← Q3 journey review packs

Journey 4 council assessment: Live bidder, in-auction and loss recovery

Verdict: CONDITIONAL

The journey's core logic is the strongest honesty case in the portfolio and the council did not weaken it: the auction owns the customer while it runs (state S2, layer L1), the close moment tells the truth per close reason with nothing celebratory and nothing salesy, and recovery touches only the 7-day non-returners, once, capped at 3 per auction. The evidence discipline holds: the only causal reads in this territory are negative (-11.8% blanket recovery; -3.73% M21_E, both confirmed), and the pack treats them as the design constraint rather than an inconvenience, including a declared, evidence-led exception to the propagated second-knock ladder. The conditions are real: the D+7 send's economics rest on a return definition Analytics has contradicted (74-79% RW vs 37.1/41.5/57.8% by tier on the Analytics event set, both confirmed on their own definitions), so DL-069 stays ship-blocking and TBC-flagged, deliberately unresolved by the council; and the close-event contract (DL-001) gates the entire outcome fork.

Strongest points: the honesty-at-loss bar enforced mechanically (the 8-branch box never says "you lost" on an auction that never ran; the payment-reassurance clause is suppressed until the payment-hold data substantiates it, a copy gate not a launch gate); the one headline test (the D+7 gate holdout, DL-074) carrying the whole GBP 92.70 LTV-gap claim with guardrails set before launch; and the Live Activity treated as contact REMOVAL (status pushes downgrade where it is active) rather than another channel.

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

1. Decision ids realigned to the consolidated log, everywhere. What: every citation now points at the consolidated Decisions and Approvals Log (J4-D01 to D11 became DL-069 to DL-078, with J4-D05 folded into DL-001 and J4-D04 into DL-004; stale DL-007/DL-014/DL-015/DL-016/DL-017 and the old token-holder DL-003 became DL-001/DL-003/DL-012/DL-075/DL-076/DL-033). Why: the log renumbered at consolidation and a pack citing dead ids sends reviewers to the wrong decisions. Customers: none directly; the review group. Where: all five tabs, the TO-BE map pins and legend, strategy.md throughout. Requires: nothing. Improves: one source of truth per decision. Action: none; housekeeping. (The map half of this edit is the one that failed mid-pass; rerun and verified, zero stale ids remain.)
2. LA push-downgrade list enumerated, closing DL-071 as a working assumption. What: where the Live Activity is active, outbid-status and time-left pushes downgrade to critical-only (the LA carries both states); the knocked-out push (action-required) and the L0 close notification (transactional) stay at full delivery. Why: the ratified cap (12 Jul) said "status pushes downgrade" without naming them; an unenumerated rule cannot be built, and unbuilt, the LA adds contact instead of removing it. Customers: every LA-enabled in-auction bidder. Where: logic tab arbitration rules, brief tab channel rows, strategy M2. Requires: ACT-LA build + DL-033 token-holder answer. Improves: the LA's de-noising case becomes buildable. Action: Gate-1 sign-off on the pack ratifies the list.
3. Day-14 episode close logged as its own new decision (J4-D12). What: the episode close at day 14, replacing the 72h timeout Journey 1's architecture assumed, is now a logged decision with the J5/lifecycle exit split, not a buried TBC. Why: it moves a Journey-1 boundary, which is exactly the class of trade-off the log exists for. Customers: every losing bidder's exit path. Where: brief exits row, logic tab episode rules, decisions-delta.json (new). Requires: journey_state. Improves: J1 and J4 cannot encode contradictory boundaries unnoticed. Action: Sheran ratifies at Gate-1, Lisa confirms canvas mechanics.
4. The J5 hand-off contract mirrored with the flags named (DL-073). What: first-time losers exit at day-14 with first_loss, close_reason and route flags into Journey 5's pool; the flag enumeration J5's pack already carries now appears on J4's exhaust node, Reference boundaries and strategy hand-offs. Why: J5 was authored and council-reviewed after this pack was drafted; both sides must name the same payload or the canvases guess. Customers: the 81%-quit first-time losers, the journey's biggest at-risk cohort. Where: TO-BE map exhaust node, Reference tab, strategy sections 3 and 5. Requires: exhaust flags from the J4 canvas. Improves: the bad-luck-not-rejection reframe actually receives the data it personalises on. Action: DL-073 ratification (cross-journey, with J5).
5. The J3 boundary stated honestly on the pack's face (DL-003 + DL-060). What: the pack now says the bid-placed boundary is agreed by all records, the DL-003 confirm (the 6 Jul ruling reopened in error) and J3's NEW no-bid fold proposal (DL-060) ride in one Sheran ruling, and the bidder slice belongs to J4 either way. Why: cross-journey review item 5: the highest-attributed canvas in the estate (GBP 4.99M + 457k attributed, not incremental) sat under three records in three states; presenting J4's record as settled would have misled the review group. Customers: pre-close abandoners and post-close losers on the M22_P/M24_P path. Where: brief existing-asset row, logic tab boundaries, strategy section 5. Requires: nothing new. Improves: one decision surface at Gate-1. Action: Sheran rules DL-003 + DL-060 together, after Lisa's M22_P entry-criteria audit.
6. Two P3s closed as working assumptions rather than escalated (DL-077, DL-078). What: Android Live Updates parked at Phase 3 (push-only this release, gated on the iOS LA read); the 159,053 losing-bidders/yr figure treated as confirmed against the Analytics sizing tracker, closing roster decision (f). Why: escalation discipline: sequencing with no customer risk and label housekeeping on a figure Analytics stood behind are not executive judgements. Customers: Android bidders (unchanged journey); the review group (correct priority signal). Where: brief and logic tabs, strategy sections 1 and 3, decisions-delta.json (closed). Requires: nothing. Improves: the log carries only genuine authority questions. Action: hub-row edit (Sheran, Ivi cc'd) remains as housekeeping.
7. Fourteen-field contact-completeness index added. What: the Reference tab now carries the standard's fourteen personalisation and contact fields with where each is decided in the pack. All fourteen were present in substance; field 12 (adaptation to response) was the least explicit and is now stated (silent exit on organic return; no second knock by evidence-led exception). Why: the standard requires every journey to state all fourteen. Where: Reference tab. Improves: nothing rests on an unstated assumption.
8. Plain-English pass completed. What: PECR (the Privacy and Electronic Communications Regulations), BUN (Buy It Now, the fixed-price upgrade), RFM (recency / frequency / monetary value), MDE (minimum detectable effect), L0 (the always-sent transactional class) and IW (instant win) expanded at first use on the tabs that use them; no em-dashes anywhere (validated); sections lead with conclusions. Why: writing standard. Where: AS-IS, logic and brief tabs, strategy section 6.

