{"slug":"vcc-cost-backfill","total":3,"limit":100,"offset":0,"since":null,"commits":[{"hash":"bfc5ef5","date":"2026-09-10 13:09:56 -0700","author":"vp-dw-commerce","subject":"wallquest cost_price loader: use net_cost not our_price (retail) (TK-11410)","body":"apply-2026-08-24.sql block 02 loaded cost_price=wallquest_catalog.our_price,\nwhich is the DW RETAIL (=net_cost/0.65/0.85), not the net cost — zeroing margin\nand false-flagging ~1,145 WallQuest products as below cost (NA508 cost_price\n$226.68=retail vs real net_cost $125.24). Fixed to read vc.net_cost. Added\ncorrective-wallquest-cost-TK11410.sql (DRAFT/gated, snapshot+undo) to repair the\n~2,484 already-contaminated rows; NOT run (live backfill is gated).\n\nCo-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>\nClaude-Session: https://claude.ai/code/session_01PuFgoHYPwznEaT5iBdCRny"},{"hash":"b3d5e82","date":"2026-08-24 13:03:25 -0700","author":"Steve Abrams","subject":"VCC cost backfill: real transaction-wrapped apply-2026-08-24.sql (4920 rows, Mac2 path)","body":"Kravet already backfilled (no-op); Kamatera staging too stale/thin to host the\nJOINs (missing cols) so target = Mac2 mirror. Proven via ROLLBACK dry-run.\n\nCo-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>"},{"hash":"4bf2f27","date":"2026-08-24 04:14:22 -0700","author":"Steve Abrams","subject":"Add VCC cost backfill dry-run analysis for TK-10785","body":"Read-only analysis of cost data backfill potential:\n- Baseline: 4,076/89,904 ACTIVE products have cost (4.53%)\n- 13 vendor catalog sources identified with joinable cost data\n- Best join: mfr_sku (more reliable than dw_sku)\n- Total fillable: 17,227 distinct products after dedup\n- After backfill: 21,303 products with cost (23.7%)\n- Largest source: kravet_authoritative_pricing = 12,242 products\n- Full gated SQL provided; write to shopify_products is Steve-gated\n  (Kamatera-canonical table, blast_radius > 500)\n\nCo-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>\n"}]}