Best practices for monitoring system health

Mindful's dashboards and reporting features provide insights into system health that administrators can monitor at any time. By monitoring system health weekly, daily, or even throughout the day, you can identify potential issues as soon as possible, giving you time to investigate and resolve issues before they become noticeable to customers. This article provides guidelines and a checklist for administrators tasked with monitoring your Callback Organization.

Overview

This article covers the following topics to help you monitor the efficiency of your Organization and overall system health:

  • High-priority alerts
  • Real-time system status
  • Daily metrics
  • Notable call events
  • System monitoring checklist

High-Priority Alerts

We recommend designating at least one administrator on your team to receive system alert emails and take action as needed when an alert is triggered.

All system alerts in Mindful Callback are important and should be addressed as much as possible. However, the Disconnect Rescue and Priority Queue Timed Out alerts can be especially critical since they indicate that agents are having trouble connecting with customers. If you see one or both of these alerts triggered frequently, we recommend investigating further to determine the potential cause and identify opportunities for improvement.

See the following articles to learn more about these alerts and how to address the underlying issues:

To learn more about any of the alerts that you may receive, see the Alerts reference article.

Realtime System Status

The Callback Status page is designed for real-time system health monitoring, and it should be checked by an administrator throughout the day. There are a few key pieces of information on this page that can provide insight into the interaction between Mindful Callback and your telephony environment. Scroll down until you see the chart:screenshot of the callback status table with numbered fields

  1. Current ECBT: Monitor the Current ECBT column for each Call Target to determine if it is similar to the holding queue's wait time shown in your ACD. Don't expect the two values to match exactly, but if you see large discrepancies arise, it may be worth investigating.
  2. Calls Waitlisted: Keep an eye on this value to determine if it is manageable based on the time of day. Calls Waitlisted represents the number of callbacks that are ready to launch but must wait for queue conditions to change first (based on the Max Queue Depth and Max Active Calls settings).
  3. Calls in Queue/Max: Monitor the Calls in Queue value to ensure that calls are consistently being sent to the high-priority holding queue. If Calls in Queue is consistently reaching the Max value, it could indicate that dial pacing is being slowed down more than your agents need. In this case, it could be useful to increase the Max Queue Depth setting to allow more calls into the high-priority queue at once.

Mindful Callback does not have visibility into your holding queues, agent staffing, or ACD wait time for the purpose of dial pacing. Callback only has information about the calls it has sent to the call center and the status of those calls. Thus, it is important for an administrator to analyze the volume of callbacks being sent to the call center to determine whether the dial pacing is appropriate.

To learn more, see How to Impact the Dial Pacing of Callbacks.

Executive Summary Dashboard

Quick access: Insights > Callback Executive Summary

On the Callback Executive Summary page, monitoring the percentage of reconnected callbacks compared to abandoned callbacks can provide insights into how reliably the system is connecting customers with agents. Pay special attention to the statistics and tips below to maximize the number of successfully reconnected callbacks.screenshot of the aggregate outcomes graph with the reconnected section highlighted

  • Aggregate Outcomes > Percentage of reconnected callbacks: Take a look at the Reconnected section of the Aggregate Outcomes graph and the Successfully Reconnected column in the Stats by Call Target section. If you find the percentage of reconnected callbacks dropping, or if the percentage is lower than you would like, there are a few key areas you can investigate to identify opportunities to improve:
    • Voice prompts: Listen to the callback voice prompts to determine whether the instructions are confusing or unclear. If the voice prompts are unclear, customers may not understand how to accept a callback when it arrives.
    • Punctuality vs. the spoken ECBT: Compare quoted ECBT to the actual callback times to evaluate the expectations being set for ECBT. If the spoken ECBT and the actual callback times vary too much, customers may not be expecting their callback when it arrives.
    • Customer behavior: Consider whether the affected Call Targets serve a group of customers who are likely to visit your website, search online, or call into a different queue before their callback arrives. If so, they may be finding solutions on their own and declining the callback when it arrives.
  • Percentage of abandoned callbacks: Look at the Abandoned statistic in the Reconnected section of the Aggregate Outcomes graph. If the percentage is growing or is higher than you expect, then consider the Average RCHT statistic. If the Return Call Hold Time (RCHT) is long for a Call Target, that can be a potential cause of abandoned callbacks. To address this, consider slowing down the dial pacing on the affected Call Targets by adjusting the Max Queue Depth or Max Active Calls settings on the Call Target page.

Callback Status Page

The Call Volume Timeline on the Callback Status page can provide a good indicator of overall system health by revealing how stable or unstable the volume of incoming and outgoing calls is throughout the day.

In a healthy and efficient system, you can expect to see the Calls Processed (purple) bars remain more steady than the Calls Inbound (blue) bars. Calls Inbound represents inbound traffic, which is not controllable by Mindful Callback. You can expect to see Calls Inbound vary based on your call center volume. However, Calls Processed is governed by the dial pacing capabilities of the Callback application, via the Max Queue Depth and Max Active Calls settings. Because of this, Calls Processed should show a more consistent value over time.image of the call volume timeline graph

If you find that the Calls Processed bars vary widely, it is possible that you are not utilizing the full capacity of your agents to handle callbacks. In this case, you might increase the efficiency of the system by speeding up the dial pacing by a small amount and monitoring the results.

Notable Call Events

Events on the Call Detail page do not need to be part of a daily system monitoring routine, but we recommend paying attention to these events after the initial deployment of your Callback Organization, after making significant configuration changes, or while investigating issues with your Call Targets.

Using the Event Filter

To search for potential issues, use the Event Filter on the Call Detail page to search for the following events:

Registration eventsPending eventsConnecting eventsFinal events
  • Chose Messaging Failed
  • Outside of Hours
  • End of Business Day
  • Max ECBT Exceeded
  • Max Scheduled Exceeded
  • SMS Timeout
  • Confirmation Notification Failed
  • Reminder Notification Failed
  • Priority Queue Timed Out
  • Rescued from Agent Disconnect
  • Early Hangup
  • Final Attempt Notification Failed
  • Customer/Agent Leg Error
  • Customer/Agent Busy
  • Customer/Agent Leg Rejected
  • Customer/Agent Leg Timeout
  • Success Notification Failed

Any of these events can indicate an issue worth investigating or help you to identify the source of an issue you are already investigating. To learn more about interpreting Call Detail events, see Understanding Call Detail Events and Outcomes.

Analyzing Final Stages

As you review the Call Detail data, take note of any calls that ended in the Registering, Pending, or Connecting stages. A normal and successful call should end in the Talking stage after the customer and agent have spoken. Any calls ending prior to the Talking stage could be worth investigating.example of the final stages in the call detail report

Take particular note of calls that ended in the Pending stage. This means that the call ended after the customer registered a callback, but before the callback was launched. This can be caused by administrators manually canceling callbacks or changing Call Target configuration in the middle of the business day.

System Monitoring Checklist

This section recaps the content from this article in a checklist for quick, daily use. Check each of the items below to fully monitor system health and efficiency at any time.

  • High-priority alerts
    • Disconnect Rescue
    • Priority Queue Timed Out
  • Real-time system status (Callback Status)
    • Current ECBT
    • Calls Waitlisted
    • Calls in Queue/Max
  • Daily metrics
    • Executive Summary
      • Percentage of reconnected callbacks
      • Percentage of abandoned callbacks
    • Callback Status
      • Call Volume Timeline
  • Notable call events
    • See the table in the previous section
    • Analyze end stages