Working assumptions adopted (closed rather than escalated)

• DL-071 closed: the LA downgrade list above; Gate-1 sign-off ratifies.
• DL-077 closed: Android push-only this release; Phase-3 line opens on the iOS LA read plus platform availability; no moment ever depends on LA on either platform.
• DL-078 closed: the 159,053 figure is confirmed per the sizing tracker; hub edit is housekeeping.
• Held, not closed: DL-069 stays open and ship-blocking by design. The council checked that the 74-79% vs 37.1/41.5/57.8% clash is TBC-flagged wherever either figure appears (it is, on the logic and brief tabs and in strategy D1) and adopted neither number.

What genuinely requires approval

Sheran: DL-003 + DL-060 (one ruling on the M22_P/M24_P confirm and the no-bid fold) · DL-004 (recovery owner of record, three packs concur) · DL-069 definition ratification after the Analytics re-statement · DL-070 (AN-2 send fold) · DL-072 (ACT-23 pre-agreed fold rule) · DL-073 (J5 hand-off) · DL-074 (headline test) · DL-075/DL-076 (loss-close estate verdicts and the M21_E rework, with the C1 ordering constraint: retirements land BEFORE the new journey ships) · J4-D12 (day-14 close). Build hard gates: DL-001 (close-event contract, Eng/PMM), DL-011/BRZ-83 (the live suppression never ships on the interim RW list), DL-012 (batch exclusions), DL-033 (LA token holder, Lisa) and the PECR class of the stripped L0 (Lisa + compliance, inheriting the DL-034 precedent).

Evidence gaps (feeding the validation plan)

1. The pinned returned_within_7d definition and the re-stated organic-return rate (DL-069), the journey's top gap: send volume, holdout power and expected GBP all move ~2.5x with it; the v1 tracker flagged the first-time rate as unreliable (n=4), so the re-statement must give the first-timer variant its own read. Owner: Data/Eng pin, Analytics (Ivi) re-state.
2. Outbid volume and re-bid headroom; knocked_out volume (the ACT-23 go/no-go with the DL-072 threshold); iOS share of active bidders (bounds the LA build value). Owner: Analytics.
3. Whether the M21_E rework holdout can be powered at ~45 sends/wk, or which leading indicators replace a causal read. Owner: Analytics, stated up front.
4. The payment-hold lifecycle (gap ⑦): a copy gate, not a launch gate; every reassurance clause stays suppressed until it lands. Owner: Eng.
5. Whether T2_E and T30_E both fire on one close event (decides the one-comm-or-two collapse, DL-075). Owner: Eng via Lisa's Braze audit.
6. Suze's Avanti WC knocked-out screen retrieved and confirmed current (the ACT-23 Gate-2 starting point). Owner: Design.