Get value from every customer touchpoint using Amazon Connect as a data gathering mechanism

The current pandemic and the impossibility of meeting consumers in individual has actually made two-way contact focuses an effective tool for sales agents to reach to clients. Amazon Connect is the perfect service to manage these contacts, and its adoption provides you the opportunity to collect new service insights. Thanks to Amazon Connect, you can set outbound calls to reach to clients and build a video contact center to improve the client experience.
Amazon Connect provides an unique opportunity for collecting information from these engagements and customer touchpoints that helps improve your organization. With Amazon Connect, you can empower sales management with call transcriptions, sentiment analysis, recommendation systems, chatbots, combination with client relationship management (CRM) systems, and call note search, among others.
In this post, we walk you through the procedure of configuring an Amazon Connect two-way contact center to enable call recording and transcription. We explain three use cases that use this data to supply value: proprietary sentiment analysis, intelligent call note search, and suggestion systems. We likewise demonstrate how to construct your own applications and personalize these use cases.
Call note storage and transcription, sentiment analysis, and text search are available out of package from Connect Lens for Amazon Connect. For more details, see Real-time customer insights utilizing artificial intelligence with Contact Lens for Amazon Connect.
Solution overview
The following diagram shows the service architecture.

The circulation of this architecture is as follows:

A customer calls your call center in the cloud.
Amazon Connect links the consumer to an agent. As an alternative, the agent can begin an outbound call to reach a client.
When the call is total, Contact Lens starts transcribing the tape-recorded call and runs belief analysis on the records. It stores all artifacts in Amazon Simple Storage Service (Amazon S3) buckets.
As brand-new documents are conserved to the matching Amazon S3 areas, 2 AWS Lambda functions, one for chat the other for voice contacts, extract information of interest and write the wrangled information back to Amazon S3.
On a few of the information the Lambda functions stored, Amazon Kendra regularly updates a search index.
A similar scheduling concept is applied using an AWS Glue crawler.
The spider updates our AWS Glue Data Catalog, that makes it easy to search our terms, for example, with Amazon Athena.

Your Amazon Kendra data source is scheduled to upgrade itself every day at 10 AM. This method, your Amazon Kendra index keeps up to date. If you produced the solution parts after 10 AM, you can either wait up until the next automated sync, or trigger the synchronization of your data sources in your index via the Amazon Kendra console. For more details, see Using an Amazon S3 data source.
Deploy the information gathering resources
As an initial step, we want to deploy all resources, other than for the Amazon Connect circumstances, utilizing an AWS CloudFormation design template. You can do this by choosing Launch Stack:

Import the file contact-flow/contact-lens-transfer- circulation, offered in the GitHub repository.

Specify the name of your new S3 pail and task.

For this post, we have a predefined contact circulation design template that you can import.

For directions on importing contact circulations, see Import/Export contact streams.
The imported contact flow should look similar to the following.

On the Amazon Connect console, choose Analytics tools in the navigation pane.
Select Enable Contact Lens.
Choose Save.

Youre now all set to establish Amazon Connect and the associated contact flow.
Develop an Amazon Connect circumstances
The initial step is to develop an Amazon Connect circumstances. Make sure you use the S3 bucket you specified and produced in your CloudFormation design template when youre asked to provide an information storage place.
For the remainder of the setup, we utilize the default worths, however do not forget to produce an administrator login.
After the circumstances is developed, which can take a minute or more, we can log in to the Amazon Connect circumstances utilizing the admin account produced formerly. Were now ready to create our contact flow, declare a number, and connect the flow to that number.
Establish the contact circulation
Prior to we set up our contact flow, we need to make it possible for Contact Lens.

Comprehending the contact flow
The contact flow does the following:

SageMaker endpoints (if they were deployed).
CloudFormation stack.
Amazon Connect instance.

