---
id: 04-contact-channel-analytics
title: Contact Channel Analytics
---
import useBaseUrl from "@docusaurus/useBaseUrl";
In addition to the CTI adapter's native ability to provide direct
playback links to call recordings in Amazon Connect, the AWS Serverless
Application Repository for Salesforce includes several functions that
allow you to process recordings, perform quality analytics functions,
and bring data into Salesforce.
This processing is done post-call, using the Contact Trace Record (CTR)
as the initiation path. The following quality analytics options are
available:
- **Call Recording Streaming:** streams the actual audio file into
Salesforce. This option is not mandatory for the others to function.
- **Recording Transcript:** you can choose to have your call
recordings transcribed to text and presented in a visual format that
resembles a chat conversation. This allows for quick scanning of a
call to identify key segments of conversation. This option is
required if you wish to include the next level of analysis
- **AI-Driven Contact Analysis:** once the recordings have been
transcribed to text, you can also indicate that you wish to do
further analysis of the conversation using [Amazon
Comprehend](https://aws.amazon.com/comprehend/). Available options
are:
- **Sentiment Analysis:** returns the overall sentiment of the
conversation (Positive, Negative, Neutral, or Mixed).
- **Keyphrase Extraction:** returns the key phrases or talking
points and a confidence score to support that this is a key
phrase.
- **Language Detection:** returns the dominant language with a
confidence score to support that a language is dominant
- **Custom Entities:** allows you to customize the AI to identify
terms that are specific to your domain
- **Syntax Analysis:** analyze the transcript using tokenization
and Parts of Speech (PoS), and identify word boundaries and
labels like nouns and adjectives within the text.
If you would like to set up streaming with Contact Lens, please finish the
[Call Recording Streaming](#call-recording-streaming) section below and then
follow the [Contact Lens Streaming](docs/lightning/cti-adapter/07-contact-lens)
instructions and possibly the [Post Call Contact Lens Import](docs/lightning/salesforce-lambdas/06-postcall-contact-lens-import)
instructions.
### Call Recording Streaming
You can stream Call Recordings in your Salesforce Org. This
allows for easy access to the recordings from within Salesforce and can
be used in conjunction with the other contact channel analytics features
to provide a complete view of the customer interaction.
The import of call recordings is not required to activate the other
contact channel analytics features.
Once enabled during the AWS Serverless Application Repository for
Salesforce, recording import is activated on a call by call basis by
adding a specific contact attribute. This attribute is used during
Contact Trace Record processing to trigger the call import.
NOTE: After Call Work time is a part of the Contact Trace Record. As
such, CTRs are not generated until the agent leaves the after call work
state. If you are not seeing a recording import, please make sure the
agent has completed the call and left the after call work state.
##### Cloudformation Template
To make sure that the AWS resources are set up, make sure that the
_PostcallRecordingImportEnabled_ parameter is set to true in your
Cloudformation stack:
})
> **Note:** If you are expecting more than 1000 concurrent calls, you may have to increase the timeout for the `sfCTRTrigger` lambda.
#### Enabling call recording streaming
1. Login to your Amazon Connect instance as an Administrator
2. From the left navigation, choose **Routing** then select **Contact
flows**
})
3. Open the contact flow that you want to use to enable call recording
import. This contact flow must have Amazon Connect's native
recording turned on.
4. In you contact flow, before you transfer to queue, add a new **Set
contact attributes** block
5. Configure the block to set a contact attribute as follows:
a. **Destination key:** postcallRecordingImportEnabled
b. **Value:** true
})
6. **Save** the Set contact attributes block. Make sure it is
appropriately connected to your contact flow, and **Publish** the
flow.
7. Wait approximately 2 minutes to give the contact flow time to
publish.
8. Place a call, connect to your agent, speak for a few moments to test
the audio, then end the call. Make sure the agent exits after call
work
9. After a minute or so, a new Contact Channel Analytics record should be
imported, and when opening it, you should be able to stream the audio.
(See section [Adding Contact Channel Analytics to the Service Console](#adding-contact-channel-analytics-to-the-service-console). below).
#### Adding users to the AC_CallRecording permission set
This step is only necessary for non admin user accounts for the non contact lens case.
1. In the setup search box, search for "Permission sets". Select the "AC_CallRecording" permission set. Select "Manage Assignments".
})
2. Select "Add Assignments". Add the users that should have access to the audio recordings and select "assign".
})
#### Adding Contact Channel Analytics to the Service Console
1. Log in into your Salesforce org and go to the **Service Console**
2. Expand the **navigation menu** by selecting the down arrow and
choose **Edit**.
})
3. On the Edit Service Console App Navigation Items page, select **Add
More Items**
})
4. Select the **+** next to **AC Contact Channel Analytics**
5. Select **Add 1 Nav Item**
6. Change the order of your Navigation Items if desired, then choose
**Save**
})
7. Once the save completes, expand the **navigation menu** by selecting
the down arrow and choose **AC Contact Channel Analytics**
})
8. Change the list view from Recently Viewed to **All**
})
9. Once the view refreshes, you should see your record(s)
})
10. Select the recording to open it
11. In the top right, you will see a button to stream the recording.
})
12. NOTE: The recording playback, waveform, and transcript views are
only active when you also choose to activate recording transcripts.
