Why I Run the Entire Pipeline Twice to Match Products

Why I Run the Entire Pipeline Twice to Match Products The pricing spreadsheet hit 4,162 rows and the retailer stopped updating it. Not because it was hard. Because it took 90 minutes every morning and the prices were wrong by the time he finished. The system I built to replace that process worked on the first deploy. Pulled supplier data, applied markup rules, synced to Shopify. Clean run. But 600 products had no supplier price attached. The logs showed them as "unmatched." Fifteen percent of the catalog, sitting in the store at stale prices. The first-pass matching was straightforward: take the product's store SKU, find the same SKU in the supplier's price feed, pull the price. For 3,500 products, this worked. For the rest, the store's SKU and the supplier's SKU for the same physical product were completely different strings. One supplier uses one numbering scheme. Another uses something completely different. The same inverter might be STP8.0-3AV-40 in the store and a nine-digit catal