← Q3 journey review packs

Journey 2 council assessment: Upgrade winner, pre-travel and travel day

Verdict: CONDITIONAL

The strategy is sound and correctly protective of the one asset that matters: the estate's only proven-incremental send (the GRW-16 day-before family, +16.9%/+16.3% confirmed) is guarded by a non-inferiority A/B in which the current creative stays live as the control, replace only on win or tie, and no other beat may borrow the figure. The disruption spine is right (the entire arc yields to L0/OPS-DR; zero commercial in disrupted contexts per the 20 Jul Darwin rulings; the seat-swap sell lives only in non-disrupted contexts). The pack could not stand as authored for one structural reason: it cited decision ids from the superseded baseline numbering, which, after the 20 Jul consolidation, point at the WRONG items in the consolidated log (its "DL-018" is now the holdout-structure item, its "DL-028" a hub-relabel item). Every citation is now aligned, and three cross-pack facts that moved since authoring are corrected in the pack.

Strongest points: the trip-anchored arc with one consent-free state surface as the floor for the 18.0% non-contactable (127,649/yr, confirmed); the honest sizing caveat that 85% of upgrades land within 24h of departure, which right-sizes GRW-18 to a tail beat and stops the mid-window read being oversold; suppression-as-treatment for disruption, refund and opt-out, stated as designed experiences; and measurement honesty (everything except the GRW-16 family is attributed, not incremental; no model triggers the journey and the pack says so).

What the council changed in the pack (applied, rebuilt, validated)

1. Decision-id citations aligned to the consolidated log. What: every old-baseline id remapped (DL-010→DL-045, DL-018→DL-047, DL-019→DL-053, DL-021→DL-054, DL-028→DL-055, DL-030→DL-056, DL-031→DL-057) and every local J2-nn reference replaced with its consolidated DL id across all tabs and strategy.md. Why: after the 20 Jul consolidation the old ids resolve to unrelated items; a builder following "DL-028" would land on a hub-relabel decision instead of the trip-state surface call. Customers: none directly; every team reading the pack. Where: all six tabs, traces, related documents, strategy.md. Requires: nothing new. Improves: the pack and the Decisions and Approvals Log now reference one id space; disputes trace to one record. Action: none; done.
2. The yield-order and F3-clamp provenance re-based to verified-ruled (DL-047). What: replaced "provenance conflict, one record must win" with the verified position: tobe-composition carries the RULED labels (yield order RULED Sheran 5 Jul, Q12 ratified 7 Jul; F3 clamp RULED Sheran 6 Jul; checked in the file), the baseline compile held it open, and the pack now builds on the ruled order as its working assumption; the pack's earlier "6 Jul" date is corrected. Why: DL-047 already verified the labels; presenting a settled ruling as a live either-way dispute invites re-litigation at build time. Customers: every winner on the travel morning (the order decides which push survives the cap). Where: TO-BE map rules, Reference contact rules, matrix GRW-17 row, Process step 0, strategy §3/§6. Requires: Sheran's one-line confirm-and-close. Improves: build sequencing can proceed without a dispute window. Action: Sheran confirms DL-047 in one line.
3. The extraction ask re-framed as the paired ratification (DL-005). What: the pack's headline ask now reads "ratify J5x as ONE paired ruling with Journey 1's J4c consolidation" everywhere the ask appears (masthead, brief status, document status, Process step 0, footer, strategy). Why: the consolidated log merged the two restructures into one paired decision because ratifying one without the other leaves the legacy win walks half-dismembered; the pack still asked for J5x alone. Customers: all 710,616 departures/yr plus Journey 1's entry cohorts. Where: five slots + strategy. Requires: nothing new. Improves: one ruling closes both packs' existence questions. Action: Sheran ratifies DL-005.
4. The post-win slot closure mirrored with Journey 1 (DL-052). What: the pack now states that the J1 pack carries the same rule (pre-travel owns the slot per rule 6 L1; J1 rebook floors at travel-completion + 1 day) as its working assumption, so one ruling closes both. Why: the J1 council applied the rule on 20 Jul; without the mirror note the two packs read as independently proposing the same thing. Customers: the ~85% short-lead majority whose post-win and pre-travel windows are the same days. Where: Process step 0, strategy §5. Requires: rule 6 layer model (ratified) + journey_state. Improves: cap arbitration by design, not accident. Action: Sheran ratifies DL-052.
5. Fourteen-field completeness index added (Reference tab). All fourteen personalisation and contact fields stated with where each is answered; the two thinnest flagged honestly: field 8 (customer history: only the register model is used; first-time vs repeat winner proposed as a phase-2 overlay) and field 12 (adaptation: the arc adapts to the trip, not to opens, deliberately).
6. The GRW-16 consent forks made buildable. Dual-consented users keep the proven email leg alongside the new push (the proven family is never switched off on an assumption; the one interaction design reads the overlap); not-email-opted carries the working assumption suppress, never re-class as transactional (DL-045, Lisa rules with the DL-034 precedent); the T-2h not-opened recovery is deleted from the TO-BE design as an adopted working assumption (DL-058).
7. Cross-pack facts refreshed. The settled Live Activity strategy (15 to 20 Jul: iOS experience surface, API-started via webhook, replaces the push, never stacks; the orchestration-home decision settles once for J2+J4+J8, and DL-057's re-scope rides the cross-journey shared build); OPS-DR's unauthored disruption comm now cited as the cross-journey review's top experience gap; volume standardisation on 710,616/yr labelled as the adopted working assumption (DL-059) with Airtable row updates to follow on Sheran's go; monitoring/alerting on the time-critical morning sends added to the CRM build actions (~1,947 departures/day, confirmed).
8. Plain-English pass. PECR expanded at first use; NBA spelt out; L0/L2/Class B defined in plain English at the matrix and in the footer key (JTBD, LA, TOC, BUN were already covered); accessibility requirements kept in the design brief's format rules. No em-dashes anywhere (validated).

