Zia - Field Predictions
Better customer service that eventually results in long-term engagement and builds stronger relationships with the customers is an asset that every business wants to invest in.
While on the surface, customer support service may seem just about providing appropriate solutions within the stipulated time. There is a lot that happens in the backdrop which facilitates quicker resolutions. For example, support agents may not always be equipped with the information needed to resolve an issue on their own.
- They may have to seek approval from seniors to proceed.
- Share the ticket with other team members to get details.
- Get a confirmation on the solution from peers.
Sometimes, even then, the issue is not completely resolved due to technical or business limitations. At times, missing or inaccurate information in the tickets can also delay the resolution.
While keeping up customer expectations and satisfaction is challenging, an important measure that businesses can take to optimize the resolution time and give appropriate solutions is by having accurate and necessary information in the tickets during creation.
These details mentioned in the tickets play a crucial role in determining:
- who should be assigned the ticket
- what action and solution is needed
- how long would it take to resolve?
Zia's role in reducing resolution time and improving customer satisfaction
Support agents have multiple responsibilities and perform a spectrum of activities as part of their routine.
These activities take up a considerable amount of time and attention, that makes it challenging for the agents to ensure every detail is included in a ticket especially when they receive high volume of requests from various channels.
Entering accurate details during ticket creation is crucial because based on this information the tickets are segmented, severity and service cost is determined, and SLAs are implemented.
Zia's predictive ability can take the load off from the agents by skillfully analyzing the historical data and predicting the right value for various fields. Its learning, analytical, and reasoning abilities are constantly enhanced and get more accurate over time.
Let's find out how Zia's field predictions can help businesses enhance agent productivity and improve customer satisfaction.
Ticket segmentation and auto assigning them to the right agent
Every customer support team has a dedicated group of people that handle specific issues; this facilitates appropriate resolution within the stipulated time. Issues are often categorized based on,
- their severity
- problem or product type
- support tier
- subcategory
Zia can analyze the incoming tickets and predict values for the above fields based on its learning from past tickets that are present in the database. If Zia's prediction matches the accuracy score, then it will auto-update the predicted value in the respective fields.
This will automatically trigger the workflow rule (configured for the fields) and the tickets will be assigned to the respective teams or individuals. It saves time and rules out inappropriate handling of tickets ensuring timely action and quick ticket closure.
Identifying customer requirement from the ticket description
Customers raise support tickets to request a service or report an issue. A detailed account of the issue or requirement is usually provided in the ticket description.
Sometimes the subject of the ticket is generic and doesn't provide much information for eg.,"Urgent requirement", "Following-up on subscription issue" or "Transaction inquiry".
The support agents have to read through the descriptions carefully and identify the exact requirement. This can be time-consuming, especially when there are large volumes of tickets to be attended. Zia can help by analyzing the "subject" and "description" fields in a ticket and suggesting the exact values for fields such as "service requested" or "issue type" and so on.
Understanding field predictions
Zia is Zoho's AI-powered tool that harnesses data to unveil valuable business insights. By analyzing a vast amount of historical data and leveraging algorithms, Zia can identify patterns, detect trends, and make accurate predictions about future outcomes.
Field predictions can help agents by predicting and auto-updating the right values in different picklist fields such as ticket category, priority, issue type, etc. based on its analysis of existing tickets that have similar information in them. It can also predict the ticket owner based on its learning.
Zia can train using the data from the existing tickets in the Desk account and create a pattern, interpreting values for different fields in experimental tickets.
Points to remember
- Zia can predict only picklist values. It can predict both system and custom picklist field values.
- You must ensure that the picklist fields are added to the default layout because prediction can be executed only in the default layout.
- A department must have at least 500 tickets for Zia to train. It is recommended to have at least 500 tickets for each picklist value for Zia to predict effectively.
- Zia will predict based on the tickets that are available at the time of configuration. Tickets received after this period must be manually added to the prediction in order to retrain Zia.
Training Zia to analyze tickets and predict values
Zia can predict picklist values and also determine the ticket owner. Once field prediction is enabled in the Desk account, Zia begins its training using the existing tickets. During configuration, as you set the field(s) and the picklist values that you want to predict. Also, you can decide whether Zia should train only using specific tickets.
- Training using specific tickets - Consider you want Zia to classify the issue and predict if it's a bug fix, feature request, incident, or data loss, then it's recommended to use specific tickets for training. This will allow Zia to analyze relevant tickets and interpret the values based on the pattern.
- Training using all tickets - Likewise, if you want to assign tickets to a particular person and want Zia to predict the owner, then it's best to train Zia using all the tickets so that the training data is vast, versatile, and provides a wider scope to learn, analyze, interpret, and predict.
