Framework
The assessment standard behind every gate. Must-meet knockout criteria versus weighted should-meet scoring, the evidence standard, the U (unscoreable) rating and why it exists, score bands mapped to the five outcomes, criteria fixed in advance at the prior gate, full criteria for all six gates, and the scored record — including why the Gate 1 recycle was not a low score but an unscoreable one. 15 sections, 3 parts.
Prior Concept Cancellation Record
A program this same board cancelled. The Harborline Buffer Series — a registered index-linked annuity — passed Gate 0 in March 2025 and was cancelled at Gate 1 that August, carried 1–4. Grounds, the vote including the dissent, all seven cancellation protocol steps executed, final accounting with the unspent tranche returned to capital, and the auditable test for why the successor is a new program and not a zombie restart. 9 sections.
Gate 0 Concept Screening Package
Where the program started. A screening gate’s job is to be cheap and to be wrong cheaply — so the operative section is not the recommendation but the six questions Stage 1 was funded to answer, each pre-committed as a Gate 1 must-meet. Preliminary economics are scored 3 and explicitly barred from being carried forward. A point-in-time record: two of its assumptions did not survive Gate 1, and it is not retrofitted to pretend otherwise. 11 sections.
Gate 1 Business Case Package
The decision package as it went to the Board at the second convening. Product structure and market, five-year volume forecast, capital strain and return against hurdle, written channel commitments and the premium gap they leave open, a downside case that fails the hurdle, the regulatory route, and the two dissents tabled against the recommendation. Records what changed after the recycle — including a 120 basis point fall in projected return. 19 sections, 5 parts.
Gate 1 Recycle Memorandum
The Chair’s record of the gate that did not carry, issued two business days after it. The two unscoreable findings, an explicit list of what the Board did not find, a root cause the Chair attributes to himself, the itemized $232,000 cost of the loop, and the schedule consequence — the launch date did not move, so development was compressed by six weeks against a fixed end date. 11 sections.
Gate 1 Recycle — Executive Committee Session
The session held two business days after the gate did not carry — a 20-slide deck to the Executive Committee with the full mathematics and five options analyzed with pros and cons, including CANCEL. Opens by correcting its own premise: Gate 1 did not fail on the numbers, it was never scored — two criteria came back U. Quantifies the 87 bp capital-charge error, bridges the return 14.6% → 13.4%, and lands on the central finding: every lever that buys back return spends sales volume — the one variable the model proves actually moves the answer.
Gate 1 Recycle Deck
A crisp, 10-slide house-format read of the same recycle decision as the executive session above — the five gate outcomes, the two correctable flaws (the 87 bp capital-charge understatement and unvalidated distribution volume), the corrected economics side by side (14.6% → 13.4% IRR, still above the 11% hurdle), why RECYCLE rather than GO, CANCEL, or conditions, and the schedule impact (launch held, Stage 2 compressed 46→40 weeks). 10 slides.
Gate Conditions Register
The live GC-01 to GC-05 record between gates — three closed, two open, one flagged at risk. Each condition carries its owner, the verification method as recorded at issue, the evidence, and a post-closure efficacy finding: did closing it change anything? One did not. GC-04 was correctly specified, verified and closed, and left $17,000,000 of Year 1 premium uncommitted. 10 sections.
Change Control Log
The register of scope, cost, schedule and vendor change between gates — distinct from the gate conditions, which govern gate decisions. Eight requests: five approved, two declined, one deferred. Change authority is split by threshold, and anything needing an unauthorized tranche cannot be decided out of cycle at all — the gate is the funding authorization. 8 sections.
Change Request Detail
The full form behind all eight requests, each with impact across scope, schedule, cost, quality and risk, the alternatives weighed, and — the field that does the work — what happens if the request is refused. That column is where the board found its reasoning for declining CR-005 on pricing discipline and CR-007 on the ground that money cannot accelerate somebody else’s credit committee.
