We’re introducing Inferred CSAT, an AI workflow that automatically analyzes 100% of support tickets upon closure to estimate how customers might have rated their satisfaction with the service they received. This feature provides a new way to measure customer sentiment without relying on surveys.

How Inferred CSAT Works

The workflow reviews support interactions and determines the likely customer satisfaction level based on the following categories:

1. Satisfied
   The customer’s needs were met, and they appear content with the resolution.
2. Neutral
   The customer was indifferent or their satisfaction is unclear.
3. Dissatisfied
   The customer’s needs were not met, and they seem unhappy with the resolution.
4. Unknown
   There isn’t enough conversation data to assess satisfaction. This ensures accuracy by preventing guesses.

Measuring Inferred CSAT

Inferred CSAT calculates satisfaction as:
Satisfied / (Dissatisfied + Satisfied)

Addressing Dissatisfaction

When a ticket is categorized as Dissatisfied, the workflow runs an additional AI process to identify reasons for the dissatisfaction. These reasons are grouped and tracked over time, enabling organizations to visualize trends and address recurring issues.

Key Benefits

• Comprehensive Coverage: Measure customer satisfaction on every ticket.
• Actionable Insights: Understand what’s causing dissatisfaction and use those insights to improve service.
• Accurate Data: Ensure reliability by categorizing unclear cases as Unknown, avoiding speculative results.

Inferred CSAT helps businesses better understand and improve customer satisfaction by providing detailed, actionable data on every ticket. Contact us to learn more and activate this workflow in your system.

We're thrilled to introduce the Product Contact Reasons workflow—a powerful AI-driven workflow designed to save you time and deliver organized actionable insights without needing to read support conversations. Whether you're:

• Sharing top product issues with your product team,
• Optimizing your bot or knowledge base by identifying common areas of customer confusion, or
• Streamlining workflows by addressing repetitive agent interventions,

this workflow cuts through the noise to surface the information you need, tailored to your unique goals.

How It Works

The Product Contact Reasons workflow uses AI to analyze ticket conversations upon closure, automatically categorizing them into one of four key categories:

1. Product Deficiencies
   Pinpoint bugs or product limitations that require escalation or prioritization. Use these insights to guide fixes and feature improvements.
2. Product Confusions
   Highlight instances where customers needed additional documentation or support. Empower your team to enhance self-service options and improve customer education.
3. Agent Interventions
   Identify cases where agents performed actions unavailable to customers. Discover opportunities to automate repetitive tasks and optimize workflows.
4. Uncategorized
   Tickets that do not fit into the other categories.

After categorization, the workflow activates a generative AI process to uncover specific details about the underlying issue—whether it's a deficiency, confusion, or intervention. Qvasa's generative workflows can reuse existing tags or create new ones, surfacing novel trends and insights from ticket content.

Why It Matters

This out-of-the-box workflow transforms raw support data into meaningful analytics, empowering you to:

• Share targeted, actionable reports with different teams.
• Identify and address trends before they become widespread issues.
• Save time by automating the categorization and tagging process.

Getting Started

To maximize accuracy, users can seed the system with known deficiencies, confusions, and interventions. Over time, this makes the AI smarter, delivering even greater value to your analytics.

Please contact the Qvasa team if you are interested in enabling this feature for your account.

We released two new metrics. 'Zendesk agent ticket message count' quick stat and basic chart. These show the total count of messages sent by agents. Both the quick stat and the basic chart can be filtered by the agent object to get the specific number of messages sent by an agent. 

We’re excited to announce a new permissions model which gives admins more granular control over user access and actions across the platform. This update replaces our previous "limited," "basic," and "admin" roles and is fully backward-compatible, so users retain all existing access.

We’ve added a set of basic chart widgets to our dashboards for tracking key ticket comment metrics over time. 

