Basware’s new Approval Confidence Index predicts the likelihood of a requisition’s approval. It not only saves time, but also makes the job easier. Read here, to explore the opportunities the ACI offers.
Basware Leverages Machine Learning to Improve Procurement
Basware, the global leader in networked source-to-pay solutions, e-invoicing and innovative financing services, has added a machine learning-enabled, predictive capability to its procurement technology: the Approval Confidence Index (ACI). The ACI is a score assigned to every item on a requisition that provides feedback to both the requisitioner and the approver about the probability, or confidence, of that requisition item being approved. The idea for the ACI was sparked during a Basware “hack day.”
According to customer data, in many cases approvers are busy and may approve requisitions without properly reviewing each item. In other cases, requisitions are rejected because they fall outside the norm for a particular item or spend category. Order requests that are declined during the approval process cause a delay. With the ACI, the approver and the requisitioner are served up insights that could prompt behavior changes. The benefits of this functionality are twofold: requisitioners get instant feedback on the likelihood of an item being approved and, if the ACI level is low, they have the option to select different items or vendors or to provide appropriate justification to increase the likelihood of approval. The system intelligently guides approvers to focus their valuable time on anomalies rather than on routine transactions. By expediting the process and prioritizing transactions for review, the ACI can easily optimize a company’s procurement process.
Bhavin Shah, Director of Product Management at Basware:
“The system leverages transaction history and machine-learning capabilities to evaluate the parameters that result in an approval or rejection. The model utilizes data specific to the customer, which results in an ACI prediction that is unique to that customer. Companies can be assured that the ACI is customized to their process, products, vendors, and services. Of course, machine learning is as good as the data it has access to, so the more data that is available, the more accurate the model is. The system will also update the model periodically to continuously improve the prediction results. The longer-term vision is that the system would become so ‘smart’ that it could automatically process routine transactions so that the approver would only need to approve the exceptions, saving even more time. The ACI is a first step towards that goal.”