Amazon Brings More Machine Learning to the Contact Center

It’s fair to say the contact center market was one that basically stood still for decades. There were a handful of vendors that offered on-premises solutions that met the needs of most customers. But then along came the cloud, it introduced a number of new companies, and innovation exploded. One of the benefits that SaaS apps have over on-prem solutions is that new features can be developed and delivered faster, because the new capabilities are available as soon as the SaaS vendor adds them.

AWS jumped into contact centers in 2017

One of those new vendors was Amazon Web Services when it announced Connect in 2017. It was certainly a late entrant to the market, but its pay-as-you-go pricing model made the offering unique and promised to save customers huge sums of money. Another aspect that makes Connect unique is that it was built from the ground up with artificial intelligence (AI) and machine learning (ML) as part of the feature set. AI and ML will evolve the contact center in ways that are hard to imagine, and AWS is using its deep expertise in these areas to add a number of features that can help organizations step up their customer-service game.

The Connect product is the same customer-service technology currently used by Amazon. Customers can set up and configure a contact center in literally minutes, and there’s no infrastructure to rack and stack or manage. Although AWS was late to the cloud contact center game, it certainly has an impressive list of customers that include Intuit, Best Western, John Hancock, GE and Capital One. During a press pre-briefing, AWS claimed customers can save up to 80% of the cost of traditional contact centers, which is consistent with the customers with whom I have talked.

How AWS bringing machine learning to the contact center

Now that the install base has been established, AWS is looking for ways to optimize the agent experience across the interaction lifecycle. At this week’s digital re:Invent, AWS announced a number of new AI / ML-based features for Connect. Details are as follows:

Connect Voice ID uses machine learning-powered voice analytics to authenticate customers rather than the cumbersome and frustrating process of answering multiple questions to confirm identity. Anyone called into a contact center knows how time-wasting and utterly maddening the process of providing things such as social security numbers, addresses or even a mother’s maiden name. Voice ID uses the caller’s voice to verify identity. This is significantly faster and simpler and does not disrupt the natural conversation.

The feature enables a customer to create a voiceprint based on the unique characteristics of his/her voice. When they call in, the service analyzes the speech attributes such as rhythm, pitch and tone, compares the caller’s voice with past recordings and generates an authentication score. Then, depending on the score, the contact center agent and manager can determine whether further verification is required.

Connect Customer Profiles brings together customer information from multiple applications and creates a single, unified profile. The information is delivered to the contact center agents at the beginning of the interaction. The feature scans and matches customer records across a large number of disparate systems, looking for unique qualities, such as phone numbers and account IDs. It then combines the data with things such as contact history, the number of times they have been on hold and customer sentiment, along with customer information from other apps. This is done with connectors to Salesforce, Zendesk, ServiceNow and Marketo. It also comes with an SDK (software development kit) and API (application programming interface) that companies can use to integrate into their own homegrown applications.

As a result, agents can have all of the customer information they need in a single place, so they can deliver more personalized service and improved customer satisfaction. This is critically important, because customer experience has now become the No. 1 brand differentiator.

Connect Wisdom uses ML to help agents resolve customer issues faster. With legacy contact centers, searching for the right answer to a customer inquiry forces agents to look across a myriad of systems, which is long and time consuming. With Wisdom, agents can search across multiple knowledge repositories at the same time, including FAQs, wikis and CRM systems. There are some systems that do this, but Wisdom leverages speech analytics to automatically detect customer issues during the call and provide real-time recommendations of articles to agents.

This results in faster issue resolution and improved customer satisfaction. Contact center agents can rate and comment on knowledge articles and improve the quality of Wisdom recommendations.

Real-time Contact Lens for Connect uses machine learning to detect calls that aren’t going well and alerts managers when there’s a problem. Last year, AWS announced Contact Lens, which delivered historical analytics to find calls that did not go well. The manager could listen to the recording or read a transcript to find out why it happened. This feature has been popular, but customers want the ability to perform analytics on the call as it is happening and not just after the fact.

With this capability a contact center manager can create certain rules that can flag potential issues such as speech patterns or volume levels. The supervisor can then jump on to the call and provide proactive assistance to agents while the call is in progress. This service also provides real-time transcripts so that customers don’t have to repeat themselves when they get transferred over to a new agent.

This feature has been particularly valuable during COVID-19 because agents are working from home. Historically, managers would identify trouble calls by walking around a contact center and listening for audible cues. This isn’t possible when agents are in their home, but ML algorithms are always listening.

Connect Tasks, as the name suggests, automates and manages tasks for contact center agents. This addresses all of the post-call items that need to be done, such as tracking followups, updating customer information, issuing product returns and more. This is a stark juxtaposition with the inefficient methods of handling calls that includes things such as sticky notes on monitors, jotting notes on a pad or toggling back and forth between apps. These historical methods waste time, lead to errors and irk customers.

Tasks makes it easier to follow up on to-dos, like notifying a customer of a status change or initiating a product exchange. This makes it easier for the agents, and it actually enables managers to automate some of these tasks entirely. The feature lets managers assign and prioritize tasks to their agents directly through the Connect interface. Based on availability, agents are assigned tasks in the same window as all of their calls and chat interactions. Agents can also assign their own tasks as well as monitor their own followup work. With Tasks, agents can focus on customer service instead of worrying about followup items.

All of these features were made possible because AWS embeds machine learning directly into the Connect call center software. AWS has been great at reinventing ways of working through the use of technology advancements. The company has long excelled at making customer service better, and it’s using its experience to enable customers to do the same.

For a while, there was some debate as to whether AWS was serious about the contact center, but from my interactions with the company, they couldn’t be more serious. Connect is here to stay and the big winners are customers.

Bossa Nova Data Solutions Caperio AI performance platform is hosted on AWS and integrates easily with cloud and on-premise contact center solutions.