Most managed service provider (MSP) service desks already have multiple layers of automation in place. From chatbots and self-service portals to AI routing and virtual agents, this tool is widely used. Some of them successfully resolve technical issues, ensuring that our skilled engineers don’t waste time on password resets or basic access requests.
However, problems typically begin when tickets are passed between multiple teams or arrive without the necessary context for troubleshooting.
While MSPs already track the usual metrics like ticket volume, average handle time, first call resolution, and technician workload, L1 teams can spend a significant amount of their day gathering context. Some support teams jokingly refer to this as the “20 Questions” phase. Who are the users? What devices are they using? Has anyone already touched the machine? Did the VPN fail before or after the update? Is this the correct column?
This often happens because support teams don’t have a clear understanding of what users are experiencing on their endpoints. Most MSPs today operate across ticketing, endpoint management, monitoring, remote access, and documentation tools, but these tools don’t always share information well. This leaves engineers jumping between systems, trying to piece together context that should already be in front of them.
According to our analysis This was recently announced at the Gartner Digital Workplace Summit London, where saving just one minute of context switching time on 10,000 monthly tickets equates to 166 hours of support capacity recovered.
This essentially amounts to full-time technicians switching tabs and rebuilding context, which is not a good use of skilled talent.
Escalation is where things really start to get expensive
In reality, many escalations simply start the troubleshooting process all over again. Support teams have different names for this. These include verification taxes, rediagnosis, and ticket ping pong. In some environments, AI-assisted triage allows tickets to arrive well-categorized, but lacks important context, making it difficult to actually discover this.
No one completely trusts notes attached to tickets, so the next technician repeats the same check. And to be honest, sometimes they have a point. By the third reassignment, half of the ticket notes have little meaning.
When senior engineers start getting caught up in tickets they never should have received, the costs quickly add up. The L2 team can spend 15-30 minutes reconfirming information that was confirmed earlier in the support chain. L3 engineers may still insist on verifying the root cause themselves before touching on anything important. If you leave some tickets, you basically remain on a doomed escalation path from the moment the initial diagnosis fails.
If the underlying ticket data and context are weak, automation can only accelerate incorrect routing decisions. Tickets end up in the wrong queue, bounce around between teams, or get escalated before anyone properly understands the issue in the first place. Some queues are essentially dumping grounds for improperly sorted tickets.
Why AI support struggles with unpredictable issues
Every MSP wants to enable users to handle simple tasks themselves, rather than flooding the queue with password resets and printer tickets.
The problem is that self-service tends to break down quickly when the problem is no longer easy. Someone reports that their “laptop is slow.” The system will suggest some common fixes. Nothing changes. Either way, the ticket will be placed in the wrong queue. Next, L1 technicians must start from scratch to determine whether the issue is Wi-Fi, memory usage, rogue updates, or background processes hogging the CPU.
Failed self-service can quickly become expensive. Failed self-service attempts can increase incident handling costs from approximately $6 to $53 with repeated escalations and troubleshooting.
Part of this problem stems from AI-driven support tools that are trained on historical ticket data and static workflows rather than real-time device information. After all, the actual support environment is constantly changing. The device will be removed from the policy. VPN issues occurred in one office but not in another.
This is where bad assumptions creep in. Even if the underlying diagnosis is incorrect, the ticket appears to be properly categorized. By the time the problem escalates, half of the support advisories are due to AI hallucinations. A skilled technician will recognize this immediately. However, this is usually not the case with other automations.
Reduce wasted effort at your service desk
Simply throwing more AI into your service desk will not solve the underlying diagnostic problem. In reality, technicians still spend far too much time piecing together context, re-running the same diagnostics, and trying to understand how a ticket that failed the first correct assignment ended up in the queue.
Smart MSPs start tackling the problem at its root. They are rebuilding their self-service portal so users can explain what the problem is in plain English, rather than forcing them to choose from confusing categories. After all, no one reports “DHCP lease failure”. They say the internet is down. L1 technicians also gain real-time visibility into endpoints, rather than relying on speckled ticket notes or whatever a user happens to mention.
Being able to instantly see failed updates, dropped VPNs, crashed services, or machines that are about to go down can save you a lot of time playing detective. But even with these improvements, most MSPs are still hitting the same walls. All important information resides in a separate system.
Ticketing, RMM, monitoring tools, and endpoint agents do not communicate effectively. So every time a ticket escalates, the next person basically has to start over. That’s why some better-run MSPs create a single shared record for endpoint health and diagnostic history. Having L1, L2, and L3 all check the same up-to-date facts eliminates endless rechecks and prevents tickets from getting stuck on a doomed escalation path.
At the end of the day, the value of automation and AI is determined by its foundation. If self-service keeps generating dirty tickets, escalating tickets too quickly, or keeps rebuilding each time you move through the queue, AI will only help you make the same mistakes faster. Teams that see real results are those that provide all support tiers with access to the same live endpoint context. That’s when AI really starts to show its power.
Topics
