Enhancing Service Delivery with ServiceNow Predictive Intelligence

Written by Raja Mohsin Naeem

ServiceNow Developer

Predictive Intelligence for ITSM empowers agents/users with a layer of artificial intelligence across the ServiceNow platform, which helps to resolve tickets faster using different frameworks (Classification, Similarity, Clustering, Regression).

Customer Challenge

“Front Line workers have to deal with a similar type of ticket, again and again, like email not working or Password reset issue, even though they have a handsome amount of resolved tickets in their system and with good working catalog.”

Pain Point

Working on the same type of issues and searching for relevant resolution in the system was a repetitive task, wasting the time of frontend Workers. Manual categorization & assignments were also added more to it. A comprehensive solution was required to reduce the manual work to overcome the iterative tasks.

Proposed Solution

Royal Cyber, an implementation partner of ServiceNow, took the challenge and suggested implementing predictive intelligence in the environment to facilitate the frontend Workers, End Users, and Service Owners.

Predictive Intelligence

Predictive Intelligence is a product that helps Front end Workers resolve tickets faster with an automated approach. As a result, it helps to improve average resolution and customer satisfaction. ServiceNow provides four types of frameworks that could be used in a customer environment to achieve target goals and automation.

ServiceNow provides four type of frameworks which could be used in a customer environment to achieve target goals and automations.

  • Similarity framework: The similarity framework allows to train similarity solutions using the company’s historical ticket data. This framework can help agents and fulfillers quickly provide the best resolution for an incoming ticket based on a ticket’s short description.
  • Classification Framework: Classification framework could be used to train solutions to predict field values to categorize, Prioritize or Assign tickets to correct users. Any model would be trained based on given data from the customer environment.
  • Clustering Framework: Clustering Framework could be used to identify patterns from data. This could be used to recommend a resolution for a group of tickets or propose a major incident when a certain threshold meets.
  • Regression framework: Regression is a machine learning framework that could be used to predict numeric outputs like price, temperature, or resolution time.

How we started?

The implementation was divided into three major phases to facilitate the customer and accomplish the target.

  • Data Sanity

  • Predictive intelligence on Virtual Agent

  • Predictive Intelligence on Self Portal

Data Sanity

Machine Learning frameworks work on data, and then it uses the data to train models. Which are further used on a need basis to predict the values. If the model is trained with inaccurate data, the prediction would be inaccurate “garbage in and garbage out,” so the first step was to check data quality and improve it.

  • Last year, data were analyzed to check assignment, resolutions, categorization.
  • Any identified wrong assignment/categorization was fixed.

Empowering Virtual Agent with Predictive Intelligence

After a successful data sanity check, we incorporated virtual agents with machine learning powers.

Phase1: Training ML Solutions

We trained similarity and Classification solutions with historical resolved tickets to predict similar records and automatically categorize, assign, prioritize, and resolve. Similarity solutions were trained based on a short description of tickets.

Predictive Intelligence provides the ability to test solutions directly from trained versions, and our created solutions were tested and then fine-tuned to an 85 percent threshold, which resulted in good quality predictions.

Note: ML understands the context and synonyms to understand user intent and predicts/suggests accordingly.

Phase2: ML solutions for Knowledge Articles

The client had a good amount of knowledge articles, and we decided to use this data as a first arm to deal with issues. We trained similarity solutions to predict relevant knowledge articles in a published state. Knowledge articles were also fine-tuned, so article titles should match user issues.

Phase 3: Virtual Agent Topic

We created a virtual agent topic that invoked ML models based on user Issues. The topic was used as a chat starter and was built to take a short description of the issue as input and then find relevant articles and relevant resolutions.

Enhancing Self Service with Predictive Intelligence

After incorporating the virtual agent with a layer of artificial intelligence, the next phase was to leverage the catalog forms. Before creating the ticket, the user can get help from related KB Articles.

Phase1: Reinforcing catalog forms with ML

Using the similarity solution for knowledge base data, we added a field for end-users to see the best published related knowledge article as a suggestion directly on the catalog form. Suppose the system doesn’t find any related knowledge article. In that case, the field won’t show, and the system will try to search for any similar resolved ticket to find a similar resolution and show the user. This whole stuff was accomplished by invoking prediction models from client-side scripting.

Phase2: Auto population of fields with ML solutions

Once any ticket is logged in the system, we designed the solution to auto categorizes the tickets with previous ticket handling experience. Any work which was manually handled by agents, like assigning a ticket to the correct group & agent, was taken care of by ML classification solutions.

Following fields were auto populated after ticket creation

  • Category

  • Subcategory

  • Assignment Group

  • Assigned to

  • Business Service

What Was Accomplished

With the Predictive intelligence client has the following key benefits:

  • Virtual agents can predict relevant topics right away from the chat window
  • Virtual agents can provide the resolutions from historical data as a solution to user’s issue
  • Relevant Articles and similar resolution on catalog forms even before submitting the ticket
  • Auto population of fields based on historical data using ML classification framework
  • Reduced number of Tickets
  • Improved average resolution time
  • Enhanced Agent capacity so they can focus more on priority issues
  • 24/7 resolutions availability

Availability

Predictive intelligence empowers end users, front workers, and service owners. It can be incorporated into the following applications in the ServiceNow environment.

  • IT Service Management

  • Customer Service Management

  • HR Service Delivery

Conclusion

Nowadays, Predictive Intelligence cannot eliminate the need for human intervention. But it can sure free the IT agents from engaging in too much laborious and monotonous work and empower them to focus more on organization’s IT efforts. By embracing predictive intelligence soon, organizations have a great chance to put the ITSM process maturity to a higher level. With Royal Cyber you can realize the maximum value of your ITSM investments. If you are looking to implement Predictive Intelligence for your organization, you can email us at [email protected] or visit www.royalcyber.com.

Leave a Reply

Live Chat