Gate 2 Readiness Assessment
A live, rolling assessment — not a scored gate package — tracking trajectory toward Gate 2 against the five must-meet and five should-meet criteria fixed at Gate 1. Reads the two still-open Gate Conditions directly into the must-meet list, flags GC-05’s ISDA counterparty gap as the one real open question, and states plainly what has to be true by April 1, 2027 to keep GO and GO WITH CONDITIONS on the table. 9 sections.
Gate 2 Review Deck
The package the Board will receive at Gate 2 — drafted 167 days early with its evidence gaps left visible. Assembling a gate package early is a diagnostic, not just preparation: one built the week before can only report what happened; one built five months early reports what will be missing if nothing changes. 0 of 5 must-meets are fully evidenced and MM-4 is failing. Carries the mandatory case for CANCEL — the Chair doesn’t vote, so the obligation is to put the strongest argument against proceeding in front of those who do — plus a blank scoring sheet and dissent recorded ahead of the room.
Gate 5 Post-Launch Review
The only forward-dated document in this suite, labeled an illustrative end-state sample and deliberately not reconciled against the mid-flight artifacts — they describe October 2026, this describes September 2028. Its finding: Gate 5 can read four of six benefits and only partially, and the two it cannot read at all are the two most likely to move the return. Reports a −12.1% premium shortfall that decomposes exactly into policy count × case size, and reads the one measure that beat plan — rider election, +6 pts — as bad news: more guarantee exposure than was priced.
Product, Regulatory & Program Artifacts
Stage-Gate Methodology Guide
An educational reference, not a gate-specific decision document: what a gate is, the two-tier must-meet/should-meet test, the Robert G. Cooper lineage behind stage-gate and the PMI layer underneath it, the IIPRC/SERFF/NAIC #275 regulatory grounding behind a fixed indexed annuity, why the program publishes no full-program earned value baseline, and how this suite's premise — a program that can be voted out of existence — differs from the other five suites in this portfolio. 10 sections.
Program Charter
The authorizing instrument, chartered at the Gate 1 second convening: business case summary, product vision, scope with explicit exclusions (New York, SEC/FINRA, a full EVM baseline), the governance structure and Chair’s limits, the 90-person roster, the $27,904,000 tranche- funded budget ceiling, RAIDD summary, and a dedicated section on what a stage-gate charter deliberately does not authorize up front. 22 sections.
Program Kickoff Deck
The program-launch deck presented at Gate 0 — orienting the full team on the product, the stage-gate governance (the five gate outcomes, the Board’s five votes and the Chair’s no vote), the gate roadmap to a March 2028 launch, and the gate-authorized, tranche-funded model. A point-in-time Gate 0 artifact: the economics shown are preliminary, later validated and refined through the gate process. 19 slides.
Steering Committee Deck
The Stage 2 steering committee review (Oct 16, 2026, week 17 of 40) — reading Amber on hedging readiness and distribution while the recycle-corrected economics hold above hurdle. Program-health/RAG snapshot, scope & change log, the aged risk register (risks moving both ways — hedging elevated 15→20, illustration reduced 16→8), the five gate conditions, the issues log with the recycle shown resolved, and the budget forecast within the $27,904,000 envelope. 20 slides.
Project Management Plan
The PMBOK backbone document — how the program is managed, not what it builds (that is the WBS). 21 numbered sections across every knowledge area, adapted to stage-gate throughout: gate-anchored schedule, tranche-released cost, the Gate Review Board as decision authority. A baseline — archived intact at closeout as the conformance reference, not rewritten as execution diverges. Pairs with the resource-loaded WBS.
Program Budget
The financial companion to the Charter: labor derived from all 15 team envelopes, non-labor across 8 categories, the four-tranche release schedule tied to each gate, contingency draws and remaining reserve, spend to date, and why no full-program earned value baseline is published. Every dollar figure in this suite traces back to this document's derivation. 11 sections.
Resource Plan
The full named 90-person roster behind the Program Budget's labor line, grouped by all 15 functional teams, with FTE-vs-Consultant mix, blended rate, and each person's stage engagement window. Every team's rows reproduce its budget envelope exactly, in both hours and cost — allocation is derived from hours over each engagement window, never stated independently.
