Procure 2 Pay (P2P) end to end automation - Possible..?
This was for one of my previous projects, we had a challenging business case of automating end to end of Procurement. My first reaction was why do you need to automate the processes when it’s not revenue generating, when we have more opportunities/ business cases for core business functions and procurement has got all over world reach for vendor list. This was much difficult task because of the number of different kind of transactions they used to have. But, when I look at the cost of operation, all clouds in mind got cleared and we wanted to put best available technologies to get it done.
We started with a business objective to make the procurement process efficient and improve bottom-line performance. The solutions covered end to end function of procurement, asset acquisition and asset life cycle management etc.
Let’s understand what a P2P (Procure to Pay) process looks like. There could be few more processes but overall, it looks something like this.
We skipped the first two processes of “Creating Requisition” & “Finalizing Vendor”, as we hardly could automate anything from there. We picked up the next easiest process (as we thought that time) to automate using RPA, “Creating PO”. The process goes something like this.
Quotation collection was all good and easy but then came the real part how to read all those. The obvious the answer is “OCR”, isn’t it...? well then, the question arises is how the OCR going to read all different types of document formats? The quotations are coming from different vendors and all of them had their own formats. If I can remember it correct, we had more than 60-65% fallout because of innovating quotations structures (read it as beautifying).
Intelligent/ Cognitive OCR was not available that time, so we had to push those to manual. Which didn’t solve our overall purpose completely.
Similarly, during “Invoice Approval” we had the same challenge appeared, because of the different formats and varieties of invoices. Rest all processes were automated and the program was closed with close to 40% automation e2e.
Business got Benefits and we got accolades
- Automated processing invoices raised by supplier up to 40%
- Improve the cycle time to process invoice up to 2 times
- Reduce cost of procure to pay by up to 30%
What can be done more/ differently to achieve more
It felt so great that time with all praises. But now when I think about it, I feel we probably hardly automated anything & we can still do a lot. Few of them can be
- We can look at a Machine Learning model with OCR to automate more to read the quotations. May be call it as ICR (Intelligent OCR).
- We can model the format of the Quotations to use the same for extracting data from Purchase Invoices too. Generally, the companies use the same letterheads (layout) for Purchase Quotation as Purchase Invoices. So, if the letterhead format it framed, 50% work is done in data extraction..
- We can also shortlist & finalize vendor with specific conditions given, unless it’s a service. Expected Price, Quoted Price, past performance, quantity that can be provided and many more of them.
- The complete PO approval can also be automated for certain below threshold $$. This depends on compliance and audit requirements too but certainly can be thought of at least for partial approvals, at least.
- Good/services received can be tracked from ERP materials management or from project tracking systems for services. Here procurement team generally doesn’t have to do much but look for entire complete or as it’s called ‘done done’.
- Now when Purchase Invoice comes from vendor, automation can extract data better %age (as we have captured the format from the quotation module). Data can be entered in ERP and be matched against approved quotation. Deviations might need approvals. Even this deviation is possible of we can use NLP for understanding communique between vendor and consumer.
- We can also think of an API integration with bank for payment to the vendors. This is little ambitious bit with large number of transactions this can be thought of. Else, normal bank wire transfers will still work.
- We can think of an interactive dashboard to give clear view to Accounting & Finance about the $$ accounted for, Payment schedules, expected future expenses and if there is anything wrong in the whole financial chain of the procurement.
- Supplier on-boarding is one more major task performed by Procurement team which can be surely thought of. This will not be much different in terms of process, from general on-boarding processes.
What more Benefits we might get
- Increase the accuracy of invoices raised by supplier up to 80% *
- Improve the cycle time to process invoice up to 3-4 times *
- Reduce cost of Procure to pay by up to 60-70% *
- Improve budget adherence with supplier dashboard and analytics.
- Enhance working capital management with intelligent payments
- Get dynamic discounts for early payments to suppliers. This can be calculated as Savings too
* are guesstimates based on experience
This automation of end to end function of procurement, asset acquisition and asset life cycle management etc. will not only reduce cost for procurement but also for accounting & finance by -
- Better adherence to budget due to tight controls (no manipulation)
- Improved compliance & maintenance due to proactive process
- Cost savings due to visible processes end to end.
It is fair to complete by saying "P2P E2E automation" is possible.
I am sure there are still more things that we can do to improvise. What will you suggest can be done for further improvements?
Technical Sales Leader(India/South Asia). IBM Storage Portfolio including Solutions for Data & AI, Hybrid Cloud and Data Resilience
4 年Interesting point of view Sir. This very thing(p2p) can be deployed on a block chain platform because I learn there is a huge potential these technologies in this area. I enjoy reading your post .. keep sharing