Expert eye

IT Prediction Dashboards – Application of Machine Learning in IT Satisfaction Prediction

Andrzej Wawer

Artificial Intelligence and Machine Learning are becoming increasingly popular in business today. Companies use Artificial Intelligence to improve the effectiveness of business policy, stay ahead of competition, and improve their customer service. A report* by Capgemini shows that as many as 75% of companies that introduced AI and ML into their businesses have achieved a 10% increase in customer service satisfaction.

Many global corporations are experiencing increasing problems with end-user satisfaction regarding the work of IT teams designing all IT processes. What is the reason behind this? Is it the rapidly decreasing quality of support, defects and frequent breakdowns of servers, or rather a combination of several simple factors? The answer to this question is not clear, therefore solutions outside the box are becoming increasingly popular.

We live in a time when standard KPIs (key performance indicators) – calculated on a monthly, weekly or daily basis based on data warehouse – are already becoming insufficient. Companies still report in a way that does not focus on or even prevent actual user problems in real time. Performance indicators processes are being increasingly implemented with a higher frequency, e.g. in hourly intervals, which really helps with the ongoing monitoring of teamwork results. There is a possibility to activate alerts and send out all warnings regarding exceeding the permitted standards directly to all involved persons, yet the question remains: is it enough?

In my opinion, this is the simplest and the most menial way of fighting the process. In fact, we still do not have a full picture of the situation and what is really happening in a given moment. As architects of IT solutions, we want to improve the process itself, and not just “hunt” for individual tickets. The solution is developments in technology, which in this case are machine learning algorithms.

Dashboards made by Promity are able to follow a number of tickets and their statuses in real time. A manager may see how many people are working in hotline services or a company’s chat at a specific time, and additionally, may verify to whom a ticket was allocated, and which ticket was not dealt with and led to the escalation of the problem. This allows a company, particularly a big one, to detect critical situations on an ongoing basis and solve them as soon as possible.

Exemplary dashboard designed by Promity

Analysis and estimation in real time

Satisfaction surveys on IT work often “circulate” among employees of a corporation every quarter or twice a year. The main challenge in this case is to illustrate the real impact of updated KPI results and all available indicators on the results of future surveys. How can we define the most significant problems in such a case? How can we improve results gradually and in particular, mitigate the risk of their deterioration?

Such estimation is completely possible with machine learning. Decision-makers want to have access to both the updated IT satisfaction results and to their forecasts based on continuously incoming information. Algorithms analyze descriptions of reported incidents, making it possible to specify key groups of events from last few hours and be alerted to their immediate escalation, or to detect undesirable situations in contact with a support department. Such reactions have the desired effect immediately, which gives the solution to the problem.

Hooray! Reports that contribute to the business!

Machine learning is increasingly popular in the implementation of reporting projects because it enables things that so far have been perceived as unfeasible and can be used in solving actual problems in the IT world. The initial processes we have implemented for our customers have started to bring the expected results, although this solution is only taking its first steps in the industry and is being developed on an ongoing basis. Therefore, I would venture that in a few years machine learning algorithms will be present in the majority of designed dashboards.

In the future, people are still going to use dashboards, but they will be changed into smart analytical devices. They will deliver to decision-making employees not only raw KPIs, but also valuable insights and recommendations in real time. Presented in the form of intelligent alerts, they will enable employees to streamline processes in a better way, as well as implement business policies.