Monitoring your Amazon Connect instance using CloudWatch - Amazon Connect

Monitoring your Amazon Connect instance using CloudWatch

Amazon Connect sends data about your instance to CloudWatch metrics so that you can collect, view, and analyze CloudWatch metrics for your Amazon Connect virtual contact center. You can use this data to monitor key operational metrics and set up alarms. Data about your contact center is sent to CloudWatch every 1 minute.

When you view the CloudWatch metrics dashboard, you can specify the refresh interval for the data displayed. The values displayed in the dashboard reflect the values for the refresh interval you define. For example, if you set the refresh interval to 1 minute, the values displayed are for a minute period. You can select a refresh interval of 10 seconds, but Amazon Connect does not send data more often than every 1 minute. Metrics that are sent to CloudWatch are available for two weeks, and then discarded. To learn more about metrics in CloudWatch, see the Amazon CloudWatch User Guide.

Note

If your Amazon Connect instance was created on or before October 2018, you need to provide Amazon Connect with permission to begin publishing chat metrics to your CloudWatch account. To do so, create an IAM policy with the following permission and attach it to the Amazon Connect service role. You can find the Amazon Connect service role on the Account overview page for your Amazon Connect instance.

{ "Effect": "Allow", "Action": "cloudwatch:PutMetricData", "Resource": "*", "Condition": { "StringEquals": { "cloudwatch:namespace": "AWS/Connect" } } }

Amazon Connect metrics sent to CloudWatch

The AWS/Connect namespace includes the following metrics.

Metric Description
CallsBreachingConcurrencyQuota

The total number of voice calls that exceeded the concurrent calls quota for the instance. For the total number of calls that breach the quota, take a look at the Sum statistic.

For example, assume your contact center experiences the following volumes, and your service quota is 100 concurrent calls:

  • 0:00 : 125 concurrent calls. This is 25 over the quota.

  • 0:04 : 135 concurrent calls. This is 35 over the quota.

  • 0:10 : 150 concurrent calls. This is 50 over the quota.

CallsBreachingConcurrencyQuota = 110: the total number of voice calls that exceeded the quota between 0:00 and 0:10.

Unit: Count

Dimension:

  • InstanceId: The ID of your instance

  • MetricGroup: VoiceCalls

CallBackNotDialableNumber

The number of times a queued callback to a customer could not be dialed because the customer's number is in a country for which outbound calls are not allowed for the instance. The countries allowed for an instance are defined by the service quotas.

Unit: Count

Dimensions:

  • InstanceId The ID of your instance

  • MetricGroup: ContactFlow

  • ContactFlowName: The name of your flow

CallRecordingUploadError

The number of call recordings that failed to upload to the Amazon S3 bucket configured for your instance. This is the bucket specified in Data Storage > Call Recordings settings for the instance.

Unit: Count

Dimensions:

  • InstanceId: The ID of your instance

  • MetricGroup: CallRecordings

CallsPerInterval

The number of voice calls, both inbound and outbound, received or placed per second in the instance.

Unit: Count

Dimensions:

  • InstanceId: The ID of your instance

  • MetricGroup: VoiceCalls

ChatsBreachingActiveChatQuota

The total number of valid requests made to start a chat that exceeded the concurrent active chats quota for the instance. For the total number of chats requests that breach the quota, take a look at the Sum statistic.

For example, assume your contact center experiences the following volumes, and your service quota is 2500 concurrent active chats:

  • 0:00 : 2525 concurrent active chats. This is 25 over the quota.

  • 0:04 : 2535 concurrent active chats. This is 35 over the quota.

  • 0:10 : 2550 concurrent active chats. This is 50 over the quota.

ChatsBreachingActiveChatsQuota = 110: the total number of chats that exceeded the quota between 0:00 and 0:10.

Unit: Count

Dimensions:

  • InstanceId: The ID of your instance

  • MetricGroup: Chats

ConcurrentActiveChats

The number of concurrent active chats in the instance at the time the data is displayed in the dashboard. The value displayed for this metric is the number of concurrent active chats at the time the dashboard is displayed, and not a sum for the entire interval of the refresh interval set. All active chats are included, not only active chats that are connected to agents.

While all statistics are available in CloudWatch for concurrent active chats, you might be most interested in looking at the Maximum/Average statistic. The Sum statistic isn't as useful here.

Unit: Count

Dimensions:

  • InstanceId: The ID of your instance

  • MetricGroup: Chats

ConcurrentActiveChatsPercentage

The percentage of the concurrent active chats service quota used in the instance. This is calculated by:

  • ConcurrentActiveChats / ConfiguredConcurrentActiveChatsLimit

Where ConfiguredConcurrentActiveChatsLimit is the Concurrent active chats per instance configured for your instance.