Working assumptions adopted (closed rather than escalated)

• DL-058 closed: the gated T-2h not-opened recovery is deleted from the TO-BE design. It was never approved to build, collides with GRW-17 and M33_P to the minute, and its job is fully covered by the T-3h wayfinding push plus the persistent state; striking a never-approved draft touch needs no formal authority. Gate-1 acceptance of the pack confirms it.
• DL-059 closed: 710,616/yr confirmed (auction 526,025 / BUN 184,591) is the single volume everywhere in the pack. The residue is execution, not decision: Airtable matrix-row updates on Sheran's go, and Analytics' per-beat derivation moves to the validation plan (AN-6 follow-through).
• Yield order + F3 clamp (inside DL-047): treated as ruled (verified labels) pending Sheran's one-line confirm; the pack builds on the ruled order.
• DL-045 (not closed, assumption carried): email leg suppressed for not-email-opted users, never re-classed as transactional; stays open for Lisa's ruling under the DL-034 precedent.

What genuinely requires approval

Sheran: DL-005 (paired J1+J2 ratification, the pack's headline ask) · DL-046 (headline test = the GRW-16 reframe A/B; the interaction design commission rides it) · DL-052 (post-win slot) · DL-047 (one-line confirm-and-close) · DL-058 (confirm the deletion at Gate-1) · DL-050 (seat-swap boundary written into the spec) · DL-051 (GRW-TD single-job scope; it strikes an opt-in line from a Gate-1 card) · DL-049 (Darwin posture sign-off). Build hard gates: DL-045 + the PECR read (Lisa + compliance), DL-048 (GRW-18 sizing go/no-go, Ivi + Sheran), DL-054 (rollback floor X + M31 offset before the A/B wires), DL-055 (trip-state surface, Product), DL-056 (TrainSwap eligibility defect, gates only the non-disrupted sell path). Adjacent: DL-016 (Darwin alignment session, Jon).

Evidence gaps (feeding the validation plan)

1. The GRW-18 T-7+ lead-time slice is unsized (DL-048), the journey's top gap: the Sheran-ruled holdout cannot promise a Q3 significance read until Analytics sizes the slice; if underpowered, pre-agreed leading indicators or a wider window (T-5 to T-2), never a quietly unreadable test.
2. Anxiety-ticket deflection baselines per moment (pre-travel 1,753 / on-train 4,776) do not exist; without them the journey's primary KPI is unreadable, and the open question whether the existing state alone deflects most of it stays unanswerable.
3. The GRW-TD saved-£ anchor (~£38 walk-up) is unsubstantiated and the at-fire-time disrupted-trip flag is undefined (Analytics + Product/Eng); both gate the modal's figure line, not the journey.
4. The interim platform/coach payload source and freshness pre-Darwin (Eng); a wrong platform number is the exact failure GRW-17 exists to prevent.
5. Refund/disruption volumes do not reconcile (4,836/4mo vs 17,416/12mo vs ~12,359), TBC: Analytics/CS; the rollback floor X and the M31 offset validation (DL-054) must land before the headline A/B wires.
6. Consultations not yet held, stated per the standard: Product on the trip-state surface and modal; CS on ticket tagging so the deflection baselines stay readable after launch.