Let’s talk about ITSM and machine learning. We’re now well settled into 2018 and IT service management (ITSM) and IT service desks still exist, despite the regular debates and discussions that question how long they’ll survive the digital revolution. In fact, the demand on service desk is growing – with HDI’s 2017 Technical Support Practices and Salaries Report stating that 55% of support organizations saw an increase in ticket volume over the past year.
In addition, and at the other end of the spectrum, more organizations saw a decrease in support ticket volume last year (15%) than in 2016 (10%). With the biggest contributing factor to fewer tickets being self-help! However, HDI also reports that last year the cost per ticket rose to US$25, compared to just US$18 in 2016. It’s not something most IT departments want to see. And, thankfully, automation based on analytics and machine learning can improve service desk processes and performance by reducing errors and increasing quality and speed. Sometimes this goes beyond human capabilities, with machine learning and analytics a key foundation for a smart, engaging, and responsive IT service desk.
This article takes a deeper look at how machine learning can solve many of the service desk and ITSM issues related to ticket volumes and costs, and how to create a faster and intrinsically-automated service desk capability that employees gladly use.
Smarter ITSM through machine learning and analytics
My favorite definition of machine learning is one from MathWorks:
“Machine learning teaches computers to do what comes naturally to humans and animals: learn from experience. Machine learning algorithms use computational methods to learn information directly from data without relying on a predetermined equation as a model. The algorithms adaptively improve their performance as the number of samples available for learning increases.”
With the following capabilities now found in some ITSM tools, based on machine learning and big-data analytics technologies:
- Bot-driven support – Virtual agents and chatbots can automatically suggest news, articles, service, and support offerings from knowledge catalogs and public requests. This 24×7 support, in the form of suggested learnings for end-users, helps ensure that their issues are resolved in a significantly faster manner. The key benefits are better user experience and fewer tickets being logged.
- Smart news and notifications – This enables IT to proactively notify users about potential issues that may impact them. And IT can provide suggested workarounds through personalized notifications that empower end users with relevant, useful information on issues they’re likely to encounter as well as advice on how to avoid them. Informed users will appreciate proactive IT support and will also log fewer tickets.
- Smart search – As end users search for information or services, a contextually-relevant knowledge management system can provide recommendations, articles, and links using machine learning. End users will typically skip some results while clicking on others. These clicks and view counts are included in the “weighting” criteria when re-indexing the content over time, so the search capability dynamically adjusts. As end users provide feedback in the form of “thumb up/down” voting, it also affects the rank of the content they and others can find. In terms of benefits, end users can find answers quickly and feel empowered/self-sufficient, and service desk agents can resolve more tickets on first touch and meet more service level agreement (SLA) targets.
- Hot-topic analytics – Here analytic capabilities identify patterns across structured and unstructured data sources. Information on “hot topics” is graphically displayed like a heat map, showing where the size of segments corresponds with the frequency of certain topics or groups of keywords in demand by users. Duplicated incidents will be found instantly, grouped, and resolved together. Hot-topic analytics also discovers incident clusters with a common root cause and significantly reduces time to identify and resolve the underlying problem. The technology can then automatically create knowledge articles based on similar interactions or similar problems. Finding trends in any data makes IT proactive, prevents incidents from reoccurring, and consequently increases end-user satisfaction while decreasing IT costs.
- Smart ticketing – Today’s end users expect to be able to submit tickets as if writing a tweet – namely, a very short message, in natural language that describes an issue or a request, which might be sent via email. Or even just an attached photo of an issue sent from a mobile device. Smart ticketing accelerates the ticketing process by automatically populating ticket fields, all based on what the end user wrote or on and optical character recognition (OCR) scan of the image. Using a dataset of observations, the technology then automatically classifies and assigns tickets to the appropriate service desk agents. Agents can reassign tickets to different support teams and can overwrite automatically populated fields if the machine learning model was not optimal for a given case. The system learns from new patterns, enabling the best-fit function to perform better in the next situation. This all means that end users can open tickets easily and quickly, leading to improved satisfaction with workplace tools. This capability also decreases manual work and errors and helps reduce the resolution time and costs.
- Smart email – This is similar to smart ticketing. The end user can send an email to the service desk and describe the issue in natural language. The service desk tool creates the ticket based on the email content and also automatically replies to the end user with links to suggested solutions. End users get better experience, as opening tickets or requests is easy and convenient, and IT agents have less manual work.
- Smart change management – Machine learning also supports modern change analytics and management. Given the frequent number of changes that businesses demand today, intelligent systems can provide agents or change managers with suggestions that can help them to optimize the environment and increase the change success rate percentage moving forward. Agents can describe what change is needed in natural language, and the analytics capability will inspect the content for affected configuration items. Changes are scheduled, and automatic indicators will tell the change manager whether there are any issues with the change, such as risk, scheduling in an unplanned window, or not being approved. The key benefits of smart change management is a faster time to value with fewer configurations, customizations, and—ultimately—less money spent.
Ultimately, machine learning and analytics transform ITSM systems through intelligent assumptions and recommendations about ticketing issues and the change process that help agents and IT support teams describe, diagnose, predict, and prescribe what has happened, what is happening, and what will happen. End users get proactive, personal, and dynamic insights and fast resolutions. With much of this is done automatically and without human interaction. And as the technology learns over time, processes only get better. Importantly, all the smart features described in this article are already available today.
Vesna Soraic
Vesna Soraic has over 20 years of experience working with IT executives and practitioners to take advantage of automated IT management. Vesna writes and speaks on these topics regularly and is also a regular contributor to the Micro Focus ITSM blog.
In addition, she has a MS in Computer Science from the University of Ljubljana in Slovenia.
