**To:** Priya Raman
**From:** Lena Hartmann
**Re:** Carrier narrative — what's fixable, what's a ceiling, and two resolved side-questions

**1. Three weak carriers need three different responses (confident on direction, caveat on precision — see below).** Albion Parcel's collapse is *not* localized: it hit both countries it serves (UK, Ireland) and every single origin warehouse uniformly the moment November started, with pct-within-3-days dropping from ~90%+ to 30-40% across the board (destination/warehouse breakdowns, Night 1) — and it's specifically a Q4 event: ~88-95% on-time (5-day, order-level) in Q1-Q3, collapsing to ~50% in Q4 and *not* recovering by year-end (carrier-by-quarter report, Night 4). That's a "this carrier broke under load everywhere at once" story — systemic, not one bad depot, but also not (yet) chronic. **Landfall Freight** is uniformly weak across all 18 countries it ships to, all year, every quarter (30-45% within 5 days consistently) — no lane or season explains it away; this looks like a relationship-level problem. **Lowlands Express** is genuinely better on its Benelux/France home lanes than elsewhere (roughly 60-73% vs 25-34% outside), but even on home turf it's still below the standard of our good carriers (which run 87-98%) — so reassigning its non-core volume helps, but it isn't a full fix on its own.

**2. The reliability ceiling is real but I won't give you a precise number yet.** Every version of this calculation — three independent cuts across two weeks — agrees that fixing the three weak carriers would close a *substantial* chunk, not just the long tail, of our delivery-reliability gap: overall order-level on-time (within 5 days) sits around 84-85%, the "good" carriers run 88-97%, and the weak three cluster in the 35-67% range depending on exactly how the query dates are computed. **Caveat, and this matters:** when I asked for the same figures three different ways this week, the same carrier's pct-within-5-days came back meaningfully different each time (e.g. Landfall ranged 35-45%, Lowlands 59-67%, "good" carriers 89-93%) even though the underlying order counts matched exactly every time. I believe this is a day-boundary/rounding difference between queries (date-diff vs. timestamp-diff), not a real swing in performance, but I don't yet have a single trusted number. My honest estimate: closing the gap on these three carriers would eliminate **roughly 40-55%** of our current delivery unreliability — directionally large and worth acting on, but I want that number verified with one engineer, one explicit date-diff definition, before it goes in front of merchants or carriers.

**3. Restocking is not falling behind (resolved).** Once I excluded purchase orders too recent to have resolved yet, the apparent Q4 "backlog" (418 open POs) vanished — the mature, comparable subset shows ~0% genuinely open in every quarter including Q4. What *is* real and unchanged by season: about half of all POs (47-59%, every quarter) arrive after their expected date — a chronic supplier promise-miss rate, not a growing queue. Worth a conversation with suppliers, but it's not getting worse.

**4. Picking discrepancies are a separate track from delivery speed (resolved).** Apparel & Footwear and Beauty have the highest rates of orders shipped short of the full quantity ordered (4.5-4.7% vs 1.6-3% elsewhere), consistent with Tomasz's picking-accuracy finding. But Apparel is actually our *fastest, most reliable* vertical once an order does fully ship (82.4% within 5 days, best in the book). So this looks like a real, separate completeness/accuracy problem — customers sometimes getting short orders — not a delivery-delay problem. Recommend it stays its own workstream.

**Next steps:** (a) get one engineer to re-run the carrier reliability numbers with a single locked-down date-diff definition so I can give you a trustworthy ceiling figure; (b) before any carrier conversation, confirm whether Albion's capacity issue is resolving in January; (c) returns/restock cycle time, still queued from two weeks ago.