In this digital era that we live in, we’ve all heard or even experienced machine learning at some point. Machine learning (ML) is a type of Artificial Intelligence (AI) that uses algorithms to learn from the historical data that it is processing, and to predict new output values based on patterns.
ML in customer service processes is used to optimize the customer experience. It equips agents with more knowledge, for example through predictive analytics, and makes it possible for them to execute their work more efficient.
Powered by ML, SAP introduces Service Ticket Intelligence. This model learns from your historical ticket data and improves over time. With Service Ticket Intelligence, the system recognizes similar tickets, predicts ticket categorization and recommends solutions.
In this blog, I will focus on one of the functionalities that Service Ticket Intelligence covers: ticket categorization.
Ticket categorization increases agent productivity, better prioritizes incoming tickets, and automates classifications based on model accuracy.
To use ML, you require:
SAP Cloud for Customer – Enterprise license or Agent console Add-in license
Data volume – > 1000 categorized tickets
Data quality – Tickets include subject, description, proper category and priority
Note: this specific functionality is supported in the following languages: English, Chinese, French, German, Japanese, Portuguese, Russian and Spanish. Companies with a Dutch system language, can use Service Ticket Intelligence for:
Ticket NLP classification
Similar Ticket Recommendation
Ticket Time to completion
To activate Service Ticket Intelligence for automatic ticket classification, you’ll need to take the following steps:
Step 1: Scope Prediction Service in your Cloud for Customer solution
Scope Prediction Service in your change project and assign the Prediction Service work center to the relevant users.
Step 2: Quality check of the data
Check your historical ticket data to determine if you can use this data in the categorization model.
A useful tool to check if the system is ready, is the Readiness report. The report gives an overall status of the required readiness check factors. It provides insight in the minimum required value, the recommended value, and the actual value in your system (think of percentage of service categories used).
You can find this report under: Administrator > Prediction Service > Machine Learning Models > Model Setup.
Step 3: Add ticket categorization model
In the Model setup, click on the link to activate and train the model. Select Ticket Categorization and add the model. Enter a name and select ticket category levels you want to use in your model and click OK.
Step 4: Train your categorization model
Select Train to start the training and select Get Status to update the status of the training model.
Consider carefully which types you select for your prediction. In hierarchical catalogues, only the lowest level in the catalog hierarchy is predicted. When the catalog types in the structure are on one level, you can decide to include which types to use.
You can use multiple category catalogues for the categorization, but be aware that if a catalogue changes, you need to update your model. You can do this by downloading the category mapper file and update the spreadsheet. Next, you easily upload the category mapper file back into the system.
Step 5: Test and check model
Model Performance Reports can be used to check the model on its prediction accuracy and for training purposes. The reports show ML model performance with standard classification metrics.
You can use the Model test console to test your ticket categorization real time.
Step 6: Activate model and adjust settings
When the training is completed, you set the model status to Active. It is possible to adjust the settings to turn automated recommendation on or off (based on confidence percentage).
Your ML model is now ready for use! You can find category prediction results in your tickets. On the ticket object you can see a green circle that serves as indicator that your categories have been populated by your ML model.
Service Ticket Intelligence and its machine learning models help you improve your customer service processes. Incoming tickets are automatically classified and could serve as a basis for routing them to the right agent or team. This increases the efficiency of your services.
Want to know more about Service Ticket Intelligence? Contact us for more information!