### Recording Transcripts
Enabling the Recording Transcripts activates a process to run your
contact recordings through Amazon Transcribe which uses a deep learning
process to convert text to speech accurately and quickly. In addition,
this process also creates a visual waveform of the recording, enables
the in-app recording playback, and provides a visual representation of
the conversation.
Once enabled during the AWS Serverless Application Repository for
Salesforce, recording transcription is activated on a call by call basis
by adding a specific contact attribute. This attribute is used during
Contact Trace Record processing to trigger the transcription.
Make sure the Salesforce user accessing recording transcription are added to the AC_CallRecording permission set, as described in the previous section.
#### Enabling recording transcription
1. Login to your Amazon Connect instance as an Administrator
2. From the left navigation, choose **Routing** then select **Contact
flows**
3. Open the contact flow that you want to use to enable call
transcription. This contact flow must have Amazon Connect's native
recording turned on, since the transcription is dependent on it.
4. In you contact flow, before you transfer to queue, add a new **Set
contact attributes** block
5. Configure the block to set two contact attributes as follows:
1. Attribute 1: enables the transcription process
a. **Destination key**: postcallTranscribeEnabled
b. **Value**: true
2. Attribute 2: specifies the transcription language
-a. **Destination key**: postcallTranscribeLanguage
b. **Value**: en-US (See [Amazon Transcribe API
Reference](https://docs.aws.amazon.com/transcribe/latest/dg/API_StartTranscriptionJob.html#transcribe-StartTranscriptionJob-request-LanguageCode)
for valid language codes)
})
6. **Save** the Set contact attributes block. Make sure it is
appropriately connected to your contact flow, and **Publish** the
flow.
7. Wait approximately 2 minutes to give the contact flow time to
publish.
8. Place a call, connect to your agent, speak for a few moments from
both the agent and the customer side to generate a good transcript,
then end the call. Make sure the agent exits after call work
9. The transcription will take at least as long as the call did. Wait
an appropriate amount of time for the transcription to be available.
#### Accessing transcriptions
1. Log in into your Salesforce org and go to the **Service Console**
2. Expand the **navigation menu** by selecting the down arrow and
choose AC Contact Channel Analytics. If you have not previously
added AC Contact Channel Analytics to the navigation menu, complete
the steps found in
[Adding Contact Channel Analytics to the Service Console](#adding-contact-channel-analytics-to-the-service-console).
})
3. Change the list view from Recently Viewed to **All**
4. Once the view refreshes, you should see your record(s)
})
5. Select a record to view the details.
6. Once the record opens, note the recording, and the visual version of the transcription
})
7. Also note that the transcriptions for each side of the conversation
are also included as attachments.
### AI Driven Contact Analysis
Enabling the AI Driven Contact Analysis function allows you to process
the transcribed text using [Amazon
Comprehend](https://aws.amazon.com/comprehend/). Amazon Comprehend is a
natural language processing service that uses machine learning to find
insights and relationships in text.
Once enabled during the AWS Serverless Application Repository for
Salesforce, contact analysis is activated on a call by call basis by
adding a specific contact attribute. This attribute is used during
Contact Trace Record processing to trigger the Amazon Comprehend task.
There are five functions available with the integration. Each function
is triggered by a code. You can use one code in your contact attribute,
or string multiple together as a comma separated list. The available
codes and their functions are:
- **snt = Sentiment Analysis**
- **kw = Keyphrase Extraction**
- **dl = Language Detection**
- **ne = Custom Entities**
- **syn = Syntax Analysis**
#### Enabling AI Driven Contact Analysis
1. Login to your Amazon Connect instance as an Administrator
2. From the left navigation, choose **Routing** then select **Contact
flows**
3. Open the contact flow that you want to use to enable AI Driven
Contact Analytics. This contact flow must have Amazon Connect's
native recording turned on, and transcription enabled as these are
both prerequisites for the analytics function.
4. In you contact flow, before you transfer to queue, add a new **Set
contact attributes** block
5. Configure the block to set a contact attribute as follows:
a. **Destination key:** postcallTranscribeComprehendAnalysis
b. **Value:** snt,dl,kw,syn
- In this example, we are performing sentiment analysis,
language detection, and keyphrase
extraction
})
6. **Save** the Set contact attributes block. Make sure it is
appropriately connected to your contact flow, and **Publish** the
flow.
7. Wait approximately 2 minutes to give the contact flow time to
publish.
8. Place a call, connect to your agent, speak for a few moments from
both the agent and the customer side to generate a good transcript,
then end the call. Make sure the agent exits after call work
9. The contact analysis runs after the transcription, which will take
at least as long as the call did. Wait an appropriate amount of time
for the analysis to be available.
#### Accessing the AI Driven Contact Analysis
1. Log in into your Salesforce org and go to the **Service Console**
2. Expand the **navigation menu** by selecting the down arrow and
choose AC Contact Channel Analytics. If you have not previously
added AC Contact Channel Analytics to the navigation menu, complete
the steps found in
[Adding Contact Channel Analytics to the Service Console](#adding-contact-channel-analytics-to-the-service-console).
3. Change the list view from Recently Viewed to **All**
4. Once the view refreshes, you should see your record(s)
5. Select a record to view the details.
6. Once the record opens, note the Keywords, Sentiment, and Dominant
Language
})