Determining prediction accuracy
Zia can learn and train using the tickets present in the database. It learns and creates a pattern on about 80% of total tickets and then evaluates its prediction abilities on the remaining tickets. Based on the result, it calculates the probability and an average accuracy score is estimated. The higher the score, the more accurate the prediction will be.
For example, in the image below, Zia correctly predicted "Sam" as the ticket owner for 484 records, giving a probability of 97%. The average accuracy score is then estimated as 100%.
Best practices to achieve a better accuracy score
Some recommendations that can help improve accuracy score.
- Set a criteria to train Zia only on specific tickets if you want explicit predictions. Eg., predict problem type for tickets from the IT sector.
- Give a wide range of training data to enhance Zia's learning ability. Eg., predict the product type.
- Retrain Zia using the latest tickets to append the prediction.
- Set the criteria as created time to train on specific data. Eg., predict issue type based on tickets that were received from January to June.
Flow of events during field prediction
Configuring field predictions
Zia's learning and predicting abilities can be utilized to improve and automate ticket management and scale agent productivity.
Creating a field prediction includes the following essential information:
Field that Zia should predict - Zia can predict values for both standard and custom fields. Currently, only picklist fields can be predicted. The prediction will be based on the existing tickets that had used similar values.
For example, if you want Zia to predict and suggest the type of issue from the following list: "Bug fix", "New feature", "Code error", and "Data Loss"; then the training data must have tickets from all these categories for Zia to make accurate prediction.
Values that Zia should predict - If you want a picklist field to be predicted, you can select the values that Zia should analyze and give prediction for.
For example, if you choose "Bug fix", "Code error", and "Data Loss", whenever a new ticket arrives, Zia will predict the "Issue type" and categorize them under these three values only. It will not suggest "New Feature" as a value. If the new ticket doesn't match the above values, then the field will remain un-predicted.
Prediction execution time - Depending on the business requirement, you can define when Zia should predict the fields.
- When the ticket is created - For example, to automate ticket profiling, the prediction must take place during ticket creation. In a new ticket, agents have to evaluate the "issue type" and the "service that is needed"; based on that information, the ticket should be assigned to an agent. Zia can automate this by predicting "Issue Type", "Service Type", and "Ticket owner".
- When the customer response is received on a ticket - For example, if the customer replies to an old ticket and you want to predict the "problem" to assign it to the right agent, then you can choose this option.
Accuracy score - Based on the probability, an accuracy score is auto-generated by the system. The higher the accuracy score, the more precise is the prediction result. Therefore, we recommend setting the accuracy score to 70 and above for reliable results.
Mode of field update - You can choose automatic or manual update of the predicted value. This will allow Zia to autofill the predicted value in the fields and set the subsequent actions in motion. However, in some cases the agent may have to verify the predicted value before entering. To facilitate this, manual verification of the value can be selected.
Note that Zia will auto-update the predicted value only if it meets the accuracy score that you set during configuration. For eg., if you set the accuracy score as 70% and the predicted value is only 65% accurate, then Zia will not auto-update the value.
Fields used for training - Businesses can allow focused training by selecting the fields that Zia uses for learning and training purposes.
For example, if you want Zia to predict only the tickets that require service from third-party vendors, then you can select the fields using which Zia should train itself.
- "Issue" contains "Replacement", "Technical issue", "Faulty product"
- "Service charges" is "Yes"
- "Ticket category" contains "Service needed"
- "Service cost" is "Greater than or equal to 10$
To create field predictions
- Go to Setup > General > Zia > Field Predictions.
- Click Create New Field Prediction.
- In What do you want to predict?, select a field from the drop-down.
- In Execute on, select Ticket creation, Customer reply or both.
- In Field Update mode, choose Auto-update predicted value or Manually confirm the predicted value.
- Select the Accuracy Score from the drop-down list.
- Select the fields that Zia can use for training.
- Click Save.
Enabling field prediction
Once field prediction is configured, Zia will take some time depending on the number of tickets present in the database to train and arrive at a prediction accuracy score. In the field prediction home page, the list of fields and their prediction accuracy score alongside the training status will be displayed.
Once the status is completed and the accuracy score meets your requirement, you can toggle the corresponding fields to enable field prediction.
Viewing tickets used for training
As per the configuration, Zia either uses all or selected tickets to train. The tickets that are used for training can be viewed under Ticket Annotation view. Field values can be changed to retrain ZIa.
To view tickets for training
- Go to Setup > General > Zia > Field Prediction.
- Click on the Predictor, from the list view.
- In the Ticket Annotation page, click the Trained Tickets tab.
- Select a field value.
The ticket will be added to the annotated list for retraining purpose.
