With AI and machine learning technologies gaining traction in the IT support world, there is also a new term that is gaining traction in the lexicon of IT service and support. That term is Zero Touch. The commonly accepted definition of Zero Touch is that users can access IT support services without any intervention from a service desk agent. So, when a user has their incident or service request resolved without agent or technician intervention, that’s a Zero Touch transaction. The best metric to track Zero Touch is Agent Assisted Tickets per User per Month. The industry average for this metric is 1.1 Agent Assisted Tickets per User per Month. We can drive this number down in several ways, including: Problem Management, whereby agent resolved incidents are reduced by eliminating the underlying problems that drive incident volume. Self-help/self-service portals where users can access knowledge articles and resolve their own incidents without the assistance of an agent. Password reset tools where users can reset their own passwords without agent assistance. Chat and voice bots that resolve user incidents without human intervention. Endpoint bots that monitor the endpoints in the environment, and either self-correct/self-heal before an incident occurs and ticket(s) are created, or an alert is sent to IT so that they can take action to prevent incidents and hence tickets from being generated. Machine Learning AI tools that ingest the data streams from your ITSM, telephony, and other systems, and using big data algorithms will do one or more of the following: Detect anomalies in the IT environment, and then self-correct/self-heal, or send an alert to IT Intelligently route tickets to the right source of support at the right place at the right time Identify and eliminate root cause drivers of tickets (e.g., automated problem management) Automatically create or update knowledge articles (e.g., automated knowledge management) Automatically shift lift to reduce Total Cost of Ownership (TCO) In one well-documented case study, a global retailer was able to reduce their ticket volume by more than 50% in two years. Here is a summary of their raw data that illustrates the effect of their ticket reduction efforts: Over a 24 month period of time the Number of Agent Assisted Tickets per User per Month dropped from 1.11 to just 0.51. Here is what two graphical representations of this same data look like: You can see from these charts that machine learning/AI had the greatest impact on ticket reduction. A distant second place was problem management, followed closely by chat and voice bots. Finally, self-help/self-service had the least impact of the four ticket reduction strategies deployed by the company. You will also note that each strategy took time to mature. In the early months of this company’s Zero Touch initiative, the number of agent assisted tickets that were prevented or eliminated was fairly low. But as the months progressed and the technologies matured, the reduction in Agent Assisted Tickets per User per Month was quite substantial. Whatever your starting point for Agent Assisted Tickets per User per Month, this measure should be going down over time if you are making progress towards Zero Touch. A 25% reduction year-over-year is considered excellent progress. So, if you start with the industry average of 1.1 Agent Assisted Tickets per User per Month, a 25% reduction in one year would reduce that number to 0.825 Agent Assisted Tickets per User per Month (75% X 1.1 Tickets). If you continue this progress, the number of Agent Assisted Tickets would be further reduced to just 0.619 Tickets after two years (75% X 0.825 Tickets). And after three years the number of Agent Assisted Tickets would be reduced to just 0.464 Agent Assisted Tickets per User per Month (75% X 0.619 Tickets). No service desk ever gets to zero agent assisted tickets. So, Zero Touch, just like Zero Defects, is a worthy goal, but one that will never be fully realized. The key is to adopt metrics, processes, and technologies that enable you to steadily and continuously reduce the number of Agent Assisted Tickets per User per Month. The payoff for reducing this metric includes lower support costs (i.e., lower TCO), and improved user productivity, which can be quantified and monetized. In our case study above the cost of support at level 1 was slashed by more than 40% over two years, as automation, problem management, and self-help/self-service initiatives reduced the need for agent headcount in the service desk. As previously mentioned, the industry average for Agent Assisted Tickets per User per Month is 1.1. Top quartile for this metric (meaning good performance) is about 0.7 Agent Assisted Tickets per User per Month, and fourth quartile for this metric (meaning poor performance) is about 1.9 Agent Assisted Tickets per User per Month. I would encourage any support organization that is serious about maturing their problem management discipline, and/or deploying AI and machine learning technologies, chat and voice bots, and user self-help/self-service, to also begin tracking Agent Assisted Tickets per User per Month. The effective deployment of these strategies will have the effect of reducing your ticket volume over time, and that should be reflected in a steady decrease in Agent Assisted Tickets per User per Month.