Most large companies have IT Asset Management (ITAM) and IT Service Management (ITSM) systems that manage assets, which include not only hardware, software, networks and services, but also employee and customer data.

In any enterprise, service desk managers and CIOs need insight from data for making key decisions. They can manage resources better, analyze costs and trends efficiently, and improve overall customer service quality by tracking critical metrics like employee performance, service desk responsiveness, IT user preferences, customer satisfaction ratings, service-level agreement violation rates etc.

A leading Australian Telecommunications company, providing telecom, internet and digital services to individuals and large businesses, was facing issues with its ITAM/ITSM systems, leading to revenue loss and low customer satisfaction.

The Telco was using its ITAM systems to manage its daily B2B operations and customer support. Despite having a very large ITAM/ITSM support team, the issues raised by the customers (such as network connectivity, purchase orders, applications down-time etc.) were taking several weeks, even months, to resolve.

The need of the hour

The telecom company wanted to improve the efficiency of their ITAM/ITSM systems, address the issues raised by their customers quickly and ensure high Customer Satisfaction Score (CSAT) through controlled operations. They also wanted answers to what-if scenarios such as

• How many more resources will we need to reduce the lifetime of tickets from 2 weeks to 1 week?
• If we remove, let’s say 10 resources, how much will the ticket lifetime increase?
• If the number of tickets increases by 10%, what would be the impact on average time to resolve an issue?

The solution approach

Post research, we understood that the Telco’s service desk system involved very complex processes with violations of many business rules.

We performed preliminary analysis of the historical data, the issues raised by the customers, and the tickets’ lifetime and journey within the network.

We came up with different solution approaches that were machine learning (ML) based and simulation-based, among others. The ML-based approaches considered were to predict the next best action to assign the tickets to the most appropriate group so that the tickets travel through the shortest path to get resolved quickly.

The simulation-based approach mimicked the ITAM system and addressed what-if scenarios that the Telco wanted answers to. In simulations, we started with a question. Then we represented the system at a high-level abstraction, adequate to answer the question. We analyzed the raw data and extracted the input parameters such as arrival rate, service time distribution, resources distribution etc. for simulations, and then performed the simulations. Post this, the output was analyzed to answer the questions. All possible configurations were simulated under the given business constraints to find optimal solutions.

Technology used:

• Initially, we used AnyLogic (free personal learning edition) to simulate simple systems and create demos and animations for the senior management. Understanding the simulations can be complex for non-technical audience. 
• Later, we developed a code in Python for the simulation to find the optimal solution.
• We leveraged Discrete-Event Simulation method with Queuing theory.
• Power BI by Microsoft was used for easy consumption of the results from simulations and KPIs.

The impact

With the delivered solution, the telecom company was able to fix the issues with its ITAM/ITSM systems and ensure customer satisfaction.

Benefits included:

• Improvement in the working efficiency of ITAM/ITSM systems
• Reduction in time to resolve customer issues
• Optimization of the system resources and manpower
• Increase in utilization of manpower
• Understanding of the interdependencies among different teams/vendors, and reduction of the waiting times between the teams
• Revelation of several violations of business rules and malpractices
• Identification of bottlenecks and key pain points in the system
• A tool to simulate and answer what-if scenarios

Key Learnings

• Do not underestimate the challenges in data quality and data cleansing.
• Data alone is not sufficient to understand the business process.
• SME and domain knowledge is a must.
• Data does not always adhere to business rules.
• When the data/ticketing system doesn’t follow queue principles, it's not feasible to find whether it is a business principle (queue) violation or data quality issue.

Best practices

• When conveying complex models and methods to non-technical audience and senior management, use very simple toy models and animations.
• Do not confine to traditional established approaches.
• Always think about new and non-traditional approaches applied elsewhere.
• Find out what problems can be solved without big infrastructure, commercial licenses and investment.

Final thoughts

Simulation is a great approach to mimic a real-world scenario safely on a computer. They do not require big infrastructure: a couple of data scientists are enough to solve complex problems and a number of open source and free packages are available for simulations. Simulations require very low investment and the ROI is high. Surprisingly, despite its overwhelming usage, simulation has not got the attention it deserves.