Sponsored One of our favourite yarns concerns a grouchy Russian sysadmin who left his company. His colleagues checked out his computer and found a library of Bash scripts that automated his entire working life.
One of them scanned his inbox for mails from a hapless DBA called Kumar, looking for keywords such as "help" and "trouble". If it found them, it would SSH into the DBA's server, roll back his staging database to a backup, and send an email saying, "no worries mate, be careful next time."
Now that's efficient.
Most admins drowning in an ocean of support tickets probably dream of automation scripts like this. Their workload is increasing, and service desks are straining to cope with user requests.
Consequently, the manual workload is increasing for admins and support staff. In joint research with the Service Desk Institute, IT service management software firm Freshservice found 69 per cent of surveyed service desk professionals focused on fire-fighting due to heavy workloads and inefficient problem management processes.
Freshservice's 2017 State of the IT Service Desk report [PDF] polled 12,000 IT service desks from around the world. It found them using an average of 15 agents for every 1,500 employees, each handling 120 tickets at any one time. This led to an average ticket resolution time of ten hours. In other words, if the average employee reports a technical problem, they won't get a fix until the next day.
According to the 2018 Practices and Salary Report [PDF] from support and service consulting firm HDI, 79 per cent of surveyed companies are hiring analysts and technicians, and 47 per cent of them have to hit the ground running, with between one and five days training. Thirteen per cent get no training at all. Staff are difficult to retain, too; one in ten level-one support technicians quit entirely while another five per cent transfer out of the support department.
So, companies are struggling to find people to cope with user IT queries. Automated requests from software, and even tickets from devices, are making the problem even worse. In what must surely become a Black Mirror episode someday, a thermostat can now raise a ticket telling you that your server is overheating. Oh, joy. They won't even listen when you swear at them.
The answer, of course, is one that any BOFH would approve: tell users with a problem to go and look after it themselves. A self-service portal with automated capabilities can help to take common tasks off your plate by automating IT tasks like provisioning and configuration requests.
A self-service portal carries some internal benefits to the support team. It reduces the number of support tickets they’ll receive, taking a load off technicians. It also cuts costs, because it's cheaper to let customers resolve their own issues (in a controlled way) than to have the pros in IT spend their valuable time doing it.
There are benefits for the end-users, too. Automating key processes cuts resolution time, potentially to zero with some tasks. It allows for 24x7 service – even for Baz in Singapore who needs to expand his email inbox size when the team in London are off duty. It also improves IT-user relationships by reducing the need for them to actually talk to each other.
Freshservice predicted an average 25 per cent increase in service desk efficiency using automation. It also knocked ticket resolution times down by an average of 20 per cent.
So how do you build this portal to paradise? Look for low-hanging fruit, analysing historical support tickets to find the most commonly-occurring tasks that you can automate easily. An automated portal can manage tasks like tracking and status updates, storage and software provisioning, account creation, usage analytics, changing user roles, and, of course, that old chestnut: password resets.
The SaaS silver-lining with a cloud
Any automated helpdesk system must integrate with IT service management systems featuring workflow automation so that they can execute their intended tasks. You should capitalize on this as much as possible by integrating related tools so the automated systems can complete key workflows. For example, use IT service management (ITSM) software to update asset management or configuration management databases after successfully executing a task for a user, such as provisioning a software license or changing a configuration.
That still leaves you handling one of the trickiest IT service desk problems: cloud. For all the supposed advantages to the end user and the business, software-as-a-service (SaaS) can rain on your automated helpdesk parade because you won’t have full control.
Let’s take the Microsoft Office 365 example. You're an admin in a Microsoft shop, and you are quite naturally running Active Directory on-premises, which gives you a fine level of control over everything that happens in the tree. You would like to import users into an organizational unit (OU) and assign them access to Office 365, but Office 365 doesn't support OUs. Right there, in one fell swoop, you don't get the same administrative options and controls that you find in-house.
If the provider – in this case Microsoft – doesn't offer direct integration with your automation system, then you may find the bridge is a third-party, of which there are a few to pick. Among them is Quadrotech, founded the year Microsoft released the SaaS-version of its desktop suite and two years back made available a SaaS management application that runs in Azure Service Fabric and maps on-prem OUs and group policy objects to Office 365 tenants. The idea is, you have a means of automating your service desk tasks with Office 365.
Build it and they shall come
The problem with some service-desk requests is that human policies and politics get in the way. So, you could easily provision a software license to a user upon request, but that may cost money, which means their manager has to approve it. Until you can create an AI manager, you're going to have to deal with the human one, but you can still streamline the process.
The answer? Automation. In this particular example, you would automate the request to the manager using email/push notification approval. Your system can then monitor for a response and as soon as the manager sees the request and presses a button to give it the all clear, the ITSM back-end can continue servicing the request.
As you roll out this automation, you'll hopefully find users taking more advantage of the portal – and taking the strain off your support team. To realize your self-service portal's true potential, though, you need to make it as accessible as possible.
Automation in self-service is nothing without access, so it’s important to make sure your new system can be used from different types of devices given the diverse landscape of end points in corporate IT. According to Deloitte, 40 per cent of polled UK workers use a mobile (smartphone or tablet) device for their work, while 62 per cent use a desktop or laptop PC.
Naturally, the system won’t work of it own accord, so you’ll need to invest in inducting people into the new way of working. Avoid "what does this button do" syndrome, or helpdesk calls about the helpdesk portal interface, by offering resources training, and a solid set of FAQs and comprehensive support articles that use plain language and are easy to find.
Two magic words: Machine learning
If you’re feeling really adventurous there is also artificial intelligence. You can use an AI chat bot, which are now commercially available for service desk scenarios and that means you can avoid the need to build something from scratch. These have different capabilities. Some are AI workers that take basic requests like "I want to reset my password" and act on them while the more sophisticated versions can manage multi-step conversations with users to extract and service more complex requests.
Bots can respond in one of three ways. They can extract enough information to send the user to the right FAQ, effectively making them a snazzy search interface. Ideally, they would instead manage the entire user request themselves with no human interaction by kicking off an automated process. Alternatively, if they needed to escalate a query that was too complex, they could populate a helpdesk ticket for a support technician.
Robotic process automation (RPA) is making these bots more effective. By extracting semantic information from a request, they can then mimic human interactions for things like security verification for unlocking accounts, resource provisioning, or granting new permissions.
Building all of this isn’t a fire-and-forget project and you can, and should, keep refining your self-service portal by adding new tasks for automation and employing a greater number of back-end API integrations. You can – and must – also monitor user behaviour, just as you would on any website, to see what the staff are doing and what they have trouble achieving. Through this process you should gather metrics for analysis in order to see how the self-service portal is and isn’t working and how it can be improved.
What metrics should you look for? They include the number of visitors to the self-service portal (and the number of repeat visitors), what people accessed, and the bounce and abandonment rates. You should examine the automated request resolution rate versus the number of escalations (incidents where the user had to raise a ticket for manual interaction), monitor the use of the self-service portal against other service channels like telephone, email or chat, and you should run reports on the interactions between the service desk/ITSM software and other IT systems for automated tasks.
Finally, give staff with the chance to anonymously rate and comment on their experience, and to rate any knowledge bases or FAQs.
A self-service portal won’t, at the end of the day, have the charm of Kumar's database requests but will be more scalable and repeatable. And, importantly for you, it’ll help get your IT pros home on time each day with a reasonable blood pressure and leave you reasonably safe in the knowledge they are unlikely to quit thanks to a rising tide of ticket requests.
Sponsored by Quadrotech