Innovating contact centre customer experience - Amazon Connect

As an outcomes-driven consultancy, Cevo are committed to helping our customers deliver an excellent experience to their own. We recently took Amazon Connect for a test drive along with the suite of Cognitive Services designed to disrupt the contact centre industry.

Arvind Chaudhary

When it comes to delivering great customer experiences, contact centres are no exception. Organisations are continuously trying to innovate around providing a superior customer experience, facilitate their queries more efficiently and leave a lasting positive impact.

Innovation in the traditional contact centre landscape raises challenges due to the complexity of current systems, interoperability of components and increasing operating expenses. Many businesses that utilise traditional contact centres depend on human understanding, communication, and point in time decision-making to handle their customer experiences, which can lead to inconsistency, delays and hinder business growth and agility.

Cognitive technologies (machine learning, artificial intelligence, natural language processing and robots) provide innovation possibilities for these businesses to augment the use of human intelligence and achieve better results from all customer interactions.

The AWS suite of cognitive customer experience services have been built to challenge the traditional contact centre delivery methods. In this post we delve into Amazon Connect and other AWS Cognitive Services, and document how we quickly and easily set up demo environments that bring these concepts to life.


What is Amazon Connect ?

The Amazon Connect service was launched to exploit cloud-based communication platforms (CPaaS) and begin the next wave of disruption in the contact centre industry. It is a cloud-based self-service contact centre that makes it easier for businesses to provide better service for their customers at a lower cost. The platform is based on the same technology used by Amazon contact centre associates around the world to power millions of customer conversations.

Amazon Connect enables businesses to change the focus from high availability operations to continuous innovation, delivering greater satisfaction around customer experience.

The technology stack is simplified. It combines the complicated hardware and software layers of a traditional contact centre systems into a fully embedded and approachable web platform with no infrastructure management required.

The diagram below shows the main components of Amazon Connect and additional, complementary services in the Cognitive CX suite of tools.



Key Amazon Connect Features:

  • Managed Service, Pay as you use
  • A user interface to create customised contact flow
  • Dynamic contact flow based on customer attributes
  • Build natural language contact flow using AWS AI services
  • Real-time call analysis using kinesis video stream for insights
  • Skill-based routing
  • Outbound calling for sending reminders etc
  • Integration with customer relationship management tools like salesforce, zendesk etc
  • Integration with other systems using AWS Lambda
  • Call-related real-time and historical metrics available for analysis and improvements
  • A/B testing (routes customers randomly based on a percentage).


Building a Demo

To put a number of these claims to the test, and explore Amazon Connect services in more depth, the Cevo team developed a web-based contact centre with smart routing for a fictitious company. Some key components of the solution were:

  • A toll-free number (1300 xxxxx xxx) provided by Amazon Connect that can be used by customers for local-cost calling
  • An Intelligent contact flow, where customers can use their voice to navigate the choices
  • A web portal for contact centre agents to answer customer queries using a soft phone
  • Call logging and recording services to capture data and insights around call lengths, inbound origin numbers, abandoned calls etc.


High Level architecture


A number of services in the Cognitive CX suite were used in our example implementation and are expanded on below:

Amazon Connect:

Amazon Connect is the core of the contact centre. You can create your own contact centre instance from the AWS console and use it to create contact flows. Contact flows can be dynamic based on information retrieved by Amazon Connect from other AWS Services or systems. For example, an airline can design a dynamic contact flow, which finds their travel schedule in the system and offer ‘check-in’ or ‘reschedule’, if they have a flight in the next few hours.

Amazon Lex:

Amazon Lex provides the advanced deep learning functionalities of automatic speech recognition (ASR) and natural language understanding (NLU) to convert speech to text and recognise the intent of the text, enabling lifelike conversational interactions.

Lex enables customers to choose options using Dual Tone Multi-Frequency (DTMF) or voice commands. Using Lex, user goals (intents) are created, which are presented to customers when they call the contact centre.

AWS Lambda:

You can use AWS Lambda to integrate your contact flow with other systems and services and use their response in the contact flow. Lambda, for instance, was used to send SMS notifications for a general inquiry to salespeople.

AWS Kinesis Video Stream:

AWS Kinesis is a streaming platform, allowing you to video stream real-time customer conversation. You can consume these conversations to run real-time analysis on the speech and generate insights.

AWS Transcribe and Comprehend :

AWS Transcribe and Comprehend can be integrated with AWS Kinesis to recognise speech, and perform sentiment analysis on a real-time conversation with the customer. The output of these services can be streamed directly to a customer agent application using API gateways and WebSocket to help customer agents understand and deliver a suitable experience to customers.

AWS Polly:

Amazon Polly is a Text-to-Speech service that uses advanced deep learning technologies to synthesise speech that sounds like a human voice. Polly was used in the contact flow for recorded prompt and taking user’s input. You can also use Speech Synthesis Markup Language with Polly to control accent, pitch, volume, rate, breaks, and emphasis, and deliver a personalised experience to customers.

AWS S3:

AWS S3 was used to store call recording and reports. These call recordings can be used in the future to better train agents and machine learning models, or improve the performance of Lex.

AWS Cloudwatch Logs:

All the contact flow invocations and every stage executed gets logged to Cloudwatch logs. These logs can be used for debugging and monitoring.


Hi, how can I assist you? Amazon Connect in action

When a customer first calls the contact centre, they are provided with several choices. Based on their input, the contact flow triggers AWS Lex to record and answer their basic queries to help save the contact centre agent time. If the customer request is not handled within the flow or customer wants to talk to a human agent, the contact flow puts the call in a queue and the next available agent can answer their call.

In order to better comprehend the customer need, a real-time conversational and sentimental assessment will be provided to the contact centre agent.


What did we learn?

In just a matter of days, we were able to quickly and easily build and demonstrate the features mentioned above. From our experience there are a couple of points worth noting for those wanting to try this out themselves:

  • You can set up a basic contact centre in just a few steps, add agents from anywhere in the world, and start to engage with your customers right away.
  • There are some example contact flows for different use cases already available in Amazon Connect to help you quickly start with validating some of the ideas.
  • Creating contact flows however, is only possible through Amazon Connect console.
  • Contact flows in Amazon Connect are not linked to an instance and therefore can be copied from one instance to another e.g. promoting contact flows from development to production.
  • The Amazon Connect Streams API lets you embed the Contact Control Panel (CCP) UI components into any web application.
  • Real-time streaming of customer conversation does require some AWS service limit considerations because it uses one Kinesis Video stream per call.


What next?

Firstly, it is important to understand the complexity and challenges of your current contact centre business. This can assist you in identifying potential opportunities and the Amazon Connect features that will drive real benefits to your organisation.

Not sure where to start? As an AWS Advanced Consulting Partner, Cevo can work with your team to quickly set up a pilot to demonstrate how Amazon Connect can add value for your business.

Along with training and experimentation with Amazon Connect, Cevo can also help you build cloud capability as a foundation for operating with Amazon Connect and other AWS services.

Get in touch to find out more!

References:
https://aws.amazon.com/blogs/contact-center/adopting-amazon-connect-in-your-enterprise-contact-center/

https://aws.amazon.com/blogs/contact-center/perform-an-external-screen-pop-with-amazon-connect/

https://aws.amazon.com/solutions/ai-powered-speech-analytics-for-amazon-connect/

https://github.com/aws-samples/amazon-connect-realtime-transcription

https://aws.amazon.com/premiumsupport/knowledge-center/disable-recording-amazon-connect/