RAIDD Log
All 12 risks, 8 assumptions, 6 issues, 7 dependencies and 11 decisions — every ID assigned up front in the fact pack before the first artifact was written — plus a status summary of the 5 Gate Conditions, which carry their own full register. The index into every live thread this suite tracks, cross-linked rather than restated.
Communications Management Plan
How the program communicates around five decisions, not one delivery date. Stakeholder information needs mapped to the Gate Review Board, the five-phase gate-communications sequence, the mandatory assumption-breach notifications that make ‘no surprises upward’ enforceable, and the Gate 1 recycle documented as the working crisis-communications case — disclosed on the program’s own initiative, quantified without softening, corrected everywhere. 11 sections.
Weekly Program Status Report
The snapshot as of the status date — read against the next gate, not a generic percent-complete. Stage 2 at week 17 of 40, funding released versus spent, the three closed and two open Gate 1 conditions with one flagged at risk, upcoming milestones, and the one decision actually due. Carries no SPI/CPI: the program funds no full-program earned-value baseline, and the report says so rather than inventing one. 6 sections.
Glossary
Every stage-gate, annuity-product and regulatory term used across this suite, defined in plain language across four categories: methodology (gates, tranches, the U rating), product (FIA, RILA, GLWB, crediting strategies), regulatory (IIPRC, SERFF, NAIC #275), and governance — with cross-suite analogies where one genuinely helps.
Org Chart
The full reporting structure: the six-seat Gate Review Board at the top, all 90 people across 15 functional teams below, each reporting functionally to its own Board seat and matrixed to the Program Manager for coordination only — not supervision (Decision D-11). A Reporting-Line Registry maps every team to its lead and functional owner. Read alongside the Resource Plan, which carries full per-person role, hours and cost detail.
Cost-Benefit Analysis
The analytical backup to the Gate 1 Business Case, maintained independently of any single gate: the five-year capital strain schedule (live formulas, reconciled to the premium forecast), the distinction between program build cost and statutory capital strain, the three-point IRR scenario record including the downside case that fails the hurdle, and the risks most likely to move it before Gate 2.
Benefits Realization Plan
Built as an early-warning instrument, not a value claim — because this is the only suite whose business case can fail, and whose failure is detectable years before it can be proven. Six benefits ordered by when they become readable, volume trigger thresholds computed from the business-case model (breakeven at −23%, of which 40% is already consumed pre-launch), and an honest account of the governance cliff: Gate 5 sits 185 days after launch, and Years 2–5 of a five-year case have no governance body at all.
Program Dashboard
A stage-gate dashboard reports on a decision, not a delivery — so it leads with what it deliberately does not show. No program percent-complete, no CPI, no SPI, because Stages 3–4 are unfunded and there is no baseline to be on or off. What it does show: the one legitimate variance inside the released tranche (Stage 2 running 5% behind elapsed time, reported without a verdict, because under-burn is ambiguous), the Gate 2 must-meet trajectory, condition status, top open risks by score, and a section on what would make this dashboard lie.
Requirements Traceability Matrix
18 requirements traced to a gate decision at both ends — the gate that authorized each one, and the gate that will verify it. That is a stronger claim than “it has a test case,” because at a gate the question is not was this tested but is the Board entitled to rely on this being done. Critically, the two crediting strategies cut at Gate 1 are kept as withdrawn rows, not deletions — an RTM that deletes cut scope cannot answer the most common question asked of a launched product: why doesn’t it do X?
Vendor SOW — Cordelane Systems
The question no other suite’s vendor paperwork has to answer: how do you contract a vendor for a program that might be cancelled? At signature only Stage 2 is funded, so the SOW is itself gated — three work packages priced in full but each authorized by a release notice that issues only on a gate outcome. $1,680,000 authorized, $1,170,000 scoped and unfunded. Carries termination for convenience at stage boundaries with a defined wind-down, because a vendor contract that couldn’t follow the program’s own governance would become the reason not to cancel.
