Prioritize tickets based on what the customer really thinks

Prioritize tickets based on what the customer really thinks

Using HP Service Manager 9.40 and HP IDOL OnDemand.

This article describes how I tested sentiment analysis and cloud services as a complement to ticket priorization.

In many scenarios HP Service Manager (HPSM) are integrated with an email solution which allows users to send e-mail to support (ticket is automatically created based on e-mail). Customers may also use Service Request Catalog (SRC) to register tickets.

Traditionally you use “Impact” and “Urgency” to calculate the ticket priority. If the user sends an e-mail this is often set to a default value and then later updated. If it is registered by using Service Request Catalog (SRC) the end user is often prompted to specify the “Urgency”. You might also use other available information like the affected service to adjust the priority.

If the user call to the Help Desk, the operator will register the ticket and of course make up an opinion whether the ticket should be handled immediately or if it could wait. However, your ticket queue will consist of many tickets and the “Urgency” may seem to “generic” to be handled in a desirable way.

Quite often I experience tickets which may be classified as low or medium while at the same time containing important and valuable information. It’s time to do something about that!

Since HP made “IDOL OnDemand” available I decided to try something new regarding urgency and ticket prioritization. “IDOL OnDemand” offers Cloud APIs and one of the APIs provide a “Sentiment Analysis”.

“The Sentiment Analysis API analyzes text to return the sentiment as positive, negative or neutral. It contains a dictionary of positive and negative words of different types, and defines patterns that describe how to combine these words to form positive and negative phrases.You can use sentiment analysis to gain valuable insights into what users, customers, friends, colleagues are saying.”

(Text and picture from https://www.idolondemand.com)

I decided to give “IDOL OnDemand” a spin with the demo data that comes with with HPSM 9.40.

Since HPSM supports writing Javascript (in HPSM Script Library) you have almost endless opportunities when it comes to writing custom scripts. Especially today when a lot of web pages are written as “client side” applications using Javascript which means that you can replicate a lot of that functionality in HPSM.

I wrote a script which is “triggered” when new tickets are registered in HPSM. The script creates a text which basically is a compilation of the “Title” and “Description” (but this is not a limitation, you may use whatever text you want) and asks “IDOL OnDemand” to perform a sentiment analysis. The results are then parsed by the script and stored in the ticket.

In the screenshot below you can see open Interactions and their respective sentiment result and score. Neutral sentiment results means that no “sentiments” were found in the analysis.

The sentiment score will range from 0 til -100 (most) negativity or 0 to 100 (most) in positivity.

Using the new sentiment analysis to react to customer tickets (in time..?)
Consider the two tickets given in the screenshot below:

Which ticket would you have prioritized? Considering the traditional way of choosing tickets the ticket with a low sentiment score does actually have a higher urgency and will in many “views” be located at the top, while the other ticket might actually be at the bottom.

In both of these tickets the sentiment analysis pick up that a user is “not able” to do something. “Not able” is a sentiment with a negative score. But in the ticket with the highest urgency the sentiment analysis also pickup (from the title) that there is “lack of ink in printer” (another sentiment).

In this scenario I would have chosen to prioritize the ticket with the lowest urgency (thanks to the sentiment analysis). However, not everything is perfect while doing sentiment analysis. You might have picked up that the user also specifies that “there’s alway error”. This wasn’t parsed by the sentiment analysis as a sentiment. This is because there is a typo “alway”. I updated the ticket and changed it to “always” and this was picked up by the sentiment analysis which actually increased the sentiment score to -49!

Why prioritize tickets with positive sentiment?
The world isn’t only negative or neutral it’s also positive! You can use the sentiment analysis to discover positive tickets. Consider the ticket in the screenshot below:

This is a ticket which is classified as «request for information». In this imaginary scenario the user had previously registered a ticket asking for a new printer location which was “delivered” but had forgotten the name for the new location.

One might discover that positive tickets are usually registered by “active” users in HPSM. Such users may often be key users and influencers in the “community”, that is,  if you keep those satisfied you may also keep a lot of other users happy.

Using sentiment analysis to push urgency
You could also use the sentiment analysis to set the urgency (and in turn priority) of the ticket. This is quite easy “script wise” and more advanced when it comes to data and rules. You could e.g. create rules that says “if urgency is 3 or higher and sentiment score is below -40 and -80 set urgency to 2”. It’s also possible to create a rule that identifies really negative sentiments and automatically sets the urgency to the highest.

What about Surveys?
Surveys are often used to gather data related to customer satisfaction, areas of improvement, issues and ensure that operators deliver what the customer expects.

I consider surveys as a reactive measure since they often are gathered after a ticket is solved and closed (and analysed even later).

While the sentiment analysis may also be a reactive measure it is however captured much earlier which gives the operator the ability to react and act while working with the ticket.

The best scenario is of course when the customer's service just work and there is no reason to register a ticket in the first place..

Christian Berg, Solution Architect, Manag-E

Manag-E
The leading HP Software partner in the Nordic area, with headquarters in Norway and a subsidiary in Sweden. Business includes licenses, consulting and support services for all IT Management tools in the HP BTO (Business Technology Optimization) portfolio from HP Software.

Visit Manag-E for more information!

 

Louie Gonis

Co-Owner &Director of Digital Marketing at Baby Lemonade Digital

8y

Very innovative. Great article

Like
Reply
Chuck Darst

Hike, Bike, and Ski Guide

8y

Sentiment analysis via Big Data to help prioritize tickets - very nice

Like
Reply
Oded Zilinsky

Vice President Of Product Management at Oversee

8y

Great to see that our partners are able to take our Big Data innovations for the Service Desk a step further. We really believe that big data will transform the way organizations are running their service desks.

Like
Reply
Bing Gu

Product Manager at Business Operation Technologies PTE. LTD.

8y

Good approach

Like
Reply

To view or add a comment, sign in

Insights from the community

Explore topics