Retraining Zia
As new tickets are added to the database everyday, Zia's learning should be enriched to ensure the prediction remains accurate. To achieve this, businesses must periodically retrain Zia with the new set of tickets. This will help them align the prediction results to the businesses and customers' evolving needs. Also, if you notice inaccuracy in prediction, then it's best to retrain the model with recent and most relevant tickets. Retraining involves the following steps:
- Adding tickets to the prediction
- Moving the ticket from unannotated to annotated
- Moving trained tickets to annotated
- Selecting annotated tickets for retraining
I. Adding tickets to the prediction
New tickets must be manually added to the prediction data. While doing so, the users can choose values for the fields that need to be predicted.
To add tickets to prediction
- From the Tickets module, select a ticket.
- In the Ticket detail page, hover on the ticket content and click the More icon.
- Click Add to Prediction.
- Select the values for the fields.
- Click Add.
The ticket will be added to the unannotated tickets list.
II. Moving tickets from unannotated list to annotated
Tickets that are added to the prediction data, without selecting a field value is listed under "Unannotated tickets". Likewise, any new ticket for which Zia failed to predict a value also goes under the unannotated list. Values can be selected for the fields, and then the tickets can be moved to the annotated list.
To move unannotated tickets to annotated
- Go to Field Predictions, and click on the desired Predictor from the list view.
You will be redirected to the ticket annotation page. - Click the Unannotated tickets tab.
- Select each ticket and choose a value for the predicted field.
Alternately, select tickets in bulk and update the fields. - Click Add to Annotated View.
III. Moving training data to annotated view
Tickets used for training purposes will be listed under trained tickets. Users can change the predicted values; this will add the ticket to the annotated view. From here you can use it to retrain.
IV. Selecting annotated tickets to retrain
From the annotated list, the field values can be changed individually or in bulk. Users can select all or specific tickets to retrain Zia. Tickets can be filtered based on
- Field values - The picklist values can be used to filter the tickets
- Ticket type - Tickets are categorized into two:
- Annotated - Unannotated tickets for which the user selected a field value is added to the annotated ticket list view.
- Overidden - Tickets for which Zia predicted a value, however the user replaced it with another are listed under this list view.
Tickets from both annotated and overridden categories can be used to retrain.
To retrain Zia
- Go to the Annotated tickets tab and select all or few tickets.
- Click Retrain Zia.
- In the popup, click Confirm.
Zoho CRM Training Programs
Learn how to use the best tools for sales force automation and better customer engagement from Zoho's implementation specialists.
Zoho DataPrep Personalized Demo
If you'd like a personalized walk-through of our data preparation tool, please request a demo and we'll be happy to show you how to get the best out of Zoho DataPrep.
- Zoho Workerly Home
- Forums
- Connect With Us:
- Zoho Recruit Home
- Forums
- Connect With Us:
You are currently viewing the help pages of Qntrl’s earlier version. Click here to view our latest version—Qntrl 3.0's help articles.
New to Zoho Survey?
Zoho Sheet Resources
Zoho Forms Resources
New to Zoho Sign?
Zoho Sign Resources
Sign, Paperless!
New to Zoho TeamInbox?
Zoho TeamInbox Resources
New to Zoho ZeptoMail?
Zoho DataPrep Resources
Design. Discuss. Deliver.
New to Zoho Workerly?
New to Zoho Recruit?
New to Zoho CRM?
New to Zoho Projects?
New to Zoho Sprints?
New to Zoho Assist?
New to Bigin?
Related Articles
Zia - An Overview
Zia is an AI-driven support assistant who makes your customer support team more effective through data mining and machine learning. She provides support agents with analyses, customer forecasting, alerts, and suggestions through a pre-configured ...FAQs: Zia powered by ChatGPT and ChatGPT for Zoho Desk extension
How is Zia's integration with ChatGPT useful for my business? The power of generative AI (ChatGPT) coupled with Zia's capabilities can bring a spectrum of benefits to the customer support processes in businesses Contextual responses to tickets: Zia ...Integration of Zia with OpenAI (ChatGPT) in Zoho Desk
Availability: Zia Powered by ChatGPT is accessible in Standard, Professional, and Enterprise editions. Currently, it is only opened for US, AU, and India data centers in order to comply with GDPR regulations. Customers located in countries outside of ...Creating Custom Lookup Fields in Zoho Desk
The ability to establish connections between different sets of data is crucial in any business. For example, in the hospitality industry, departments like sales, marketing, and bookings can derive essential insights from customer data if it is linked ...Translate Ticket Conversations with Zoho Translate
Zoho Translate (powered by Zia) is an extension that allows Zoho Desk users to translate the tickets to a language of their preference. Customers often raise tickets in their preferred language, which may not be comprehensible to the support agent ...
New to Zoho LandingPage?
Zoho LandingPage Resources