RACI Matrix
Confronts a real problem rather than papering over it: standard RACI has no notation for a vote. Five executives decide gate outcomes — marking them C misdescribes them (consulted implies input, voting implies decision) and A is worse, since RACI expects one A per row and this program has five voters plus a chair answerable for the evidence with no vote at all. So the notation is extended by V (binding vote) and F (facilitate, no vote), and accountability splits: the Chair owns the evidence being sound, the Board owns the decision being right. 20 activities; the Chair is Accountable on 8 and votes on none.
Resource-Loaded WBS
You cannot resource-load beyond the funded horizon. Stages 1–2 are decomposed, loaded from the 90-person roster and costed; Stages 3–4 are named as a planning envelope and deliberately not loaded, because decomposing them would present a staffing commitment against $11,780,000 the Board hasn’t released. The number that falls out: only 36% of program hours can be baselined at all — an accurate statement of how much has been authorized, not a gap in the planning. Ships as Excel and MS Project XML; Stages 3–4 import as unstaffed summary tasks.
Program WBS Console
The plan as an interactive Gantt — 29 line items across the four stages, each with its accountable lead. Exports to MS Project.
Stage Budget Rollup Console
The suite’s first interactive artifact. In a stage-gate program “budget” is not one number — it is four states: authorized $27,904,000 · released $14,232,000 · spent $6,950,000 · unreleased $13,672,000 (49%). A conventional console reporting “25% consumed” would be arithmetically true and materially misleading: it implies three quarters of the budget is there to spend, when only $7,282,000 actually is. Toggle between stage and category views, and between including and excluding money the gates haven’t released — the two pictures differ sharply, and a reader is better served seeing that than being told it.
Regulatory Filing Plan
The one plan the program cannot compress. Every other schedule risk can be worked — a regulator’s review clock cannot (DEP-01), so the program commits to submission quality and nothing else, and treats approval dates as forecasts belonging to somebody else. 46 states via three routes: one IIPRC Compact filing covering 44, separate SERFF filings for California and Florida, New York excluded. Its real job is forcing a decision nobody has taken: what counts as enough states to launch? — with the Compact-only fallback called out as the option that looks safe and isn’t.
Product Requirements Document
A PRD that can change between gates is a PRD nobody can build against. Versioned by gate, not by sprint — frozen between them, so teams build against a set that cannot move under them and anyone wanting it to move must take that to a Board with authority to say no. Carries the requirements the program has no BRD to hold (a BRD would sit between the business case and this, duplicating both). The requirement set is imported from the traceability matrix, so the two cannot disagree about what the product is. Version 1.0 was tabled and never adopted — kept in the history rather than renumbered away.
Actuarial Pricing & Assumption Summary
This is the document that was wrong. The Gate 1 recycle turned on an 87 bp understatement of the capital charge, and the misstatement lived here — a confident, internally consistent, peer-unreviewed number that moved the reported return 120 bp and survived internal review because required capital compounds across every projection year rather than showing up as one wrong line. So this version is organized around the question the Board asked afterwards: an assumption summary’s job is not to state assumptions, it is to state how each one could be wrong and what would reveal it. All 8 carry a failure mode and a detection point; 2 cannot be tested before the gate that depends on them.
Master Test & Validation Strategy
You write the test strategy for the gate, not for the testers. Testing is Stage 3 work and Stage 3 isn’t funded — so this document’s first reader isn’t a QA lead planning execution, it’s a Board deciding whether the approach is good enough to release $7,450,000 against. Six levels, each naming the gate it feeds and the requirements it proves. Illustration validation is a compliance test wearing a functional test’s clothing: output reconciles to the filed methodology, not the spec — and where they disagree, the filing wins, so a whole class of defect can’t be fixed by the program alone without re-filing.