About the Authors.
Michael Wallner is a Global Data Scientist with AWS Professional Services and is enthusiastic about enabling clients on their AI/ML journey in the cloud to end up being AWSome. Having a deep interest in Amazon Connect, he delights in and likes sports cooking.
Andrea Di Simone is a Data Scientist in the Professional Services group based in Munich, Germany. He assists customers to establish their AI/ML items and workflows, leveraging AWS tools. He delights in reading, symphonic music and hiking.
Daniele Angelosante is a Senior Engagement Manager with AWS Professional Services. He is passionate about AI/ML projects and items. In his leisure time he likes coffee, sport, soccer, and baking.

# map from words to integers, developed at training time.
serializer.load _ vocab_to_tokens(./ meta/vocab _ to_token_dict. p).
# need to be the very same tokenizer utilized for training.
serializer.set _ tokenizer( word_tokenize).

Smart chat and call notes search by means of Amazon Kendra.
Amazon Connect transcribed calls and chat can are searchable by speaker, keywords, belief rating, and non-talk time. For an intelligent search, you can utilize Amazon Kendra, an intelligent search service powered by device learning (ML).
With Amazon Kendra, you can stop searching through chests of disorganized information and find the ideal answers to questions. Amazon Kendra is a fully handled service, so there are no servers to arrangement, and no ML designs to construct, train, or release. Amazon Kendra can be matched by Amazon Translate to allow multi-language search assistance.
Amazon Kendra is provisioned instantly with the CloudFormation template provided in this post, and you can utilize it to implement intelligent search of your calls.
Recommendation systems from call notes.
Sales representatives utilize call notes to capture meeting feedback and actions from an engagement with each customer. The call notes can either be taped or transcribed, after which theyre conserved in CRM solutions. The tape-recorded call notes might include insights like goals for the next consumer interaction, follow-up plan, and so on, there is minimal AI and ML involved (generally algorithms to discover individual information), so the sales representatives have to go back to the CRM service every time they want review the call notes to follow up on the discussion or the action points from the conference. This is necessary for producing continuous conversation with the customers, bridged across several engagements.
To further improve operational quality, you can rapidly incorporate and develop suggestion systems to examine the call notes in genuine time and offer instant feedback and notifies for the sales representatives on the suggested next finest action, which a sales representative can choose to dismiss or accept.
This option helps strengthen relationships utilizing AI and ML by offering a platform that enables sales agents to have more significant discussions with customers beyond the item itself, for instance to talk about clinical patterns, new publications, and tailored recommendations based upon specific customer requirements and locations of interest.
You can further improve the option with advanced AI and ML by enhancing the existing abilities to offer insights to sales representatives on how to optimize time and client engagement based on prior customer interactions and value-adding activities.
You can implement this using an Object2Vect design to categorize the call notes, comparable to what we demonstrated in the proprietary belief analysis usage case. The categories arent the different beliefs, but the next best action to advise after a given text. With this in mind, you can recycle the example notebooks and code provided earlier for this usage case.
After the design is trained, you can utilize it as an initial step in a 2nd design, which also takes into account non-textual functions as the ones originating from CRM systems (client account information, client section, consumer orders, and so on). The following diagram illustrates this architecture.

import pickle

response = predictor.predict( test_payload).

In this area, we explain how to utilize call and chat information collected by Amazon Connect in 3 use cases: proprietary sentiment analysis, intelligent chat and call notes browse through Amazon Kendra, and recommendation systems from call notes. You can also integrate these usage cases into your CRM (such as Salesforce or Zendesk) by using Amazon Connect integration features. This enables you to choose whether to use Contact Lens for consumer analysis or to embrace Amazon Connect as a data event mechanism and develop an information intake and reasoning pipeline to use your proprietary belief analysis design in AWS using Amazon SageMaker.
Amazon Kendra can be matched by Amazon Translate to allow multi-language search assistance.
We offered architectures and design templates for three usage cases that use the data gathered by Amazon Connect: proprietary belief analysis, smart search, and suggestion systems.

Conclusion.
In this post, we showed how to utilize Amazon Connect as an omnichannel data gathering system to gather data throughout consumer engagements such as chat and call recordings, notes, and transcriptions. We showed how to establish Amazon Connect to gather information from outbound calls. This crucial feature can make Amazon Connect the go-to service for sales management. We provided architectures and design templates for three use cases that use the data collected by Amazon Connect: exclusive belief analysis, intelligent search, and suggestion systems. Attempt it out today and let us understand what you believe in the comments!