Here’s what’s included:

1. Public Agent Ticket Comment Count Over Time – Shows the total number of public agent comments.
2. Tickets with Public Agent Comments Over Time –  Shows counts of unique tickets with public agent comments.
3. Average Public Agent Ticket Comment Count Over Time – Shows the average number of public agent comments per ticket.
4. Average End User Ticket Comment Count Over Time – Shows the average number of end-user comments per ticket.
5. Average Comment First Touch Time Over Time – Shows time to the first agent comment.
6. Average Comment First Touch Time After Assignment Over Time – Shows the time from assignment to the first agent comment.
7. Average Comment Reply Time Over Time – Shows the average agent reply time on comments.
8. Average Median Comment Reply Time Over Time – Shows the average median agent response time after an end-user comment.
9. Average Total Comment End User Wait Time Over Time – Shows the total end-user wait time for agent all of the replies.

Note: For the reply time metrics, when analyzing agent performance, use the "Agents" filter and not any of the assignee ticket attribute filters. The reason for this is that the agents filter will restrict the data to just that agent's data, where as the assignee ticket attributes will restrict the ticket set to tickets an agent was assigned.

We’ve introduced a new feature that allows users to download a CSV of tickets directly from reporting widgets. For any widget with "inspect" capabilities, indicated by the magnifying glass icon in the upper right, you’ll now see a ticket icon as well. Clicking this icon downloads a CSV of the same set of tickets displayed in the inspect modal. This makes exporting and analyzing your ticket data outside the app faster and more convenient. Please note, the export is limited to the first 500 tickets for performance reasons.

We’ve added a new Quick Stat to Dynamic Dashboards. In the widget select lightbox, you can now choose the 'Tickets Entering Status Count' widget and filter by a specific ticket status, such as 'Solved.' This allows you to quickly see how many tickets entered a chosen status within a selected time period.

We’ve introduced new filtering capabilities in dynamic dashboards. Users can now exclude tickets based on specific attributes. Any filter listed in the "Ticket Attributes" section on the left of the filters light-box will now have an "Exclude" option directly below it, enabling you to easily filter out tickets that match a given attribute.

We’ve improved our alerting system by adding a fixed floor to percentage threshold alerts. This enhancement ensures you won’t receive alerts during low-volume periods unless ticket volume exceeds a set minimum, making your notifications more relevant and focused on meaningful changes in ticket volume.

How It Works: Users will see a new input in the alert condition form for "Percent threshold fixed floor". If added, alerts will not trigger unless ticket volume surpasses that minimum threshold AND the set percent threshold.

We’re pleased to announce the introduction of five new agent performance metrics in the configurable agent table, designed to provide deeper insights by removing outliers and offering a fairer representation of agent response times. These new metrics complement the existing ones—average messaging reply time, average comment reply time, and average comment first touch time after assignment—giving you more control and flexibility in how you assess performance.

* Zendesk Messaging Median Agent Reply Time By Agent and Zendesk Comments Median Reply Time By Agent are not backfilled and will populate from the time of this announcement forward.

Here are the new metrics:

1. Zendesk Messaging Median Avg Agent Reply Time By Agent: Each messaging chat has an average agent reply time. This metric calculates the median of those average reply times across all chats included in the query, providing a balanced view by minimizing the impact of outliers.
2. Zendesk Messaging Median Agent Reply Time By Agent: This metric captures the median reply time on each chat and then calculates the median of those medians, offering a refined perspective on agent performance across multiple messaging chats.
3. Zendesk Comments Median Avg Reply Time By Agent: For email-based ticket replies, this metric calculates the median of the average reply times across all ticket assignments for an agent, giving a more typical representation of their performance.
4. Zendesk Comments Median Reply Time By Agent: This metric takes the median reply time on each ticket for an agent and then calculates the median of those medians. It’s a powerful way to measure consistent performance by filtering out extreme cases.
5. Zendesk Comments Median First Touch Time After Assignment By Agent: This metric measures the time it takes for an agent to send their first email after being assigned a ticket. It calculates the median of all first touch times, helping you evaluate how quickly agents respond after being assigned new work.

These new metrics give you more options to assess performance, especially when you want to account for outliers and get a clearer, fairer view of how your agents are performing across messaging and email interactions.