Mærsk uses ServiceNow to handle helpdesk tickets. After an initial analysis, it was discovered that 10% of tickets take longer than 13 days and 5% of tickets take longer than 3 weeks.

This is due to the fact that 45% of tickets are redirected, and 10% are redirected at least three times which cause longer resolution times, potential SLA breaches and poor customer service.

The KPMG NewTech Machine Learning (ML) team developed a stand-alone supervised ML solution that integrates seamlessly into ServiceNow. It periodically reads the data about previous incidents from the SQL database and updates its knowledge which is used to present recommendations to the helpdesk employee live (less than 1 second execution time).

The user sees not only the usual information but also a prioritised list of recommendations based on ML. The recommendations include an indication of certainty, so the user can be extra aware in situations where the recommendations are weak

It only took Mærsk 6 weeks to break even from the solution. 10% of the business case reduced the time spent by helpdesk, while the other 90% had savings related to faster resolution. The KPMG team was able to reduce the fraction of redirects from 45% to 26%.