In this blog post, we want to share with you a new way how to analyse and visualise user behavior in customer journeys. Within our by the FFG funded R&D project CUBIC PM, we use cutting edge, innovations and research findings from the areas of artificial intelligence, feature attribution and natural language processing. CUBIC is a cutting-edge solution that combines a novel user interface, a data warehouse, a trained AI model, a text generation application with a self-hosted large language model which is capable to answer questions on own statistic and internal data. It is used as well in combination with process models from statistics which were revealed by using process mining techniques. to enable a comprehensive understanding of customer journeys!

First, we connect the user behaviour actions of a customer journey process model with any kind of data from the environment to find out what factors affect the behaviour. For this, we train an AI model with all available data to perform feature attribution methods. For example, we can discover how the weather, the time of day, the devices or the user's mood influence the buying decision!

Second, we show the customer journey statistics with a chatbot that is linked to a large language model which has been finetuned. The chatbot provides get process-based statistics on customer journeys in natural language. For example, we can ask the chatbot how many users have finished a certain process step, how long they took to do it or what data from the environment had a role. The chatbot can also answer more questions to explain the result more clearly. In the best case, the chatbot can even give us insights or suggestions on how to improve the customer journey.

With CUBIC, you can

• Interact with a chatbot that can answer your questions about statistics using natural language and show you relevant parts of the process model
• Use a self-hosted LLM that meets the standards of the EU AI Act, runs efficiently on your hardware, and performs well on various languages and domains
• Ask questions about your own data sources, such as customer portals, and get insights into your customers' behavior and preferences
• Ask domain-specific questions and follow-up questions based in natural language which is easier to to than learning how a sophisticated UI works
• Visualize customer journey process models from event logs which helps also to identify bottlenecks, opportunities, and trends
• Incorporate environmental data into your statistics and see how it affects the customer journey

CUBIC is a unique and innovative solution that takes customer journey analysis to the next level!

You can use log data mixed with data from the environment as fuel for the own chatbot, which helps and guides you as a daily companion. Together with business processes along a customer journey, it gives you useful information and makes you have a strategic advantage over others or even a much more:

• Chatbot for process-orientated statistics on customer portals and journeys
• Support for daily decisions across departments, with no training needed 
• Customisable framework, connects to various data sources
• Local solution, data protection, with no external resources
• EU-compliant and data-efficient

• allows easy and natural access to statistics
• lowers the barrier for employees to work data driven
• reduces errors and increases the understanding of data
• helps to optimize services and to avoid bad investments
• lowers training costs for employees
• supports from small to large companies with customized statistics and equal opportunities
• adapts to human needs and relieves interpretation work in your team

CUBIC PM is one of our many innovative projects. We are always looking for new collaborators and partners who share our vision and passion. If you are interested in learning more about CUBIC PM or any of our other projects, please feel free to reach out to us or sign up for our newsletter. We would love to hear from you and keep you updated on our progress.