← Q3 journey review packs

Journey 3 council assessment · Non-transacted bidder: funnel abandon

Verdict

APPROVE AT GATE-1, WITH CONDITIONS. The journey's evidence-to-choice-to-experience chain holds: the ~72k/mo drops (confirmed) are decomposed into three evidenced jobs (dead-end, value doubt, commitment fear), every treatment is anchored to the existing mechanic, and the recovery spine reuses ruled shapes (J8 episode, channel ladder, warm-clickers) rather than inventing machinery. The conditions: the one-pass recovery ruling (DL-003 + DL-004 + DL-060, with DL-065 riding it), the phase split (DL-061) and headline test (DL-062) ratified; no J3 recovery ships while M22_P still fires on the same cohort; and Phase 1 only until the Eng event package lands, because half the journey's triggers do not exist yet (confirmed, taxonomy audit).

Strongest points (keep, do not dilute)

1. Three jobs, not one, and an honest AS-IS. The pack retires its own "zero recovery today" framing: two blanket pushes carry £5.4M lifetime attributed (confirmed as attributed, not incremental) and the three biggest hesitation points have no treatment. Absorb-and-replace at the bid-placed boundary, never stacking, is the right call and is stated with the political optics named.
2. Demand prevention over chasing (the Product/Design lens). Payment friction routes to Product, never to a message; dead-end search is treated as a service defect with CRM as the interim symptom treatment; the journey's success condition includes making its own moments unnecessary.
3. Consent realism. Most abandoners are pre-consent, so the recovery spine is honestly gated on ACT-13 (signed, referenced not re-authored) and the no-channel row carries real weight instead of assuming push reach. No incentives ride any J3 surface (peak-moment consent rule held).
4. Measurement discipline. One headline test (ACT-02 bid-add holdout, the biggest leak with the cleanest read), everything else phase-2 gated; ACT-05 runs exactly as ruled 9 Jul (diagnose price-hesitation vs payment-friction first, test scoped to the recoverable slice); the session-theatre guardrail (+108% sessions, -12% purchases precedent) sits on every in-app surface.

Changes applied this council pass (each: what · why · who · where · capability · outcome · action)

1. Fourteen-field personalisation completeness table added. What: a one-page index on the Campaign brief tab mapping each of the standard's fourteen personalisation and contact fields to the one place it is defined, with marketing, service, operational, product and human contact reviewed together. Why: the standard requires all fourteen explicit; contact priority and required personalisation level were previously implied. Who: reviewers. Where: brief tab. Capability: none. Outcome: checkable completeness at Gate-1. Action: none.
2. All decision citations aligned to consolidated log ids. What: the pack's J3-D1 to D9 citations became DL-060 to DL-068; the completion pass also replaced the stale baseline ids that now collide with different consolidated items (baseline DL-014, the M22_P clash, is now DL-003; baseline DL-015, batch audiences reading journey_state, is now DL-012). Why: the consolidated log renumbered; a pack citing "DL-015, open, P1" would have pointed readers at an unrelated J6/J10 item. Who: reviewers and the log owner. Where: strategy §2/§4/§5/§6, AS-IS table, TO-BE rules, dependency tables, status tables. Capability: none. Outcome: one citation scheme, no wrong-item reads. Action: none.
3. Demand-prevention recommendations made explicit (strategy C12). What: Product is asked whether coverage and window state can surface before results so the dead-end never renders, and ACT-02's target state is the GRO-1612 Place-Bid flow carrying the reassurance natively, with the slide-up as interim that sunsets into it. Why: two of the journey's biggest moments are symptoms of product-surface defects; chasing them without saying so is symptom-fix over root cause. Who: dead-end searchers and bid-add hesitators. Where: strategy §4 C12/§6, brief and process tabs. Capability: Product roadmap answers (GRO-1612 timeline, coverage surfacing). Outcome: prevented demand instead of chased demand. Action: Product answers the two questions (§6 list).
4. Content-card instrumentation caveat carried on the ladder. What: the pack states that content-card impressions and clicks register zero in the warehouse today (confirmed, estate-wide), so the no-channel next-open row runs blind until the shared fix lands. Why: that row carries the pre-consent majority; presenting it as measured reach would be dishonest. Who: unreachable-out-of-app abandoners. Where: strategy C11, ladder rows. Capability: the shared instrumentation fix (cross-journey dependency 8, Eng briefed). Outcome: honest reach claims and a named launch dependency. Action: rides the existing cross-journey item (DL-013 group), no new decision.
5. Four decisions closed as council working assumptions (DL-063, DL-064, DL-066, DL-068). What: ACT-10 authored as the shared Route-Watch engine; interim power-user suppression on the legacy £500 flag; the one-nudge-per-session deepest-step rule replacing five pairwise clauses; BUN confirm-abandon shipped light with no dedicated holdout. Why: escalation discipline; none needs formal authority (reasoning per item in decisions-delta.json). Who: all J3 audiences. Where: TO-BE rules, matrix, brief and process tabs. Capability: session-scoped state (Product/Eng confirm, a validation). Outcome: five fewer items on Sheran's desk. Action: none; each carries its break condition.
6. DL-065 (the 7-day J3/J5 window) corrected from closed to open-adopted. What: the pack now labels the 7-day episode window a working assumption adopted in both packs, with ratification riding the DL-004 pass, instead of council-closed. Why: the J5 pack (already finalised) adopts the window but keeps DL-065 ratification with Sheran; one item under two verdicts is the exact drift DL-003 exists to stop, and the window sets a contact boundary between two journeys' owners. Who: never-transacted abandoners shared with J5. Where: strategy §6, TO-BE episode rules, exit node, status tables. Capability: journey_state. Outcome: J3 and J5 read the same contract with the same status. Action: Sheran ratifies with DL-004.
7. The M22_P one-pass ruling recommendation. What: DL-003 (confirm the 6 Jul pre-close ruling and close the erroneous reopen), DL-004 (recovery owner of record) and DL-060 (bid-placed split, no-bid slice folds into the spine) are put to Sheran as one ruling pass, after Lisa's entry-criteria audit. Why: the estate's highest-attributed canvas currently sits under three verdicts (cross-journey inconsistency 5). Who: every funnel abandoner M22_P chases today. Where: strategy C1/§6, AS-IS and brief tabs. Capability: Lisa's audit; journey_state for the replace plan. Outcome: one verdict, no three-push pile-up. Action: Sheran rules; Lisa audits first.

