Concept: Spend Category Recommendations
You might need to take additional steps to enable this feature depending on your organization's subscription service agreement. For more information, see this Community article.
Spend category recommendations use machine learning to suggest spend categories on noncatalog requisitions and purchase orders. Workday bases recommendations on the supplier invoice data generated from these documents. Depending on the type of document, these predictions depend on information such as the:
- Company.
- Cost center.
- Item description.
- Requester history.
- Supplier.
You can enable spend category recommendations on requisitions on these tasks:
- Checkout
- Edit Requisition
- Request Non-Catalog Items
- Revise Requisition
You can enable spend category recommendations on purchase orders on these tasks:
- Copy Purchase Order
- Create Purchase Order
- Edit Purchase Order
- Review Purchase Order
- Revise Purchase Order
You can configure alerts for when a requester selects a supplier that hasn't been selected for the spend category before. The machine learning model needs to accumulate a sufficient amount of approved supplier invoice records for any new spend categories before recommending them. The model retrains weekly with fresh data to keep the predictions up to date.
Workday requires that your tenant meet these data thresholds to provide accurate spend category predictions:
- At least 1 year of historical requisition or purchase order data.
- A minimum of 10,000 historical requisition or purchase order lines.
- At least 10 unique spend categories in this historical data.