Here’s how to use Amazon Machine Learning capabilities to build an intent prediction model in Amazon Connect.
- Understanding customer intent is key for contact center agents.
- Businesses can use artificial intelligence (AI) and machine learning (ML) to assess customer data from a variety of sources and use it to predict customer intent.
- The Amazon Connect virtual contact center solution lets a business build and deploy customer intent prediction models.
- A Connect customer intent prediction model lets a business use AI and ML in conjunction with customer data from multiple sources to help contact center agents deliver exceptional customer service.
Your contact center agents must be able to determine why a customer is calling your business as quickly as possible. Yet, legacy technologies often make it tough for agents to find out why a customer is reaching out.
For example, a company may use a menu tree to clarify customer intent. The tree can use Amazon Lex or other natural language understanding technologies that let customers use their words to tell a business why they are calling.
Menu trees can be effective, but they can also be frustrating for customers. One alternative to menu trees that businesses should consider is building and implementing a contact intent prediction model that leverages artificial intelligence (AI) and machine learning (ML) capabilities.
A contact intent prediction model that utilizes AI and ML allows a business to capture customer data across multiple sources. The business can then use that data to learn, generalize, and predict caller intent. The result: Contact center agents gain insights into the customer’s contact intent that equip them to provide intelligent, personalized, fast, and efficient support.
How to use an intent prediction model in Amazon Connect
The Amazon Connect omnichannel cloud contact center solution enables businesses to develop an intent prediction model that uses Amazon ML capabilities. To set up this model, you’ll need an Amazon Connect instance configured for inbound and outbound calls. Next, claim a phone number. You can then create an intent prediction model that uses the following steps:
- 1. A Connect contact flow is invoked as soon as a customer reaches out to your contact center.
- 2. Connect invokes an AWS Lambda function to pass the customer ID information within the contact flow.
- 3. The Lambda function invokes the Amazon Personalize API, which retrieves the recommendation for customer contact intent. Personalize uses a trained model based on historical contact activity data from a similar group of callers.
- 4. If intent is predicted with a high confidence score, the Lambda function delivers a predicted intent to Connect. Conversely, if a low confidence score is generated, the Lambda function returns a value indicating that the model could not predict intent for the interaction.
- 5. In an instance where a high confidence score is generated, the contact flow in Connect proceeds with the intent capture experience using either Lex or a menu.
- 6. Before the end of the call, the confirmed intent of the customer is returned to the Personalize API. This helps provide accurate predictions for this caller in the future.
There are many ways you can use your intent prediction model. For example, you can use the model as part of your efforts to build a complete view of your clientele. Or, you may leverage your model to identify “micro-moments” — instances when customers are ready to engage your brand — and find ways to get the most value out of them.
Regardless of how you use your model, you need to leverage it consistently. That way, you’re well-equipped to collect and measure customer intent now and in the future.
3 benefits of building an intent-driven contact center with Amazon Connect
An intent prediction model in Amazon Connect helps contact center agents accurately identify customer intent. It can help you build an intent-driven contact center that allows your agents to:
1. Learn from customer conversations
Your model supports real-time listening across your contact center. It lets you gain insights into customer conversations, so you can uncover ways to optimize your contact center operations.
2. Generate actionable customer insights
Real-time customer intent data can be placed into dashboards and reports over time. From here, you can analyze the data to identify trends and patterns hidden within it. Next, you can generate insights from your data and use them to find ways to assist your contact center agents. This allows your agents to get the help they need to handle customer requests with speed and precision.
3. Deliver exceptional customer experiences
You can automate predicted intents in Connect to drive high self-service rates. If urgent calls are identified, they can be immediately routed to agents without the need for a lengthy handoff. Meanwhile, once an agent gets on a call with a customer, they are in a great position to provide an outstanding experience because that agent understands why that customer may be calling and can handle the request accordingly.
The bottom line on setting up a customer intent prediction model in Amazon Connect
AI and ML are powerful technologies, particularly when it comes to learning about customer contact intent. With Amazon Connect, a business can utilize an intent prediction model equipped with AI and ML capabilities to leverage a large collection of data to predict why customers are getting in touch. Contact center agents can then retrieve insights that allow them to predict customer extent and deliver exceptional support.
For businesses that want to deploy a customer intent prediction model in Connect but require additional assistance, it pays to partner with an Amazon Managed Services Partner. A partner ensures your company can receive expert support as it tries to bolster its Connect contact center.
CloudHesive is an Amazon Managed Services Partner that offers comprehensive Connect support. Businesses use our Customer Connect solution to maximize the value of their Connect investments. Contact us today to learn more.