Launch Readiness Criteria
A launch readiness document’s real job is to define the conditions under which you do NOT launch — everything else is decoration. Nine criteria written as no-go conditions with named deciders and pre-committed responses, adopted at Gate 4 rather than assembled in the final weeks “when every answer is shaped by how much it would cost to say no.” Says the uncomfortable thing plainly: this launch date has never moved, and an organization that has always protected the date will protect it again. §4 admits the most consequential criterion names a threshold nobody has agreed yet; §5 lists three things deliberately excluded, because a list that includes everything important stops discriminating.
Vendor Management Plan
A gated program cannot contract a vendor for scope a future gate may decline to fund — so every multi-stage engagement is decomposed into gate-authorized work packages, released one gate at a time (the Cordelane SOW is the worked example: WP-1 Stage 2 build is released; WP-2 and WP-3 are not). Carries the vendor inventory and criticality, contract structure and termination for convenience, financial control, and the realized vendor risk: Brightpath's illustration engine could not support two of five crediting strategies (I-02) — a vendor platform limit that reshaped product scope, cutting the launch strategies from five to three. 10 sections.
Distribution & Advisor Enablement Plan
Half of this plan governs other people’s behavior and half governs the company’s own throughput — and only one half is enforceable. Distribution rests on IMOs and wholesalers who have signed nothing binding; enablement rests on appointment and certification pipelines the company controls outright. The distinction that organizes the document: carrier appointment is schedule-free and can be front-loaded now, but product certification cannot begin until the forms are approved — so one half inherits every day of the filing plan’s slippage and the other does not. Carries the tiered commitment position ($168,000,000 written against a $185,000,000 Year One target, leaving $17,000,000 uncommitted after GC-04 closed — closed is not resolved), the coverage arithmetic on four wholesalers hired against six planned (67% of the field carrying 100% of the volume assumption), the $1,540,000 enablement budget Stage 4 has not yet released, and six leading indicators that move before premium does. 12 sections.
Operations Readiness & Cutover Plan
A release can be rolled back and an in-force contract cannot — so the point of no return is not go-live, it is first policy issued. That moment is an event, not a date: it lands whenever a producer somewhere submits an application that clears suitability review, which makes it the one boundary in the program that cannot be put on a schedule. Rollback here doesn’t fail, it expires, and the plan is built in two halves that obey different rules on either side of it. Carries three findings, none of them flattering: the vendor contracted to run hypercare (Cordelane WP-3, $410,000) isn’t under contract for it until Stage 4 funding releases at Gate 4 — 11 days before launch, which makes Gate 4 a launch-or-defer decision wearing a readiness assessment’s clothing; suitability review is a licensed judgment task staffed by one analyst against a Year One volume nobody sized it for, and it degrades invisibly rather than failing; and the Severity-2 workaround rule (“documented, staffed and accepted”) quietly books unbudgeted manual work to a 7-person operations team every time testing accepts one. 12 sections.
The Program Story
The whole program read end to end — five gates, one recycle, one cancellation decision that was not taken. Written as narrative rather than as governance record, for a reader who wants to know what happened before they audit how it was decided. Every figure in it reconciles to the artifacts above.
Roles & Responsibilities
Two people sit on the Gate Review Board without votes, for opposite reasons. Who owns what across thirteen teams and six board seats — why the Chair facilitates and does not vote (D-11), why the Chief Risk Officer observes but may record a dissent, and who holds the two different ways of saying no.
Why this program is shown mid-flight — on purpose
The other suites in this portfolio complete. This one is documented at Stage 2 of 4, with three gates still ahead, and that is a deliberate choice rather than an unfinished record. The artifact set itself is finished — every document listed below is built and downloadable. What remains open is the program: conditions outstanding, funding unreleased, gates unmet.
A completed stage-gate program is indistinguishable from waterfall. 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 one budget. The governance is invisible precisely because it worked. It can only be shown while the decisions are still open — while conditions are outstanding, while money is unreleased, while a gate has not yet met.
The suite still shows how a program closes: the Gate 5 Post-Launch Review is carried as an explicitly labeled illustrative end-state, and the Prior Concept Cancellation Record documents a predecessor concept this same board cancelled at its Gate 1 in August 2025. See How to read this portfolio.