According to a recent survey, 75% of organizations point to the customer experience as the area where their digital initiatives deliver the greatest value. Read this article to uncover six automation projects that can help you build an improved experience for your customers.
Content Summary
Why automation?
Why these six projects?
Refund request approvals
Route customer requests
Update customer information
Remote data upload
Compliance verification in retail sales
Respond to customer queries
Key takeaways and next steps
Why automation?
Predicting which customer behaviors will persist post-pandemic—from wildly fluctuating demand to heightened health and safety precautions—can be tricky. But one area you can control is how you manage the experiences you create for your customers. Automation, especially automation combined with AI, can help you fix or refine these experiences, resulting in higher sales, better use of resources and greater customer satisfaction.Predicting which customer behaviors will persist post-pandemic—from wildly fluctuating demand to heightened health and safety precautions—can be tricky. But one area you can control is how you manage the experiences you create for your customers. Automation, especially automation combined with AI, can help you fix or refine these experiences, resulting in higher sales, better use of resources and greater customer satisfaction.
Why these six projects?
IBM recommends these six automation projects because they’re:
- Able to be implemented quickly with impact
- Cost-effective when using cloud software solutions
- Applicable to virtually every business
- Likely to create persistent value going forward
- Able to interact with and build on each other to multiply positive effects
Refund request approvals
Issue
Meet Claire. She’s known for her speed and efficiency, but lately she struggles to keep up with customer refund requests due to a spike in requests and reliance on a largely manual process. She’s concerned that customers must wait too long for a decision or that inaccurate service fees are repeated on the next bill. This results in multiple emails or calls for the same customer issue, which further increases backlog.
Action
Using decision services and workflow management, eliminate manual approvals that delay response times, lead to inconsistencies and are expensive to scale.
Automate
What a refund approval looks like when automated
- Customer refund request is routed to a decision service
- Decision service determines approval by considering multiple inputs, such as customer propensity to churn, profitability and transaction history, and ensures the automated decision is as reliable as one made by human experts
- System issues refund using workflow capability, notifies customer and updates backend systems; manual intervention only needed for exceptions
Not that
Avoid automating a personal apology when a human, personal touch is warranted.
Route customer requests
Issue
Meet Jose. He manually reviews and routes incoming customer emails for his bank and frequently learns something new about his customers in the process. He wishes he had more time to identify opportunities to better serve customers with relevant product and service offers, but he spends most of his time routing an increasing volume of emails.
Action
Streamline how you process incoming communications using intelligent automation.
Automate
What customer request routing looks like when automated
- AI classifies communications and extracts data
- Decision service determines handling priority based on multiple factors, such as loyalty and profitability
- New case is created to process customer request using workflow capability, then the case is automatically populated with extracted data
Not that
Avoid automating actions or responses that appear to feel your customers’ pain—leave sympathy and empathy to humans.
Update customer information
Issue
Meet Erin. She enjoys her colleagues and job, but she’d prefer to spend less time manually transferring customer data into multiple backend systems. As diligent as she is, she’s concerned about making errors and knows she could be more effective if she spent her time differently.
Action
Build robotic process automation (RPA) bots that can read and update customer data.
Automate
What a customer information update looks like when automated
- Customer submits new address online, which is picked up by a bot
- New address is validated and bot updates existing systems
- If address is invalid, customer is contacted for resolution
Not that
Avoid automating any tasks or processes that a trained employee could do quicker than your customer.
Remote data upload
Issue
Meet Toyin. She regularly speaks to customers to ensure their personal and other relevant information is accurate. She believes these back-and-forth verification conversations could be easily automated for better self-service and time savings.
Action
Enable customers to quickly and reliably upload information from anywhere using intelligent mobile capture.
Automate
What remote data uploads look like when automated
- Customers capture documents using their mobile devices
- AI-based mobile capture extracts information from relevant documents, such as names or policy numbers
- Extracted data is automatically validated; customers can verify before data is uploaded to system
Not that
Avoid automating any action that would be annoying if a person did it—and would be just as annoying when automated—such as sending multiple irrelevant texts or emails.
Compliance verification in retail sales
Issue
Meet Priya. She supports the online sale of regulated products. Because her company relies on manual oversight to ensure compliance, she can’t always get the right answer to customers the first time. It’s also difficult to scale this oversight when volume increases.
Action
Automate online purchase validation in real time using decision services.
Automate
What compliance verification looks like when automated
- Regulated item is automatically evaluated against predefined set of eligibility rules at point of sale, before order is confirmed
- Order is monitored and offerings adjusted to improve customer experience based on decision outcomes
Not that
Avoid automating advice in unclear situations—instead, make it quick and easy for customers to reach a friendly, well-informed human who can help.
Respond to customer queries
Issue
Meet Abdul. He handles an increasing volume of customer queries and transactions. Even with the help of basic chatbots, he can’t always meet customer experience expectations for speed and quality. Often chatbot interactions end quickly because the chatbot has limited capabilities to respond accurately or because the customer doesn’t know what he needs.
Action
Integrate intelligent virtual agents (IVAs) into the systems used by the people who serve customers to improve efficiency and outcomes.
Automate
What a customer query looks like when automated
- Customer query is routed to an IVA, which is designed to emulate human service and support agents
- IVA guides customer using natural language processing (NLP) while interpreting the context and providing relevant answers and feedback
- If the bot can resolve the query, it executes the appropriate action but, if needed, the bot routes customer to human employee
Not that
Avoid automating actions that cause human employees to manage laborious tools or systems instead of connecting with customers.
Key takeaways and next steps
If you’ve already started to digitize and automate key customer-facing processes, great news. You’re in a better position to adapt and change to whatever customer behaviors persist as you build a pipeline of customer experience (CX)-focused automations, for now and later.
The capabilities that automate each of these use cases are part of IBM Cloud Pak® for Business Automation, a fully equipped, modular and open platform for automating virtually any type of work at scale. Its modularity is designed to help you fill urgent tactical gaps while laying the foundation for growth-driven automation. A subscription-based model allows you to start new projects quickly and affordably.
