The balance of human and machine collaboration is shifting; Artificial Intelligence (AI) is a driving force behind that shift. AI is already changing entrenched ways of thinking and operating, and the domain of services procurement is no exception. This highly traditional, analog function is beginning to embrace the benefits of AI.
The foundation for any successful AI initiative is data and the patterns that can be derived from analyzing it. Accumulating data in a way that makes it most useful for AI requires structure in the data collection platform that will readily allow for making correlations and sound inferences. If we examine the evolution of social media platforms, and certainly the exponential increase in tracking (and track-able) data, we see that unstructured data can now be cataloged, creating extremely accurate predictions based on users’ behaviors, preferences, and even relationships.
The recent developments within Deep Learning have to date most benefitted industries that were data-oriented to begin with: search engines, marketing, and social media platforms ― where the winners in these domains organically ended up amassing immense proprietary data sets. The next wave of industries to undergo such a transformation will include those who are traditionally more data limited. Any given industry and organization will need significant planning and preparation, defining a strategy.
We will eventually see this same kind of evolution in data-centric procurement applications, but it must begin by building up large data moats with information that has long gone untapped. We match this to the commonly described ‘levels’ of autonomous driving that involve progression for both the vehicles and drivers:
Level 1: No Automation
Early adopters of automated procurement have a unique opportunity to capture advantage of the cutting edge, gathering data and feedback in a structured form that works for them – and sets a bar for their industry.
Level 2: Partial Automation
At this phase, vehicles can steer, accelerate, even brake in many situations; however, the human driver must be fully alert, ready to take over if the vehicle is unable to continue.
For procurement, an AI-based platform that has been amassing data can use Machine Learning to make suggestions based on how you’ve acted historically. Relying on pattern-matching and recognition from what you’ve done before, it makes its best guess based on available information. At this stage, its suggestions will have lower accuracy; humans will accept some and reject others.
This is where data volume really starts making the difference. Most companies haven’t spent the time tracking data from their service projects year-over-year. For the time being, procurement involves “small data” but of high value. A single transaction can be worth millions of dollars, as opposed to the potential value of one recorded hour of driving. Therefore, it’s very important to get service procurement data right, and make sure it’s actionable.
Companies that started early with collecting data about services spend are already getting to that increased value which can only happen as the data is amassed and indexed over time. The sooner companies shift procurement to a digital platform, the sooner their data moat progressively deepens, helping models generated through machine learning to become ever-more precise.
Level 3: High Automation
At this stage, the vehicle can assume all driving tasks under almost all conditions without needing human involvement, thanks to cumulative data. However, there should be a passenger who could notice and take control if the vehicle becomes unable to continue.
Combining data moats leads to more subtle understanding of the signals that a Machine Learning system can begin to understand. For procurement, one user’s responses get combined with the many collected, so the machine can find correlations between traditionally siloed data and experiences across an entire organization.
At this stage, the machine can also be exposed to the interactions between people engaged in the buy cycle and start to learn from subliminal signals. Those can include things like provider response time, client feedback, and proposal value, which all help to dictate buying decisions. However, humans will still remain involved to eliminate obviously wrong options.
Level 4: Full Automation
At this point, we’re fully on the move. We have witnessed the performance of autonomous cars in diverse situations and have built up enough empirical evidence that they are able to perform under any circumstance. This performance crosses a wide range, and there is no need for human intervention at all.
The same will be true for service procurement. At full end-to-end AI, well-informed assumptions about what the user wants to do are served up proactively. Now, the role of procurement can shift from tactical to strategic, acting as a partner to business stakeholders. Stakeholders will dictate the goals around how it will impact the business instead of how they can find the right matches, greatly streamlining the entire procurement process while engaging the providers that are the best fit for any unique requirement.
This future may seem far away, but it’s within reach. The time to start the AI transformation of your services procurement is now. As you think through your AI strategy, it will help to seek AI-based solutions that can automatically capture procurement signals and data points. Standardization is critical; that’s what will allow you to look at relevant data in the same way. That, ultimately, is what will enable AI’s predictive ability to find the correlations that will radically transform the efficiency and value of your organization’s entire procurement process.