You may have currently a proprietary sentiment analysis design fine-tuned for your consumers. With our option, we collect all touch points that are likewise utilized by Contact Lens. This enables you to decide whether to use Contact Lens for customer analysis or to adopt Amazon Connect as a data gathering system and build a data ingestion and reasoning pipeline to use your proprietary sentiment analysis design in AWS utilizing Amazon SageMaker.
To get going, we offer a complete example in a Jupyter note pad that shows you how to train and release your proprietary design as a SageMaker endpoint.
In the example, we process the text from client calls and utilize a text classification algorithm (Object2Vect) to perform belief analysis. We assume that the sentiment analysis model is already readily available.
In our example, we use random labels to train the model. A totally operational option has to include a custom-made label-gathering system (for example, utilizing Amazon SageMaker Ground Truth), however those information are beyond the scope of this post.
For more details, see Deploy a Model in Amazon SageMaker. When the endpoint is active, you can utilize Lambda to send out information and receive forecasts. We also need to convert the text to the numerical input Object2Vect needs.

Clean up.
To save money on expenses, make sure you erase all the resources you utilized when you do not need them any longer:.

class O2VTextSerializer( SimpleBaseSerializer):.
# a dictionary
def load_vocab_to_tokens( self, file_name):.
self.vocab _ to_tokens = pickle.load( open( file_name, rb)).
# a callable: string -> > list of strings.
def set_tokenizer( self, tokenizer):.
self.tokenizer = tokenizer.
def sentence_to_tokens( self, sentence):.
“”” converts sentences to tokens”””.
words = self.tokenizer( sentence).
def serialize( self, information):.
for row in data [ circumstances]:.
new_row = row.
, if type( new_row [ in0] == str:.
new_row [ in0] = self.sentence _ to_tokens( row [ in0].
if type( new_row [ in1] == str:.
new_row [ in0] = self.sentence _ to_tokens( row [ in0].
js [ instances] append( new_row).
return json.dumps( js).
serializer = O2VTextSerializer( content_type= application/json)

text = event [ text] If the text belongs to category 0, # this tests.
# loop on all classifications to get the complete category result.
label_to_test = 0.

def lambda_handler( occasion, context):.

from sagemaker.serializers import SimpleBaseSerializer
import sagemaker.predictor

predictor = sagemaker.predictor.Predictor(.
endpoint_name= endpoint_name,.
serializer= serializer,.
deserializer= sagemaker.deserializers.JSONDeserializer()).
test_payload =

Specify call recordings for both consumer and representative when you enable Contact Lens in your contact flow. We also make it possible for Contact Lens for speech analysis for English (United States) and post-call analytics.
Claim your phone number
Declaring a number is simply a couple of clicks away. For directions, see Step 3: Claim a phone number. Make sure to choose and attach the previously imported contact circulation while declaring the number. If no numbers are offered in the country of your option, you can raise an assistance ticket.
After you claim the telephone number, the representative can start receiving consumer calls and starting outgoing calls. The calls can be taped, transcribed, and kept in Amazon S3. You can then use this information to supply additional value to your organization, as shown in the next area.
Usage cases
In this section, we describe how to utilize call and chat information collected by Amazon Connect in 3 use cases: exclusive sentiment analysis, intelligent chat and call notes search via Amazon Kendra, and suggestion systems from call notes. Example notebooks can be found here. Amazon Connect is the data gathering system for all these usage cases and can supply security and personal privacy functions to fulfill your requirements, such as sensitive data redaction.
You can utilize extra AWS services with Amazon Connect to provide extra value. You can also incorporate these use cases into your CRM (such as Salesforce or Zendesk) by utilizing Amazon Connect integration functions. For more info, see Set up applications for job creation.
Exclusive belief analysis
Contact Lens is empowered with belief analysis of tape-recorded calls. This indicates you can review your contacts directly on the Amazon Connect console and learn how each call went.

Leave a Reply

Your email address will not be published. Required fields are marked *