The impact of machine learning on ITSM

The impact of machine learning on ITSM

5 ways IT help desks could benefit from machine learning

ServiceDesk PlusITSM Machine learning


Machine learning, according to TechTarget, is a type of artificial intelligence (AI) that provides computers with the ability to learn without being explicitly programmed. For people like me who need it in simpler terms, machine learning deals with systems that can learn from past data and experience to improve performance of a particular task.

Machine learning has already started making more of a difference in our daily lives than anyone could've imagined. Say, for example, a couple trains their sprinkler system to switch on automatically when cats go on their lawn to shoo them away.

Machine learning also has other daily uses, such as processing online search requests, filtering spam automatically out of our email inboxes, as well as understanding and replying to our speech commands on smart phones. Not surprisingly, machine learning can also benefit ITSM by:

  • Predicting issues and problems proactively
  • Improving search capabilities and knowledge management
  • Classifying and routing issues with greater ease

1. Efficient handling of Level 1 incidents

End users will be able to search for solutions, and self-resolve incidents without the involvement of any technicians. Through machine learning,help desks can be trained to scan incoming tickets and provide end users with solutions automatically, based on the system's previous experience. Google Assistant-style chat boxes will also help end users self-resolve incidents, or get information without even logging a ticket into the help desk.

For example, a user would just have to ping the help desk that "the printer is not working," and the help desk would be able to check the printer's print threshold level, check if it needs a toner replacement, and create a service request for the same. Or if that is not the case, the system would also be able to immediately and automatically send any relevant KB articles that might help the end user.

Help desks could also learn from past experience and data to route tickets or tasks to the appropriate technicians or support groups, thereby automating the ticket assignment process without having to create any rules/workflows. Machine learning would help reduce resolution times and improve the efficiency of the help desk team.

2. Auto-approvals and custom workflows for service requests

With the implementation of machine learning, help desks can be trained to auto-approve service requests based on the employee's role, department, work site, and other parameters. For example, when a designer requests additional design tools/software, the help desk will be able to automatically approve the request and initiate a workflow without waiting for the manager's approval. Further, the help desk can be trained to automatically check the workstation assigned to that designer for minimum system requirements to install the requested tools/software, and create a request to upgrade the system, if necessary. All by itself.

Help desk systems might also be able to learn from past onboarding experiences and suggest the type of software and hardware the user needs, the access permissions they need based on their role/department, a printer configuration setup, etc. These are all options for improving the speed of service delivered to end users.

3. Proactive problem prediction and prevention

With machine learning, help desks will be able to analyze incident patterns and anticipate problems. On top of that, trained help desks might be able to automatically trigger notifications or create a problem ticket for anticipated issues, so that the help desk technicians can look into it at the earliest. Say the performance of an application server starts deteriorating. Help desks would be able to anticipate any application failures from the past performance data of that particular server, warn end users who might be affected, create a problem ticket, and associate any relevant incident tickets with the problem ticket.

4. Highly dynamic change workflows

Change implementations are always associated with a certain level of risk. Without a proper plan and workflow in place, change implementations can be costly. Help desks can learn from previous change implementation data and experience to help create highly dynamic workflows.

For example, with the implementation of machine learning, help desk systems might recognize potential signs of change implementation failure, and prompt, administrators to stop the implementation and execute the backout plan even before the failure occurs. Change management modules guided by machine learning will also be able to make recommendations during the planning phase based on previous experiences.

5. Intelligent asset life cycle management

A sizeable amount of incidents occur due to old IT assets that have degraded performance. Machine learning can help automatically identify which assets might repetitively break down, based on factors like their performance levels, incidents associated with them, etc. Once those assets are detected, the help desk can use machine learning to send notifications to technicians and even help order replacements. The simplest case could be the help desk automatically creating requests for printer toner replacements after a printer has printed N number of pages.

These are some of the simplest use cases showing how machine learning can make life easier for both the help desk team and end users. Though these might not be readily available as out-of-box solutions, they are also not too far away into the future.

Let's support faster, easier, and together