Unit: Percent (Output displays as an integer. For example, 1% of chats is shown as 1, not as 0.01.)

Dimensions:

  • InstanceId: The ID of your instance

  • MetricGroup: Chats

ConcurrentCalls

The number of concurrent active voice calls in the instance at the time the data is displayed in the dashboard. The value displayed for this metric is the number of concurrent active calls at the time the dashboard is displayed, and not a sum for the entire interval of the refresh interval set. All active voice calls are included, not only active calls that are connected to agents.

While all statistics are available in CloudWatch for concurrent voice calls you might be most interested in looking at the Maximum/Average statistic. The Sum statistic isn't as useful here.

Unit: Count

Dimensions:

  • InstanceId: The ID of your instance

  • MetricGroup: VoiceCalls

ConcurrentCallsPercentage

The percentage of the concurrent active voice calls service quota used in the instance. This is calculated by:

  • ConcurrentCalls / ConfiguredConcurrentCallsLimit

Unit: Percent (output displays as a decimal)

Dimensions:

  • InstanceId: The ID of your instance

  • MetricGroup: VoiceCalls

ConcurrentEmails
Note

Data for ConcurrentEmails is sent to CloudWatch every 5 minutes.

The number of concurrent active emails in the instance at the time the data is displayed in the dashboard. The value displayed for this metric is the number of concurrent active emails at the time the dashboard is displayed, and not a sum for the entire interval of the refresh interval set. All active email are included, not only active emails that are connected to agents.

While all statistics are available in CloudWatch for concurrent emails you might be most interested in looking at the Maximum/Average statistic. The Sum statistic isn't as useful here.

Unit: Count

Dimensions:

  • InstanceId: The ID of your instance

  • MetricGroup: Email

ConcurrentEmailsPercentage
Note

Data for ConcurrentEmailsPercentage is sent to CloudWatch every 5 minutes.

The percentage of concurrent active emails service quota used in the instance. This is calculated by:

ConcurrentEmails / ConfiguredConcurrentEmailsLimit

Where ConfiguredConcurrentEmailsLimit is the Concurrent emails per instance configured for your instance.

While all statistics are available in CloudWatch for concurrent emails you might be most interested in looking at the Maximum/Average statistic. The Sum statistic isn't as useful here.

Unit: Percentag

Dimensions:

  • InstanceId: The ID of your instance

  • MetricGroup: Email

ConcurrentTasks
Note

Data for ConcurrentTasks is sent to CloudWatch every 5 minutes.

The number of concurrent active tasks in the instance at the time the data is displayed in the dashboard. The value displayed for this metric is the number of concurrent active tasks at the time the dashboard is displayed, and not a sum for the entire interval of the refresh interval set. All active tasks are included, not only active tasks that are connected to agents.

While all statistics are available in CloudWatch for concurrent tasks you might be most interested in looking at the Maximum/Average statistic. The Sum statistic isn't as useful here.

Unit: Count

Dimensions:

  • InstanceId: The ID of your instance

  • MetricGroup: Tasks

ConcurrentTasksPercentage
Note

Data for ConcurrentTasksPercentage is sent to CloudWatch every 5 minutes.

The percentage of the concurrent active tasks service quota used in the instance. This is calculated by:

  • ConcurrentTasks / ConfiguredConcurrentTasksLimit

Where ConfiguredConcurrentTasksLimit is the Concurrent tasks per instance configured for your instance.

Unit: Percentage

Dimensions:

  • InstanceId: The ID of your instance

  • MetricGroup: Tasks

ContactFlowErrors

The number of times the error branch for a flow was run.

Unit: Count

Dimensions:

  • InstanceId: The ID of your instance

  • MetricGroup: ContactFlow

  • ContactFlowName: The name of your flow

ContactFlowFatalErrors

The number of times a flow failed to execute due to a system error.

Unit: Count

Dimensions:

  • InstanceId: The ID of your instance

  • MetricGroup: ContactFlow

  • ContactFlowName: The name of your flow

LongestQueueWaitTime

The longest amount of time, in seconds, that a contact waited in a queue. This is the length of time a contact waited in a queue during the refresh interval selected in the CloudWatch dashboard, such as 1 minute or 5 minutes.

Unit: Seconds

Dimensions:

  • InstanceId: The ID of your instance

  • MetricGroup: Queue

  • QueueName: The name of your queue

MissedCalls

The number of voice calls that were missed by agents during the refresh interval selected, such as 1 minute or 5 minutes. A missed call is one that is not answered by an agent within 20 seconds.

