One Airtable system for a fashion brand, from sketch to purchase order
A growing fashion brand ran design approvals, factory samples and purchase orders across spreadsheets, email and chat. Now it all lives in one Airtable base, with about 75 automations doing the routine work.
- Client
- Fashion brand
- Tools
- Airtable, n8n, Shopify, Slack, AI
- Scale
- 50+ tables, about 75 automations
- Since
- 2026, ongoing
The problem
Every product this brand makes goes through a long chain: design, sign-offs, samples from factories, reviews, more samples, a purchase order, invoices and customs paperwork, and finally Shopify. Each step had its own spreadsheet, email thread or chat. Nobody could say at a glance where a style was stuck, and the same details were typed again at every step, so small mistakes travelled all the way to the factory.
What I built
One Airtable base that follows a product from the first sketch to the shop:
- Design sign-offs for style, fabric and logo, so a style can't move on to sampling until the right people have approved it
- Sample rounds with factories. Each department reviews its own part (fit, fabric, artwork, labels), and the sample's status is worked out from those verdicts instead of being typed in by hand
- Purchase orders built inside Airtable and turned into PDFs automatically, with a standing message thread per order so conversations with the factory stay attached to it
- Documents on demand. Tech packs, mood boards and exports for Shopify and product-safety paperwork are generated from the data, not assembled by hand
- Daily sales by product pulled from Shopify, so the design team can see what sold when planning the next season
- An AI first draft. An n8n workflow sends sample photos to an AI model, which drafts review notes for the team to check before anything goes to the factory
Behind the scenes, n8n runs on the brand's own server to handle the PDFs and the heavier jobs, so costs don't grow with every document.
Keeping the data clean
A system this size, with tens of thousands of records across more than 50 tables, only works if people trust it. I cleaned out several thousand duplicate records left over from older automations, fixed the race conditions that created them, and set up a daily check that sends the team a short Slack digest of anything filled in wrong, such as a duplicate product code or a sample that was approved but never closed off.
The result
- One place to look. Anyone can open a style and see its whole history, from sketch to purchase order.
- About 75 automations do the routine work: creating the next record, notifying the right person and generating documents.
- Fewer mistakes reach the factory, because details are entered once and checked every day.
- The team runs it themselves. Each department has its own simple interface showing only what it needs.
Want something like this for your products? Book a free rant session.