Get ahead of customer demand with “citizen” predictive process monitoring

Apromore has added predictive process monitoring to its process mining platform. While predictive analysis is not new to the world of process mining, the Australian-founded business is out to make it easier and faster for enterprises to deploy, with a “citizen”-friendly approach and no need for coding or data scientist involvement.

Real-time predictions help enterprises get ahead of customer demand, with typical models foretelling events such as whether a customer will accept a quote, loan offer, or insurance claims settlement. Machine learning (ML) and artificial intelligence (AI) models can be trained to become accurate predictors. To develop, deploy, and train such models often requires access to in-demand data scientists and other specialists in short supply. Even with the right resources, the models can take months to train.

A point-and-click interface requires neither coding nor data scientists

Apromore enables users to build predictive machine learning models for a range of business processes using a point-and-click interface with no need to write code.

The predictive process monitoring add-on includes training and runtime modules. The training module applies machine learning algorithms to build its predictions from your event log, ingesting historical cases. The trained model delivers predictive dashboards showing predictions for each open case in the process.

Those predictions will include the time until completion of the process and any binary or categorial outcomes such as “Will the customer pay on time?” or “Will the customer complain?”

A European insurance firm has already proven the module

The module has already been proven in a deployment at a large European insurance company to reduce SLA (service level agreement) violations in its claims handling. It allowed claims managers to monitor performance across various claims and anticipate which needed intervention to reduce SLA violations.

Customers will appreciate the ability to automatically trigger warnings about predicted KPI breaches, feeding this to decision makers via email, messaging apps, or dashboards, and offering guidance on the next best action to take. Integration via Workato, MuleSoft, or Apromore’s APIs means those next best actions can automatically trigger actions for other automation software to perform.

HFS believes this module may put Apromore in direct competition with rivals such as ABBYY Timeline or Celonis. ABBYY offers predictions based on observing past events. It also offers solutions, using Timeline’s prescriptive analytics to both diagnose problems and make changes toward improved outcomes. Celonis has a Python-based Machine Learning Workbench for predictive insights and adds to this with an AI-powered “Action Engine” providing recommendations for improvements.

The Bottom Line: Process predictions can get you ahead of customer demand, but it’s down to you to design the right response to the insights gained

Apromore’s “citizen”-friendly route to process predictions should entice enterprise leaders looking for a fast route to getting a step ahead of customer demand. The data you need is already in your event logs; Apromore offers a fast way for non-technical users to surface it and apply ML to predict with it, presenting the information for managers or systems to act on. It’s only as effective as your teams’ decisions or your systems are directed to act on.
By making this available to non-technical users, Apromore’s solution makes it easier to scale the capability across business processes without having to wait for the availability of data scientists—an advantage skills-constrained CIOs and CTOs will be eager to consider.