|Predictive analytics is still relatively new in procurement, but it’s increasingly important, particularly in organizations that have already been through the cycle of spend analytics, supplier leveraging, segmentation, and consolidation. Prediction requires a deep understanding of the technical and commercial attributes of the supply chain ecosystem, as well as advanced statistical and modeling capabilities. This provides the ability to forecast revenue, mitigate disruption, identify market opportunities, and much more. It is often built on a strategic platform that provides data visibility and near real-time availability, as well as advanced data warehouses for collection of relevant third party datasets. They can also deliver an in-depth view into supply chain operational outcomes that are not readily apparent to executives in the business.
Furthermore, prediction requires highly sophisticated levels of procurement analytical modeling capability. This represents an important opportunity for organizations: To harness the power of multiple computing systems that enable the convergence of data pulled from machines, systems, and social media into a powerful real-time predictive system of the future. There is an awful lot of hype about this capability, so let’s think a bit more about what this means.
One of the most important elements in designing predictive systems for procurement is deciding what data is important, and what data is available in a cleansed form. Being able to pull from many of the different pools of data requires the right type of analytical talent, and an ability to integrate data into an object oriented database is key. This means that an important requirements for predictive analytics is having the right people to engage in data modeling approaches. More than data scientists, what’s needed is subject matter experts who know how to engage with the business in a consultative manner, and who also understand systems, data structures, and modeling approaches such as statistical testing, decision analysis, simulation, sampling methodologies, and clustering techniques. These types of analysts are more difficult to find, but they’re critical for driving actionable insights.
A strong basis in analytical influence is needed; but this is difficult to achieve without investment in procurement systems such as spend analytics, contract management, and transactional procurement systems. In this research, we suggest that to become a strategic partner, procurement must first seek to understand the problem stakeholders are facing, and then be able to effectively articulate the question that needs to be answered. Can a supply chain organization produce supply analytics without a robust set of spend, contract, and supplier life cycle data? Many executives we interviewed in our research responded “yes”. Is it significantly more difficult to produce analytics absent reasonable spend data and contract visibility? Absolutely. The executives we interviewed described significant challenges in constructing reliable and trustworthy analytics for stakeholders, due to a lack of systems. But it was also rare to find organizations that have achieved a high level of spend data integrity across all of their business units. The message is clear: procurement must seek to build analytical insight in the absence of perfect data, and be able to leverage “whatever data is available”. This may involve importing third party external datasets to augment the lack of robust data internal to the organizations existing databases, and creative thinking around how to exploit this data to advantage.
As Bob Murphy, IBM’s former Chief Supply Chain Officer, put it: “Fueled by analytics, procurement can derive insight from disparate sources of information and uncover intelligence for competitive advantage. This paves the way for us to develop an even deeper understanding using cognitive technologies that will help us further transform the procurement landscape as we engage across our supply base and with business partners to unlock value from all types of data that have been hidden in the past.”