How AI disrupts the P2P process

Emerging TechnologiesS2P/ P2P
How AI disrupts the P2P process

Which procure-to-pay (P2P) manager has not dreamt of optimising their processes into a single standardised process they could apply to all purchasing categories, business units and countries?


For many years, procurement functions have tried to unify, standardise and optimise the number of processes while they were centralising their activities. The mantra was fewer processes, fewer specifications, fewer suppliers.


Blinded by the sake of optimisation, many leaders tried to severely reduce the number of existing processes to deliver efficiencies. Or so they believed. Indeed, the cost of maintaining a single process is far lower than having many exceptions. In addition, employees can easily remember one process.


Nonetheless, standardisation and centralisation have their limits.


After 2008’s financial crisis, regulations tended to converge. This is no longer the case. Increasingly stringent local regulatory requirements have made a simple process, such as P2P, more complicated. Think about the impact of the European Union’s General Data Protection Regulation. We are entering an increasingly complex world. This often makes it difficult to meet the simplification objectives.


On the other hand, not abiding by local specific needs creates dissatisfaction among process stakeholders in the long run. Subsequently, there comes a point where employees look for any means to avoid such a rigid process. Compliance would be at its lowest again.


Shall we then allow a different process for any specific requirement?

Doing so will multiply processes and subprocesses to the point nobody would ever remember them, not even experts. Specific processes cost more time to go through and require greater resources to maintain. Thus, P2P managers seek a compromise between a reduced number of processes and a need for specific requirements. What number allows this compromise? Having more than three ways of doing things is already too challenging for most people, whose primary job is not to purchase through a maze of processes. But for any global P2P team, specific requirements will surely demand tens of exceptional processes.


This compromise, which P2P managers too often have to make, often tends to go beyond stakeholders’ capacity to remember processes. Process simplicity is then sacrificed due to specific requirements built in. This time, compliance will be impaired by employees being lost in the maze of processes.


Informing by any means has been the first response procurement departments have thought of to solve this problem. Newsletters, blogs, banners, leaflets, process libraries and repositories have multiplied in every company. Procurement has become marketing savvy. However, only the most curious employees would read through this content and only the most frequent users of the process would remember it.


The second solution that procurement departments have implemented is to train people. As a result, the procurement training program blossomed. But, classrooms are empty most of the time! Or only the same people keep coming every year to these training sessions. Again, this format may work for heavy process users, but most will ask “Why do we need training to buy something within a company while any newbie can shop online pretty easily?”.


So, are we fighting a useless war? A puzzle without a solution?

The answer might have been ’yes’ for the last couple of years. The answer has been switching to ’no’ these days. How? Here is where the power of artificial intelligence (AI) comes into play. “What can AI do about that?” you ask. Well, several things.


First, AI’s memory is far greater than ours. It can remember all the various processes. But one could argue this is no different with traditional computer technology, which also memorises our data. This is where the second point comes into play: AI can retrieve the information in a wide variety of formats and locations that could not have been done before. But what is the greatest advantage of AI is its superior user interface capabilities. It is able to translate both ways: human language into machine actions or machine language into meaningful content for us.


Let us imagine the future of P2P with AI. A random user will think about the product or service he or she needs to buy. The user will not think about the process to get this product or service; they will directly address this need to the AI interface from their procurement platform, either by typing on a keyboard or through a verbal command. AI will then interact with the user through relevant questions to qualify the need (volumes, product specifications, budget, deadline) and refine the answer it will provide the user. Eventually, this answer might be an order to a preferred supplier in the case of a mere commodity purchase. It may, alternatively, become a sourcing request, which would go to a relevant sourcing category manager because no preferred supplier is found and a budget threshold has been passed. AI will have the job of determining the appropriate process to meet a user need and providing guidance on the process to follow, the right goods or services to buy, the right suppliers, and so on. In the end, a user will not need to know the process – he or she only cares about when that need will be satisfied and the price.

AI will not do everything.


First, procurement teams will need to map their processes and translate them into rules that would be understood by AI. The machine needs to be properly fed upfront to be able to run afterwards. Second, they will need some maintenance with the business and process evolution. On this side, we can hope machine-learning capabilities will reduce this maintenance work.


As a result, AI will empower true guided buying. Maverick spend, noncompliance or user dissatisfaction will be things of the past.


Arnaud Malarde is senior product marketing manager at Ivalua.


This contributed article has been written by a guest writer at the invitation of Procurement Leaders. Procurement Leaders received no payment directly connected with the publishing of this content.

Arnaud Malarde
Posted by Arnaud Malarde

Want to learn more? Please fill in your details to hear from us.