To monitor the total missed calls in a given time period, take a look at the Sum statistic in CloudWatch.

Unit: Count

Dimensions:

  • InstanceId: The ID of your instance

  • MetricGroup: VoiceCalls

MisconfiguredPhoneNumbers

The number of calls that failed because the phone number is not associated with a flow.

Unit: Count

Dimensions:

  • InstanceId: The ID of your instance

  • MetricGroup: VoiceCalls

PublicSigningKeyUsage

The number of times a flow security key (public signing key) was used to encrypt customer input in a flow.

Unit: Count

Dimensions:

  • InstanceId: The ID of your instance

  • SigningKeyId: The ID of your signing key

QueueCapacityExceededError

The number of calls that were rejected because the queue was full.

Unit: Count

Dimensions:

  • InstanceId: The ID of your instance

  • MetricGroup: Queue

  • QueueName: The name of your queue

QueueSize

The number of contacts in the queue. The value reflects the number of contacts in the queue at the time the dashboard is accessed, not for the duration of the reporting interval.

Unit: Count

Dimensions:

  • InstanceId: The ID of your instance

  • MetricGroup: Queue

  • QueueName: The name of your queue

SuccessfulChatsPerInterval

The number of chats successfully started in the instance for the defined interval.

Unit: Count

Dimensions:

  • InstanceId: The ID of your instance

  • MetricGroup: Chats

TasksBreachingConcurrencyQuota

The total number of tasks that exceeded the concurrent tasks quota for the instance. For the total number of tasks that breach the quota, take a look at the Sum statistic.

For example, assume your contact center experiences the following volumes, and your service quota is 2500 concurrent tasks:

  • 0:00 : 2525 concurrent tasks. This is 25 over the quota.

  • 0:04 : 2535 concurrent tasks. This is 35 over the quota.

  • 0:10 : 2550 concurrent tasks. This is 50 over the quota.

TasksBreachingConcurrencyQuota = 110: the total number of tasks that exceeded the quota between 0:00 and 0:10.

Unit: Count

Dimensions:

  • InstanceId: The ID of your instance

  • MetricGroup: Tasks

TasksExpired

Tasks which have expired after being active for 7 days.

To monitor the total number of tasks that have expired in a given time period, take a look at the Sum statistic in CloudWatch.

Unit: Count

Dimensions:

  • InstanceId: The ID of your instance

  • MetricGroup: Tasks

  • ContactId: The ID of the task contact

TasksExpiryWarningReached

Tasks that have been active for 6 days 22 hours and reached expiry warning limit.

To monitor the total number of tasks that have reached expiry warning limit in a given time period, take a look at the Sum statistic in CloudWatch.

Unit: Count

Dimensions:

  • InstanceId: The ID of your instance

  • MetricGroup: Tasks

  • ContactId: The ID of the task contact

ThrottledCalls

The number of voice calls that were rejected because the rate of calls per second exceeded the maximum supported quota. To increase the supported rate of calls, request an increase in the service quota for concurrent active calls per instance.

To monitor the total throttled calls in a given time period, take a look at the Sum statistic in CloudWatch.

Unit: Seconds

Unit: Count

Dimensions:

  • InstanceId: The ID of your instance

  • MetricGroup: VoiceCalls

ToInstancePacketLossRate

The ratio of packet loss for calls in the instance, reported every 10 seconds. Each data point is between 0 and 100. The ratio of packet loss for calls in the instance appears as a percent between 0 and 1.

Unit: Percent

Dimensions:

  • Participant: Agent

  • Type of Connection: WebRTC

  • Instance ID: The ID of your instance

  • Stream Type: Voice

Amazon Connect CloudWatch metrics dimensions

In CloudWatch, a dimension is a name/value pair that uniquely identifies a metric. In the dashboard, metrics are grouped by dimension. When you view metrics in the dashboard, only metrics with data are displayed. If there is no activity during the refresh interval for which there is a metric, then no data from your instance is displayed in the dashboard.

The following dimensions are used in the CloudWatch dashboard for Amazon Connect metrics.

Flow metrics dimension

Note

Queue names must contain only the following characters in order for it to be displayed in CloudWatch: alphanumeric characters (0-9A-Za-z), period (.), hyphen (-), underscore (_), forward slash (/), hash (#), colon (:), comma (,), dollar ($), at sign (@), vertical bar (|), ampersand (&), opening brace ({), closing brace (}), plus (+), question mark (?), percent sign (%), and the space character.

