Service level
Service level measures the percentage of conversations answered within a defined time threshold. It helps you evaluate how effectively your organization responds to customer interactions within a target timeframe.
How is service level measured?
Service level is measured based on how long a conversation waits in a queue before an agent answers it. The timer starts when the conversation begins, waiting in a queue. The timer stops when the conversation is answered. The total wait time is compared with the configured service level target. If the wait time is less than or equal to the target, the conversation meets the service level.
Service level calculation
Service level is calculated as:
Service Level = Conversations answered within the target time Total conversations included in the calculation
The number of conversations included in the calculation depends on your configured service level options. To set these service level options, see Configure contact center analytics options.
Service level options
You can use service level options to determine the conversation type to be included in the calculation. These options do not affect how wait time is measured; they only determine which conversations are included in the calculation of the final service level percentage.
You can enable or disable each of the following service level options independently:
- Include flowouts in the calculation: Flowouts are conversations removed from the queue by a flow action before being answered by an agent. When enabled, flowouts are included in the calculation and count as not meeting the service level target. When disabled, flowouts are excluded from the calculation.
Service Level calculated by: (tAnswered − nOverSla) / (tAnswered + tFlowOut) - Include short abandons in the calculation: Short abandons are conversations that disconnect within a defined short time threshold. When enabled, these conversations are included in the calculation. When disabled, they are excluded. This option helps you ignore brief interactions that do not represent meaningful customer intent.
Service Level calculated by: (tAnswered − nOverSla) / (tAnswered + tAbandon − tShortAbandon) - Include abandons in the calculation: Abandoned conversations are the conversations that disconnect before being answered. When enabled, abandoned conversations are included in the calculation and count as not meeting the service level target. When disabled, they are excluded from the calculation.
Service Level calculated by: (tAnswered − nOverSla) / (tAnswered + tAbandon) - Include Callback Virtual Wait in SLA: Determines whether callback conversations contribute to service level calculation based on Callback Virtual Wait. Callback Virtual Wait measures the time from when the callback offer enters the queue until a callback outcome is reached. This duration is compared against the configured SLA target instead of using the standard Answer event that applies to most other conversations. This setting applies to both Agent First Callback and Customer First Callback. When enabled, the service level is calculated as:
Service Level calculated by: (tCallbackConnect − nCallbackOverSla) / tCallbackConnect- Include callback abandons in calculation: Includes callbacks where the customer was reached, but the interaction ended before an agent was selected. When enabled, the service level is calculated as:
Service Level calculated by:(tCallbackConnect − nCallbackOverSla) / (tCallbackConnect + tCallbackAbandon) - Include unreachable callbacks in calculation: Includes callbacks where the customer could not be reached, including no answer, busy, invalid number, voicemail, or when the maximum number of callback attempts was reached. When enabled, the service level is calculated as:
Service Level calculated by: (tCallbackConnect − nCallbackOverSla) / (tCallbackConnect + tCallbackNotReached)
- Include callback abandons in calculation: Includes callbacks where the customer was reached, but the interaction ended before an agent was selected. When enabled, the service level is calculated as:
How the options work together
Service level can be understood as follows:
Service Level = Conversations that met the target Conversations included in the calculation
The numerator always represents conversations answered within the target time. The denominator changes depending on which options are enabled.
The following example illustrates how different conversation outcomes and service level options affect both inclusion in the calculation and whether each conversation meets the target, using a service level target of 20 seconds.
| Conversation | Outcome | Wait Time (seconds) | Meets Target | Included (example configuration) |
|---|---|---|---|---|
| A | Answered | 10 | Yes | Yes |
| B | Answered | 25 | No | Yes |
| C | Abandoned | 15 | No | Yes (on enabling abandons) |
| D | Abandoned | 3 | No | No (on excluding short abandons) |
| E | Flowout | — | No | Yes (on enabling flowouts) |
In the preceding example, only Conversation A meets the target. The total number of included conversations varies based on the enabled service level options.
Service level for callback conversations
By default, callback conversations use the same service level calculation as other conversations. The calculation uses the standard Answer event for the callback media type. For Agent First Callback, the Answer event occurs when an agent accepts the callback. Therefore, the default calculation provides a meaningful indication of whether the callback met the service level target.
For Customer First Callback, the callback media type does not have an Answer event because the system handles the callback offer directly. As a result, the callback leg does not have a meaningful "met target" signal for determining whether the service level target was met under the default calculation.
The Include Callback Virtual Wait in SLA option, described in the preceding Service level options section, addresses this difference. When you enable this option, "met target" and "included" signals for callback conversations for both Agent First Callback and Customer First Callback are based on Callback Virtual Wait instead of the standard Answer event.
Callback Virtual Wait measures the time from when the callback offer enters the queue until the callback reaches an agent. A callback meets the service level target when Callback Reached occurs before the configured SLA target elapses. Callbacks that exceed the SLA target are counted by the Callback Over SLA metric.
Default—Include Callback Virtual Wait in SLA off | Include Callback Virtual Wait in SLA on | |
What "met target" is based on | The standard Answer event on the callback media types | Callback Virtual Wait measured against Callback Reached |
What counts as an outcome | Answer, abandon, or flow-out, consistent with voice conversations | Callback Reached, Callback Not Reached, Callback Abandon, based on the callback lifecycle |
| Reflects | When a human agent (AFC) or the system (CFC) accepted the callback offer | When the customer reaches to an agent, providing an end-to-end measure of the callback wait |
The Include Callback Virtual Wait in SLA option is off by default to maintain backward compatibility. Existing service level reporting remains unchanged unless you enable this option. You can configure it with the other service level options. For more information, see Configure contact center analytics options. Enabling this option does not create a separate service level metric. Instead, it changes how callback conversations contribute to the existing service level calculation.
When you query service level across multiple media types, including voice and callback for the same queue, the numerator and denominator contributions from each media type are combined to calculate a single overall service level ratio. Callback conversations contribute to this calculation based on the service level configuration enabled for your organization. For more information about callback metrics and reporting, including the differences between Agent First Callback and Customer First Callback, see About callback reporting.
Service level and operating hours
Service level calculations do not consider queue operating schedules. Wait time is not paused when a queue is closed, and time spent waiting during closed hours is fully included in the total wait time.
For example, if a conversation enters the queue at 5:55 PM, the queue closes at 6:00 PM, and the conversation is answered at 9:00 AM the next day, the entire overnight period is included in the wait time. As a result, the conversation likely does not meet the service level target.
How service level is reported over time
Service level is reported based on when a conversation is answered, not when it enters the queue. Metrics are recorded at the time that the conversation is answered, and reporting intervals—such as hourly or daily views—group conversations using this timestamp.
For example, if a conversation enters the queue at 11:50 PM on Day 1 and is answered at 12:10 AM on Day 2, the service level impact is reflected in Day 2 reporting.
Aggregation and averages
Service level is calculated using totals rather than by averaging percentages across intervals. To determine service level over multiple intervals, add the total number of conversations that met the target and the total number of conversations included, then divide these totals. This approach ensures an accurate overall service level. For more information about aggregates and averages, see the developer center.
[NEXT] Was this article helpful?
Get user feedback about articles.