It is difficult to prioritize AI and machine learnings when even simple reports are currently unreliable
A recent report by The Hackett Group finds that 63% of procurement executives believe there will be mainstream adoption of advanced analytics within the next 12 to 24 months
It’ll probably come as no surprise to most people that artificial intelligence (AI) can help solve a wide variety of business problems. In fact, we almost expect it. Procurement, in particular, is a great fit for AI and machine learning, because it generates a lot of data to learn from and there’s generally a clear path to return on investment (ROI). The anticipated advantages of AI go beyond the automation of simple downstream functions, with more strategic advantages lie in the upstream activities. However, despite the potential benefits of AI in procurement, it’s application is minimal.
In 2017, Deloitte conducted its annual Global Chief Procurement Officer (CPO) survey and found that “the application of predictive and cognitive analytics is almost non-existent.” This is because procurement leaders are focused on baseline analytics. Deloitte found that 65% of CPOs believe that analytics is the technology area that will have the most impact in the next two years (only 20% said the same about “emerging technology”). Additionally, 88% believe automation and robotics will impact procurement over the next five years. So why is AI not being prioritized? Well, it’s hard to prioritize AI and machine learnings when even simple reports are currently unreliable.
The Deloitte survey found that data quality is the biggest barrier to using digital technology in procurement. Other issues include data integration and data availability. Obviously, if data is unreliable, incomplete and fragmented, then fixing it becomes the top priority, followed by analyzing it. Feeding this data to machines to obtain new insights is thus a lesser priority.
But is it possible for AI to help with the first priority of fixing the data? Companies such as Orpheus are leveraging AI to cleanse, combine and classify spend data, making it more reliable and meaningful. As far as data availability, several companies are trying to fill gaps in internal data sets to enrich them. A startup called Scoutbee uses AI to analyze 3.2 million companies, and then recommends the best vendor for a use case through a simple query.
A recent report by The Hackett Group finds that 63% of procurement executives believe there will be mainstream adoption of advanced analytics within the next 12 to 24 months. And 79% believe that improving their capabilities in analytics, modeling, and reporting will be highly important or critically important. However, similar to the findings from Deloitte, many feel that they are not doing a great job at analyzing and quantifying operational risks, risk drivers, and operational impact and exposure currently. Actually, 39% said they were not using analytical tools to support their procurement planning/budgeting and forecasting processes.
Traditional Approach vs. Strategic Sourcing
The traditional approach to spend analytics is aimed to help procurement reduce, avoid or recover costs with their suppliers. Historical analysis of spend data can also help in the supplier selection process. Yet, processing a typical quarterly batch of data is not the most actionable information on which to form a strategic sourcing effort.
Instead the more recommended route is to expand the role of spend analysis in strategic sourcing. By using a robust data acquisition, cleaning and classification process that is enabled by machine learning, procurement can run spend analysis reports before, during and after a sourcing event.