Filters metric data by flow. Includes the following metrics:

  • ContactFlowErrors

  • ContactFlowFatalErrors

  • PublicSigningKeyUsage

Contact metrics dimension

Filters metric data by contacts. Includes the following metrics:

  • TasksExpiryWarningReached

  • TasksExpired

Instance metrics dimension

Filters meta data by instance. Includes the following metrics:

  • CallsBreachingConcurrencyQuota

  • CallsPerInterval

  • CallRecordingUploadError

  • ChatsBreachingActiveChatQuota

  • ConcurrentActiveChats

  • ConcurrentActiveChatsPercentage

  • ConcurrentCalls

  • ConcurrentCallsPercentage

  • ConcurrentTasks

  • ConcurrentTasksPercentage

  • MisconfiguredPhoneNumbers

  • MissedCalls

  • SuccessfulChatsPerInterval

  • TasksBreachingConcurrencyQuota

  • ThrottledCalls

Instance ID, Participant, Stream Type, Type of Connection

Filters metric data by connection. Includes the following metrics:

  • ToInstancePacketLossRate

Queue metrics dimension

Note

If a queue has a dimension name in non-ASCII characters, you won't be able to see it in CloudWatch.

Filters metric data by queue. Includes the following metrics:

  • CallBackNotDialableNumber

  • LongestQueueWaitTime

  • QueueCapacityExceededError

  • QueueSize

Amazon Connect Voice ID metrics sent to CloudWatch

The VoiceID namespace includes the following metrics.

RequestLatency

The elapsed time for the request.

Frequency: 1 minute

Unit: Milliseconds

Dimension: API

UserErrors

The number of Error counts due to bad requests from user.

Frequency: 1 minute

Unit: Count

Dimension: API

SystemErrors

The number of Error counts due to internal service error.

Frequency: 1 minute

Unit: Count

Dimension: API

Throttles

The number of requests that are rejected due to exceeding the max rate allowed for sending requests.

Frequency: 1 minute

Unit: Count

Dimension: API

ActiveSessions

The number of active sessions in the domain. Active sessions are sessions that are in pending or ongoing status.

Frequency: 1 minute

Unit: Count

Dimension: Domain

ActiveSpeakerEnrollmentJobs

The number of active Batch Enrollment Jobs in the domain. Active Jobs are those which are in Pending or InProgress status.

Frequency: 15 minutes

Unit: Count

Dimension: Domain

ActiveFraudsterRegistrationJobs

The number of active Batch Registration Jobs in the domain. Active Jobs are those which are in Pending or InProgress status.

Frequency: 15 minutes

Unit: Count

Dimension: Domain

Speakers

The number of Speakers in the domain.

Frequency: 15 minutes

Unit: Count

Dimension: Domain

Fraudsters

The number of Fraudsters in the domain.

Frequency: 15 minutes

Unit: Count

Dimension: Domain

Amazon Connect Voice ID metrics dimensions

The following dimensions are used in the CloudWatch dashboard for Amazon Connect Voice ID metrics. When you view metrics in the dashboard, only metrics with data are displayed. If there is no activity during the refresh interval for which there is a metric, then no data from your instance is displayed in the dashboard.

API metrics dimension

This dimension limits the data to one of the following Voice ID operations:

  • DeleteFraudster

  • EvaluateSession

  • ListSpeakers

  • DeleteSpeaker

  • OptOutSpeaker

Domain metrics dimension

The Voice ID domain where the enrollment, authentication or registration is conducted.

Amazon AppIntegrations metrics sent to CloudWatch

The AWS/AppIntegrations namespace includes the following metrics.

RecordsDownloaded

The number of records that were successfully downloaded as part of an AppFlow flow execution for a data integration.

Frequency: 1 minute

Unit: Count

Valid Statistics: Maximum, Sum, Minimum, Average

RecordsFailed

The number of records that failed to download as part of an AppFlow flow execution for a data integration.

Frequency: 1 minute

Unit: Count

Valid Statistics: Maximum, Sum, Minimum, Average

DataDownloaded

The number of bytes that were successfully downloaded as part of an AppFlow flow execution for a data integration.

Frequency: 1 minute

Unit: Bytes

Valid Statistics: Maximum, Sum, Minimum, Average

DataProcessingDuration

The time it took to process and download data as part of a single AppFlow flow execution for a data integration.

Frequency: 1 minute

Unit: Milliseconds

Valid Statistics: Maximum, Sum, Minimum, Average

EventsReceived

The number of events that were successfully emitted from your third-party source application (Salesforce, Zendesk) and received on your event bus.