Working assumptions adopted (labelled, with break conditions)

• ACT-10 is the capture-and-alert engine behind ACT-25/ACT-26, not a fourth moment (closes DL-063). Breaks only if the crosswalk needs a customer-facing ACT-10 card, which nothing currently requires.
• Dwell recovery ships with power users suppressed on the legacy £500 flag, caveat carried in every read (closes DL-064). Breaks when AN-4/RFM lands (re-cut) or if flagged-user complaint rates spike.
• One funnel nudge per session, deepest step reached wins; ACT-13's user-initiated ask stays outside it (closes DL-066). Depends on Product/Eng confirming session-scoped state (validation); monitor per-step CVR for over-suppression.
• BUN confirm-abandon ships light, guardrail monitoring only, no holdout (closes DL-068). Analytics can argue otherwise before build; no causal read is accepted knowingly.
• The 7-day episode window and deepest-step-wins spine run as designed ahead of ratification (DL-065, DL-004: open, adopted in both J3 and J5).
• The 27.5k/34.1k/10.3k step composition stays assumed pending Snowflake verification; no per-moment sizing claim graduates past assumed until Ivi's run lands.

What genuinely requires approval (open, in the consolidated log)

• DL-003 + DL-004 + DL-060 ruled in one pass (Sheran; Lisa's M22_P/M24_P entry-criteria audit first): confirm the 6 Jul pre-close ruling, ratify the recovery owner of record, rule the bid-placed split. DL-065 rides the same pass.
• DL-061 the phase split: Phase 1 ships the no-new-event moments; the dwell/abandon-gated moments are briefed blocked on the J3 Eng event package (Sheran ratifies, Sophie/Eng slot the package).
• DL-062 the headline test: the ACT-02 bid-add reassurance holdout, everything else phase-2 gated (Sheran ratifies, Analytics sizes).

Evidence gaps (REQUIRES VALIDATION, owner named)

1. The step-drop composition behind the confirmed ~72k/mo: 27.5k/34.1k/10.3k are Tableau derivations, unverified in Snowflake. Owner: Analytics (Ivi). Top gap: phasing, sizing and the headline test's power all lean on it.
2. Dwell thresholds per screen and segment (what separates hesitation from reading), defined before Eng builds the events. Owner: Analytics.
3. M22_P/M24_P entry criteria (who they actually chase), plus the M29 numbering collision and dead X67 variants. Owner: Lisa (Braze estate).
4. The ACT-05 price-hesitation vs payment-friction diagnostic and recoverable-slice size (the DL-067 follow-through, ruled 9 Jul). Owner: Analytics + Product.
5. M29 opt-in baseline and not-opted-in in-lead volume (baselines the whole ACT-13 family), dead-end volumes, confirm-abandon rate. Owner: Analytics.
6. Content-card impression/click instrumentation landing in the warehouse (the no-channel row is unmeasurable without it). Owner: Eng, Analytics verifies.
7. Canonical figures for the shared stats in two versions (accidental-bid 3,009 vs 1,349 tickets; confusion reviews 225 vs 127; cohort-labelled non-contactable shares). Owner: CX/Analytics.