Purchase orders from supermarket chains arrive as PDFs and spreadsheets in a shared inbox. An extraction step reads stores, dates, item codes and quantities, maps each chain's codes to internal SKUs and creates the sales order in the ERP, with exceptions held for review.
Typical results
≈15 hrs/wk
Order desk time saved on typing PO lines
≈80%
Fewer order entry errors
Every PO
Checked against the agreed price list
Figures reflect typical results for this kind of automation.
The challenge
Each supermarket chain sends purchase orders in its own layout, some as PDFs and some as spreadsheets, with its own item codes. The order desk types hundreds of lines into the ERP each week, about 20 hours of work, and a wrong code or missed delivery date means short shipments and penalty deductions.
Before and after
Order desk hours per week on PO entry
−75%Order entry errors per month
−80%Hours from PO received to sales order created
−96%The automation
Purchase orders land in a shared inbox, and an extraction step reads the store, delivery date, item codes and quantities from each PDF or spreadsheet. A mapping table turns each chain's item code into the internal SKU, and the sales order is created in the ERP. Unknown codes, prices that differ from the agreed price list and delivery dates inside the minimum lead time go to a review queue, where the order desk fixes them before release.
How it works
- 01
POs land in a shared inbox
Purchase orders from each supermarket chain arrive by email as PDFs or spreadsheets, each in the chain's own layout.
- 02
Order lines extracted
An extraction step reads the store, delivery date, item codes, quantities and prices from every line.
- 03
Chain codes mapped to SKUs
A mapping table turns each chain's item code into the internal SKU and checks the price against the agreed price list.
- 04
Sales order created
Clean orders are created as sales orders in the ERP, ready for picking and delivery planning.
- 05
Exceptions to the order desk
Unknown codes, price differences and delivery dates inside the minimum lead time go to a review queue for the order desk.
Starts from
- Chain POs in shared inbox
- Item code mapping table
Delivers
- Sales order in the ERP
- Exceptions review queue
- Price and lead time flags
The impact
Order entry time falls by about 75%, from roughly 20 hours a week to 5, most of it spent on the review queue. Entry errors drop from around 15 a month to 3, and price mismatches are caught before invoicing instead of after a chain deducts them.
Questions about this automation
What if a chain changes its purchase order layout?
The extraction reads fields by meaning rather than fixed positions, so small layout changes usually go through. If a new layout produces low-confidence fields, those orders go to the review queue instead of the ERP, and the order desk confirms the new layout once so later orders process normally.
How are new products or new chain item codes handled?
An unknown item code stops that line from going to the ERP and lands it in the review queue. The order desk picks the matching internal SKU once, the mapping table is updated, and every later order with that code maps automatically without another review.
Which ERPs can it create sales orders in?
Any ERP that accepts sales orders through an API, a database connection or a structured import file, which covers most mid-market systems. This is a multi-step build, usually rolled out in stages, often starting with the two or three chains that send the most orders.