Frequency: 1 minute

Unit: Count

Valid Statistics: Maximum, Sum, Minimum, Average

EventsProcessed

The number of events that were successfully processed and forwarded to be evaluated against the rules you configured on an event integration.

Frequency: 1 minute

Unit: Count

Valid Statistics: Maximum, Sum, Minimum, Average

EventsThrottled

The number of events that were throttled because the rate of emitting events exceeded the maximum supported quota.

Frequency: 1 minute

Unit: Bytes

Valid Statistics: Maximum, Sum, Minimum, Average

EventsFailed

The number of events that failed to process due to malformed or unsupported third-party events, and other processing errors .

Frequency: 1 minute

Unit: Bytes

Valid Statistics: Maximum, Sum, Minimum, Average

EventProcessingDuration

The time it took to successfully process and forward an event to be evaluated against the rules you configured on an event integration.

Frequency: 1 minute

Unit: Milliseconds

Valid Statistics: Maximum, Sum, Minimum, Average

Amazon AppIntegrations metric dimensions

You can use the following dimensions to refine AppIntegrations metrics.

Dimension Description
AccountId

AWS account ID

ClientId

Service principal of the client

IntegrationARN

ARN of the event or data integration

IntegrationType

DataIntegration or EventIntegration

Region

Region of the data or event integration

Amazon Connect Customer Profiles metrics

The AWS/CustomerProfiles namespace includes the following metrics.

Real-time export metrics sent to CloudWatch

The two following metrics will be published to CloudWatch for every export task. These metrics will provide information on your export stream tasks and will allow you to configure your Kinesis streams based on your use case. In the case of being throttled, these metrics will enable you to provision your Kinesis stream to ensure delivery to your destination.

EventsProcessed

Number of records successfully streamed into a Kinesis Stream.

Unit: Count

EventsThrottled

Number of PutRecord attempts that encountered throttling exception.

Unit: Count

Amazon Connect Customer Profiles metric dimensions

You can use the following dimensions to refine Customer Profiles metrics.

Dimension Description
DomainName

Customer Profiles domain name

DestinationType

Type of destination. Available value is: Kinesis

DestinationName

Name of destination. Kinesis Data Streaming name for DestinationType: Kinesis.

Use CloudWatch metrics to calculate concurrent call quota

Important

The ConcurrentCallsPercentage calculation information is not the same as ConcurrentTasksPercentage and ConcurrentChatPercentage.

  • The metrics emitted for ConcurrentCallsPercentage are in decimal and not multiplied by 100. The metric represents a percentage of your total quota.

  • For ConcurrentTasksPercentage and ConcurrentChatPercentage the value is multiplied by 100. That gives you your total quota.

  • The metrics emitted are correct and there is no discrepancy in data.

Here's how to calculate your quota usage for concurrent calls.

With calls active in the system, look at ConcurrentCalls and ConcurrentCallsPercentage. Calculate how much of your quota has been used:

  • (ConcurrentCalls / ConcurrentCallsPercentage)

For example, if ConcurrentCalls is 20 and ConcurrentCallsPercentage is 50, your quota usage is calculated as (20/0.5) = 40. Your total quota is 40 calls.

Use CloudWatch metrics to calculate concurrent active chats quota

Here's how to calculate your quota for concurrent active chats.

With chats active in the system, look at ConcurrentActiveChats and ConcurrentChatsPercentage. Calculate the quota:

  • (ConcurrentActiveChats / ConcurrentActiveChatsPercentage) * 100

For example, if ConcurrentActiveChats is 1000 and ConcurrentActiveChatsPercentage is 50, your quota is calculated as (1000/50)*100 = 2000. Your total quota is 2000 chats.

Use CloudWatch metrics to calculate concurrent task quota

Here's how to calculate your quota for concurrent tasks.

With tasks active in the system, look at ConcurrentTasks and ConcurrentTasksPercentage. Calculate the quota:

  • (ConcurrentTasks / ConcurrentTasksPercentage)*100

For example, if ConcurrentTasks is 20 and ConcurrentTasksPercentage is 50, your total quota is calculated as (20/50)*100= 40. Your total quota is 40 tasks.

Use CloudWatch metrics to calculate concurrent email quota

Here's how to calculate your quota for concurrent email.

With email active in the system, look at ConcurrentEmails and ConcurrentEmailsPercentage. Calculate the quota:

  • (ConcurrentEmails / ConcurrentEmailsPercentage)*100

For example, if ConcurrentEmails is 20 and ConcurrentEmailsPercentage is 50, your total quota is calculated as (20/50)*100= 40. Your total quota is 40 emails.