Artificial Intelligence to Improve Decision-Making in Procurement
We’re entering the age of AI procurement. What is the best way to secure a sustainable competitive advantage from new technologies?
Until now the focus of attention has been process automation or augmentation. Our published report, “Digital Transformation in Procurement: How Close Are We?”, which is based on a global survey with responses from 391 CPOs and other senior procurement decision-makers, showed that most progress had been made with the automation of supplier management, procure-to-pay and spend management processes. Some of these are supported by artificial intelligence, for example to map inbound supplier invoices for automated data entry.
Momentum and Direction
The Gartner research note published in September 2018, entitled “The Impact of Artificial Intelligence on Procurement Software Applications” argues that while the benefits of such automation are significant, they are not able to deliver sustainable competitive advantage:
“Product managers have traditionally implemented AI technologies to enhance procurement processes through automation or augmentation, or by improving user experience. While these benefits are significant, they are not able to deliver sustainable competitive advantage.”
Once you have automated a data entry process, there is nothing further to gain from it: it’s a one-off. Unsurprisingly, therefore, our own survey surfaced considerable doubt about where to go next with artificial intelligence technologies.
Improving Decision Making Power
The Gartner research note recommends that as a next step, procurement software vendors should harness the powerful machine learning algorithms developed in recent years to create solutions that improve procurement decision-making. These should be used in combination with conversational platforms (e.g. chatbots, smart assistants) that improve user experience.
“The evolution of machine learning algorithms and conversational platforms helps [vendor] technology product managers to invest in developing business applications that improve quality of decisions and enhance user experience. They can focus their investment in developing decision guidance systems,” according to the Gartner paper.
It goes on to recommend that vendors trial such systems at selected customer sites by developing robust procurement workflows from multiple sources of data and validating these in short, focused proofs of concept. Leveraging machine learning it should then be possible to define a decision matrix for each node across the workflow; the solution can then recommend optimal pathways to users and continuously improve these recommendations through an iterative process.
“The first step in developing successful decision guidance systems is to develop robust procurement workflows from multiple sources of data.”
Deployed through conversational platforms, these decision guidance systems can deliver guidance in a seamless, non-intrusive manner to enhance the procurement user experience.
“They can deploy these decision guidance systems through conversational platforms like chatbots, VPAs and VEAs to deliver contextual guidance in a seamless, non-intrusive manner to enhance procurement user experience”
The Importance of Supplier Data
Key to the success of such projects will be the creation of robust supplier collaboration solutions through the continuous collation of data from a large number of data sources and incorporating the lessons learned from such data into an algorithm and delivering these solutions autonomously.
These data sources typically include the end-customer’s approved supplier network, third-party supplier networks, and private company information supplier networks. JAGGAER, with a global network of four million suppliers, is particularly well placed to provide such data.
“AI is well-positioned to deliver robust supplier collaboration solutions by enabling product managers to continuously collate data from a large number of data sources, incorporate new lessons from such data into an algorithm and deliver these solutions autonomously.”
A JAGGAER Digital Assistant: A closed-loop machine learning system
Based on a similar analysis of the next steps for digital transformation in procurement, JAGGAER has itself been moving towards a closed-loop feedback system that relies on machine learning for continuous improvement. We call it the JAGGAER Digital Assistant.
JAGGAER, ERP and third-party data is fed into a central data layer. The information is used in traditional analytics and reporting, but what is new is that algorithms are now providing real-time support for decisions, recommendations and actions.
Typically, there might be several recommendations and the end-user makes a decision based on which of these makes most sense. These decisions drive machine learning in an artificial intelligence application, which returns information on these decisions to the JAGGAER application set, enabling continuous improvement (see graphic).
Digital Assistant Functionality
For example, a decision could result in a change in the status of a supplier, the supplier’s performance rating, or it could mean the creation or updating of an RFQ, revising a contract agreement, etc. The point is that this is done automatically, without manual intervention. (In practice this means the updates are triggered by an API, not requiring an email or other human action.)
There are three building blocks to the logical architecture behind the JAGGAER Digital Assistant:
- the Recommendation Engine (which interrogates and checks the base data);
- the Workflow Engine (which enables the user to execute an action);
- and the Artificial Intelligence Engine (by means of which the system learns and feeds back the new knowledge into applications).
This means that the application could recognize an action on which a decision has been based in the past, so it can make a recommendation for similar action.
The Benefits of a Digital Assistant
AI procurement benefits, and specifically those from a digital assistant, fall under two headings, strategic and operational.
Strategically, a digital assistant simplifies strategy development and execution within the procurement function. Available data sources are analyzed on an ongoing basis, recommendations and actions are proposed automatically, and these proposals are based on the expertise and best-practice know-how of specialists.
This means that specialist knowledge is shared and taken advantage of more broadly across the organization (for example, know-how that was previously only available within the procurement function at corporate HQ is now at the disposal of employees working at regional or local level).
The system learns independently and evolves, and the benefits are accessible and transparent to everyone involved across the procurement value chain – a clear and effective basis for collaborative work.
In terms of operational benefits, i.e. through process optimization and automation, the digital assistant greatly simplifies interaction with procurement software. Some manual tasks are completely automated through robotic process automation (RPA). Elsewhere, a single human input command can initialize an entire process chain. Recurring tasks are performed by the system, with human beings only needing to intervene to take care of exceptions.
There are no outstanding technological obstacles to such developments. However, there are a few criteria that must be met. Most notably, all the data must be of high quality and easily accessible by intelligent systems.
Secondly, all of the expertise that is in people’s heads must be made available in digital format. And thirdly, processes must be capable of being automated. However, it does not require the implementation of a complete suite of software. The JAGGAER Digital Assistant can work with individual modules in the JAGGAER ONE suite.
The JAGGAER OTD Predictor
In March 2019, JAGGAER announced the successful trialing of the world’s first AI-based algorithm for predicting the probability of on-time delivery of goods and materials in direct procurement.
The JAGGAER OTD Predictor provides immediate information about the likelihood of delays to deliveries from suppliers, enabling supply chain managers to mitigate risks of disruptions to production flows and reduce the costs that these can cause. ZEISS, the internationally leading technology enterprise operating in the fields of optics and optoelectronics, has partnered with JAGGAER to ramp up the OTD Predictor.
The algorithm predicts if an order will be delivered on time. In testing, the accuracy was greater than 95 percent, which potentially of huge and sustainable benefit to manufacturing companies, especially those that rely on just-in-time components and materials delivery. By using the OTD Predictor, companies can identify where there is a risk of late delivery, and take actions to mitigate that risk, for example by spreading an order over a second or third source.
The OTD Predictor was “trained” by feeding millions of line items through the algorithm to learn from previous events. It uses 50 separate data dimensions to predict outcomes. Its machine learning algorithms mean that these predictions should get even more accurate over time.
In the final article in this series we will consider in greater depth why better access to supplier data will enable the development of more AI-powered solutions that improve supplier collaboration.
Gartner, The Impact of Artificial Intelligence on Procurement Software Applications, Balaji Abbabatulla, Magnus Bergfors, Patrick Connaughton, 14 September 2018
Gartner does not endorse any vendor, product or service depicted in its research publications, and does not advise technology users to select only those vendors with the highest ratings or other designation. Gartner research publications consist of the opinions of Gartner’s research organization and should not be construed as statements of fact. Gartner disclaims all warranties, expressed or implied, with respect to this research, including any warranties of merchantability or fitness for a particular purpose.