← back to work

Cut a 40-hour catalog refresh to under 2 hours of review time

A multi-location retail brand · Retail · Product catalog image sourcing

Someone on the team was manually Googling every product, picking an image, downloading it, renaming it, and uploading it — for hundreds of SKUs. A full catalog refresh was a week of one person's time, and the quality depended entirely on who ran it. Feed it a spreadsheet, get back a ranked image list in the time it used to take to do ten products.

The results

  • 40-hour catalog refresh task cut to under 2 hours of review time
  • Any non-technical team member can run it — no training required
  • Image quality standardized across the entire catalog, not dependent on individual judgment
  • Failed lookups flagged explicitly, never silently dropped
n8n workflow showing the product image finder pipeline — Google Sheets input, batch splitting, SerpAPI image search, AI agent selection, Shopify product creation, and failure logging
n8n workflow — spreadsheet input to Shopify catalog
Shopify product list showing populated product images after the pipeline ran — hair care products with thumbnails, inventory counts, and active status
Shopify catalog after pipeline run — images sourced and attached automatically
Google Sheets source data showing product names, cost prices, full prices, categories, brands, and supplier columns — the input spreadsheet the pipeline reads from
Input spreadsheet — product catalog fed to the pipeline
Product Image Finder ArchitectureFour stages left to right. A SKU List CSV feeds Image Search via SerpAPI. Scored and Ranked selects the best image per product. Output CSV delivers the ranked list ready for catalog import. A dashed branch drops below Scored and Ranked to a Flagged for Review node when no suitable match is found.SKU ListCSV inputImage SearchSerpAPIScored & RankedBest pick selectedOutput CSVReady for catalogno matchFlagged for ReviewManual needed

Tech stack

  • n8n
  • SerpAPI
  • JavaScript
  • CSV I/O