We Ran Sukoshi Mart's Inventory, Warehouse, and Catalog Data across 15,000+ SKUs
Sukoshi Mart is the destination for K-beauty and Asian lifestyle, home to COSRX, Anua, and Beauty of Joseon alongside curated anime merchandise, stationery, and collectible blind-box toys. Every week new product lands from dozens of vendors.
Each unit has to be received and counted into inventory, listed, described in enough detail that a customer searching for a specific skin concern actually finds it, and moved into the warehouse ready to sell, with stock levels kept accurate across every channel the whole time.
That work had outgrown spreadsheets and the people doing it. We took the whole path, from the receiving dock and the inventory count to the product page, and we run it.
- 2.6M+ product attributes maintained across 15,000+ SKUs
- 80+ hours a month returned to the internal team
- 95%+ catalog accuracy without adding a single hire
The Problem
Before working with Aquifer Growth, Sukoshi Mart was holding the whole operation together by hand. Inventory, receiving, attribution, and merchandising each lived in their own place, and none of it scaled with the catalog.
Every product needed 175 decisions
Twenty-five metafields and a hundred and fifty tags per SKU. Across the catalog that is over 2.6 million attributes, each one entered by a person who could get it wrong.
Customers couldn't find products the store already had
Attribution drives search, filters, and collections. When ingredients and skin concerns are missing or inconsistent, the product exists but nobody sees it.
Merchandising ran on instinct
Deciding what to feature meant guessing. There was no reliable view of what was selling, what had just landed, and what was sitting.
Product arrived before anyone knew what to do with it
Receiving, listing, attributing, and putting away happened across separate systems and separate people.
Stock counts couldn't be trusted
Inventory lived in more than one place and drifted as product moved between receiving, the warehouse, and the storefront. Overselling and stockouts followed.
The only obvious fix was another salary
Keeping pace manually meant hiring for it, and headcount on catalog work is headcount not spent on growth.
The Solution
We own every attribute on every SKU
AI generates the attribution, ingredient parsing, concern mapping, and tagging. We review it, correct it, and stand behind the accuracy. Sukoshi's team stopped touching a spreadsheet.
Merchandising decisions now have data underneath them
Clean attribution made collections, filters, and featured placement something we can act on rather than guess at. Slow movers get surfaced. New arrivals get placed while they are still new.
We run the path from vendor to storefront, inventory included
Product gets received against the purchase order, counted into inventory, listed, attributed, then moved into the warehouse ready to sell. One count follows each unit the whole way, so what the store shows matches what's on the shelf. One workflow, one owner, one place to check where anything is.
The systems talk to each other
We built the integrations between their platform, their catalog data, and their warehouse operations so the same numbers appear everywhere.
Structured for how people search next
Product data is built to perform in traditional search and in LLM and agentic search, so the catalog stays discoverable as buying behaviour shifts.
How it looks
Illustrative. Real data sits behind access controls, but this is the shape of it: every SKU received, attributed, tagged, and made searchable in one system.
The Results
Sukoshi Mart no longer spends 80 hours a month on catalog work. Over 2.6 million attributes stay accurate above 95% without an internal person assigned to it. Product moves from arrival to live faster because receiving, listing, and attribution are one workflow instead of four handoffs. Merchandising runs on what the data says rather than what someone remembers. The team scaled the catalog without scaling the headcount that normally comes with it.