For ticketed activities, automation is integrated with the clients’ or enterprise’s ticketing system (ITSM) to reach the future state of IMS. An Artificial Intelligence (AI) engine triggers a work flow for standard incident or service request. Currently, available technology makes it possible to automate 30-40% of incident management tickets and 20-30% of service requests. However, by using AI-backed cognitive automation in conjunction with analytics, those figures can go up to 70%. The performance gains, both for systems and human resource, is non-linear.
The building blocks to achieve this boost in performance include:
- Sufficient volume of quality data related to tickets, monitoring and alerts
- An extensive and dependable knowledge base and SOPs
- Machine Learning
- Algorithms and models
The historical ticket data is used to train, test and validate automation bots that take over the task of managing tickets. The bots use the knowledge base, SOPs, algorithms and models to select the right processes and apply them quickly, accurately and autonomously. Machine Learning helps the bots improve through continuous learning, enriching the system’s capabilities with each iteration.
Non-ticketed activities are simpler to handle. Automation is built for fixed frequency intervention or through a dynamic threshold-adjusted frequency. The two methodologies are adequate for handling routing and health checks, alerting and reporting requirements, compliance related and housekeeping tasks.
Identifying Right Opportunities for Automation
What’s the best way to discover available opportunities for automation in IMS? Our suggestion is to keep it simple: Examine data and system logs as these are readily available, and use standard requirements elicitation techniques like interviews and focus group discussions. Also, some of the following should be useful in identifying automation opportunities:
- Existing tools, monitoring tools, event correlation, orchestrators and analytical engines
- Ticket data
- Run Book Automation (RBA) templates
Tools, system and ticket data along with team interviews throw considerable light on the possibilities for automation. Once identified, narrowed down and prioritized, the next step is to validate the plan and create a realistic estimation of effort and cost savings versus the investments required to develop and deploy the automation. It is a quite widely known that automation infrastructure has upfront costs related to tools, process re-engineering, testing and people. Often detecting and fixing automation glitches is several times more expensive than for manual processes. Organizations would do well to remember that automation has the best returns when applied to processes that have a long lifespan[i].
So, What's the Upshot?
Automation via AI, cognitive bots, policy engines, virtual agents, etc., typically leads to a reduction in human intervention in day-to-day IMS activities, leaving engineers to manage higher-value tasks that are critical to business. How many engineers get released from repetitive non-value added tasks is a factor of how completely (or partially) the automation has been implemented. But the direct impact of automation can be measured in:
- Effort and cost saving through left shift; activities that require L3 resources can be done by L2 in collaboration with bots and virtual agents
- Skill rationalization with existing engineers and IMS team members adding to long-term value such as enhancing capacity utilization instead of fire-fighting
- Standardization of processes across the organization making enterprise-wide integration and collaboration simpler
- Reduction in human errors and improved Quality of Services with the correct classification of tickets. This leads to right skill allocation and faster resolution
- Cycle time reduction and consistency in service delivery (bots, unlike human, don’t suffer from fatigue and hence their response never varies)
- Tighter service level management
The bottom line is that these gains accrue only by making astute choices for automation across IMS. It can be surprisingly easy to make poor decisions based on low hanging fruit or by mimicking peers for quick gains. On the other hand, all it takes is a thorough examination of opportunities before applying automation for long-term value.