While ChatGPT has thrust AI into the spotlight, customer experience (CX) leaders have been leveraging AI in critical parts of the customer journey for years. A 2021 report from Observe.AI, for example, found that 71% of contact centers were already leveraging AI. The motivation for using these tools is simple: better and cheaper customer interactions.
While there is no doubt that AI tools, such as chatbots, are seeing broad adoption in the contact center, there are still several pitfalls CX leaders can avoid when choosing and implementing AI tools. In this article, I'll look at seven mistakes to avoid when leveraging AI in your customer journey:
1. Not Considering How Automation Fits With Your Brand
Your brand has an identity and voice. Deciding the “personality” of your chatbot should be step one in your AI transformation. Is your chatbot playful? Does it joke with your customers? Is its voice compassionate or authoritative? How does it respond to failure? How are your customers greeted?
The marketing and other stakeholders who protect your brand are often never consulted in the AI selection and implementation process. Billions of dollars are spent annually teaching agents how to live the brand. Yet little to no money is spent on conversational design for your chatbots that are supposed to act in the same capacity as your highly trained and compassionate agents.
2. Not Knowing The Locations Of The Exit Rows
If your AI is going to fly, it’s important to know the location of the exit rows. We often don’t teach our AI tools to constantly evaluate their chances of successfully completing a transaction.
Have we kept our customers in an interaction for far too long and have them going in circles?
Is our customer getting angry or frustrated?
To ensure customers have high levels of satisfaction with self-service tools, those tools must know when to turn the interaction over to a human. All too often, though, there is no exit from AI jail once a customer enters an AI workflow.
3. Not Locating The Circuit Breakers
Even before initiating an AI interaction, we should determine if automation is the right path for the customer. Just because something is “automated” doesn’t mean automation has to be used in every instance.
• Is there a specific impediment from the customers' history that means solving their problem through AI has a low probability of success?
• Have they called about this order three times in the past 24 hours? Do you really want to route them to AI for the fourth time?
• Perhaps, it’s a high value customer to your business. Do you want them interacting with a human or a machine?
Different tools have different capabilities when it comes to their ability to plan for short-circuiting an automated workflow. You should have a deep understanding of those capabilities prior to procuring an AI solution.
4. Not Considering Agent Enablement Before Automation
Many of the dollars flowing into AI in the contact center are focused on automating interactions. Businesses may find more value can be created by focusing on AI tools that create more effective agents.
A Forrester report highlights the revenue impact better agent interactions can bring. According to the report, across industries ranging from retailers to auto manufacturers, answering all the customers' questions can lead to up to $324 more revenue per customer, increasing first call resolution can drive up to $2,003 more revenue per customer, and reps that can resolve problems without a supervisor can drive up to $1,942 more incremental revenue per customer.
5. Starting From The Top And Not The Bottom
All too often, the journey of implementing AI in the contact center starts with the question: “What can we automate?” The problem with this top-down approach is it can take too long to evaluate, design and implement.
If we start with a bottoms-up approach, based on helping our agents become more effective, you can see benefits sooner and drive revenue (see mistake four above) instead of just cutting costs. Eventually, the interactions you can most effectively automate will reveal themselves as you build out your "agent assist" tool kit.
6. Not Thinking About All The Use Cases
So much focus on AI in the contact center is spent on the customer journey. Customer experience leaders also need to understand how AI is enabling increased compliance.
AI has enabled a world where every chat, call and customer interaction can be audited and scored. Long gone are the days of only being able to audit 3% to 5% of interactions to evaluate agent performance. You can now evaluate 100% of human lead interactions and determine if they asked the right questions, followed protocols for safe handling of customer information and gauged customer sentiment at each step in the interaction.
7. Not Planning For Care And Feeding
Without a care and feeding plan your AI will never live up to its promise of providing better and cheaper interactions. To be successful, it must constantly adapt and learn, yet CX leaders aren’t prioritizing evaluating a tool's ability to learn and improve in their selection process.
The ability of a tool to improve over time should rank highly in your evaluation criteria. For example, ask yourself:
• What is my investment in professional services to implement and update the tool?
• Who is going to be responsible for “teaching” the tool?
• What is the training path and timeline for my people to learn how to improve the AI tool over time?
• What skill sets are required to update and improve the tool?
• How often should I “train” the tool with new data?
• How do I know where an interaction went “wrong,” so I can teach the tool the “right” way to handle this interaction?
You should ensure you know the answers to these questions prior to selecting any AI tool.
Need help avoiding these sins? Want to talk to a trusted resource to absolve you from your past sins? We can help. Contact us to get started.