Measuring helpdesk processes with artificial intelligence

IT has always had an obsession with measuring things — disk space, CPU speed, network throughput, and so on. The modern helpdesk typically has to juggle dozens of monitoring tools and agents, constantly feeding back alarms, alerts and performance statistics.

Every now and then something flashes up red or amber, clamoring for attention among the general digital hubbub. As soon as someone has the time, or has worked out what to do about it, the issue may well get fixed. In most cases, that’s usually the end of the matter.

But that’s surely not the most efficient approach, is it? Firstly, do those particular alerts really matter? And secondly, if they do, what improvements can be made so they don’t recur?

Conveyor belt thinking

If you’ve read my previous blogs, you’ll know that I’m a strong advocate of adapting industrial processes to drive IT thinking and digital strategies. And if you look at the way many helpdesks operate, it’s clear that there really are compelling financial and commercial reasons for investing in better ways of managing tickets.

Instead of getting the alert, fixing the problem, and moving onto the next ticket, what are we on the helpdesk learning each time? Are we doing any root cause analysis that can prevent the problem arising again, and that can improve the efficiency of our operations? Can we save money by automating future responses so that helpdesk staff are better engaged on solving more complex issues?

As each ticket comes rumbling down our virtual conveyor belt, it’s really our duty as IT helpdesk professionals not only to fix each problem, but to make that conveyor belt runs as smoothly as possible.

As far as measurement goes, that means understanding the context of every incident or request. Knowing you’ve dealt with 200 tickets one week and 300 the next is next to useless if you don’t understand the business impact each has had. A better process is to prioritize, categorize, assign, fix, learn and re-apply the solution should a similar problem happen in future.

The good news is that with machine learning and artificial intelligence, creating dynamic continuous improvement is not only easy but highly cost-effective.

Artificial intelligence and the helpdesk

Everyone accepts that automation can detect and measure incidents. Automation can often actually solve the ticket without manual intervention. What concerns me is how few people realize that with modern machine learning, you can actually automate everything from prioritizing, categorizing and assigning the ticket to carrying out root cause analysis.

Then, because everything happens automatically, you have a clear audit trail, a record of what action has been to solve the problem, a suggestion of how to fix the root cause, and a model for dealing with future incidents. Suddenly we can provide measurements that really mean something to the business.

Going back to our virtual conveyor belt of tickets, how useful would it be if you could remove the problem from the belt (so as not to interrupt anything else), take it over the bench, look at it from every angle, see what could be improved, and then feed that information back? That’s what machine learning and artificial intelligence can do for you.

If you’re skeptical about how much manual intervention this might save, just think about everyday things like arranging file shares, VPN access or ERP permissions. To action those, your helpdesk staff are only authorized to act within policy guidelines that already exist — they’re not actually taking decisions. So there’s no reason why everything can’t happen automatically, based on learned policies. (That’ll even reduce the chances of error, and it’ll all be properly trackable and measurable).

Dealing with high volumes of tickets

Naturally, if you’re a smaller business where the workload on the helpdesk is relatively light there’s less to be gained from automation. But the benefits kick in pretty quickly as the volume of requests escalates.

This has been perfectly demonstrated by our customer experience so far, and sheer scalability would also be particularly relevant to outsourced IT service providers who run helpdesks on behalf of clients. One service provider might work for hundreds of separate businesses, all of which have different requirements, policies and demands. One of the areas in which such service providers could show added value would be by demonstrating continuous improvement and methods for improving process management.

Automation is the key, because it both helps deal with each ticket, creates new processes to deal with future tickets, and provides reliable, meaningful measurements that show each customer how great their service provider is.

Want to show how great your helpdesk is? Invest in machine learning and artificial intelligence, and start measuring things that really matter.