Supplier Invoice Automation
Supplier invoices are processed and their status updated automatically from creation through to payment, reducing manual processing time and limiting errors and duplicates.
Manually processing supplier invoices across multiple systems was time-consuming and error-prone
Each invoice had to be manually checked and updated across several systems — identifying the payee, checking for duplicates, creating the invoice, and tracking its payment status. This process consumed significant time and carried the risk of missed steps or duplicate invoices.
Invoice processing and status updates are automated end to end
The system automatically identifies the payee, checks for duplicates and data errors, creates the invoice, and continuously tracks and updates its payment status. If an error occurs at any step, the relevant team is notified immediately so it can be resolved promptly.
Integrations
Technical Details
Impact
Step by Step
How It Works
Upload the invoice
Users upload the invoice file directly in the internal application — the starting point of the whole process.
Validate the information
Power Automate automatically identifies the payee and screens for duplicates or data-entry errors before proceeding.
Create the invoice in Xero
The data is sent to Zapier, which creates the draft invoice and matching contact in Xero.
Update the invoice status
Zapier calls back into Power Automate whenever the invoice status changes in Xero (draft created, draft approved, invoice approved, paid), keeping everything in sync.
Notify the invoice status
Power Automate raises the disbursement in Actionstep once approved, and sends status notifications through the relevant channels.
Why It Matters
Key Highlights
Each invoice is processed ~93% faster
Manually processing one invoice (resolving the payee, checking for duplicates, creating the invoice, tracking status, raising the disbursement, sending notifications) takes about 60 minutes; with automation, human time drops to under 4 minutes, mostly uploading the file and reviewing flagged errors. This is a reasoned estimate based on typical processing times, not measured data yet.