When someone mentions artificial intelligence (AI) or machine learning (ML), it can be difficult to tell if they’re really talking about a grand new technology or just using impressive buzz words. But in the case of customer care, automatic testing and AI customer experience bring real, tangible benefits.
AI allows you to converse with customers using conversational experience design, which creates personalized interactions for customers. But beyond that, a conversational solution can learn from its own interactions to make self-service even more successful through automatic testing.
Supplemented with manual experimental design, an AI solution can pay dividends for your company. But it’s important to first question whether a technology solution truly uses AI or if it’s just designed to look intelligent.
This article is part of the Inside CX series from Intrado, through which we’re creating a detailed, focused and actionable library of CX content. Use this player to hear another version of this content designed for podcasts and subscribe to future episodes.
Defining AI: Decision Making and Automatic Testing
Since AI and ML are so commonly used as buzz words, there are misconceptions about what these terms really mean. In the context of customer experience, AI and ML refer to automated technology solutions (i.e. self-service) that can determine the next best action.
The key to driving that type of automation is creating a feedback loop for automatic testing so the solution learns from every interaction that occurs. An AI customer experience solution should work off a predictive or prescriptive model to determine what to do next based on whatever data it has.
But as new components get added to the system, like new user data or business-supplied best practices, you need to understand the impact of those new introductions. Instead of taking in pure data, a feedback loop supplies end results back into the system. That enables the solution to make better choices for users. Therefore, the abilities to learn from itself through automatic testing and make decisions are what defines an AI customer experience solution.
Outcomes of AI Customer Experience
So beyond discussing AI, ML and automatic testing to sound cool at a cocktail party, what’s the point of all this? At the end of the day, there are tangible benefits of AI solutions in two aspects: customer experience and business outcomes.
From a customer experience standpoint, you may get immediate feedback from your customers. Will customers want to use your system, or will they find a way around it? Are you creating a positive experience for them?
From a short-term perspective, customers who have a good experience will potentially share that experience on social media. That creates a quick win. But looking at the long-term perspective, a single bad experience could drive a customer toward a competitor. That’s a permanent loss of lifetime value. Both in the short-term and long-term, the frictionless experience created by AI customer experience pleases customers and reduces churn, both of which help your business.
Next, there are two benefits from a business outcome perspective. First is immediate impact to revenue. Maybe somebody wants to order a movie. Maybe somebody considers subscribing to a new streaming service. Creating a frictionless experience makes it easier to complete the transaction and reduce last-second dropouts. That’s additional revenue.
Second, AI customer experience results in cost savings. AI enables people to enjoy the experience and find a resolution within a cheaper communication channel. It keeps customers away from the call center or an SMS chat with agents, which adds labor expenses and costs more than automated self-service.
Supplement Automatic Testing with Experimental Design
Regardless of how great your AI customer experience solution was when you put it in place or how well it runs automatic testing, it’s still important to run periodic checks to make the solution the best it can be. The customer experience can always be better, but how do you keep raising the bar?
To do that, you must interject change into the system. This is done in part through automatic testing, but you can take an active role through experimental design. At Intrado, we call this “champion-challenger” testing. Others call it A/B testing.
Regardless of method or what you call it, the most important factor is introducing changes and pumping results back through the feedback loop into your AI/ML system. This lets the solution learn from business outcomes that may not happen naturally during day-to-day interactions with customers. If these new strategies perform better than current practices, set them as the new baseline.
Best Practices: How to Introduce Change
The outcomes of this level of experimental design and automatic testing depends on the extent of the implementation. In essence, the results can be as grand as you choose. Here are a few examples of good, better and best practices for introducing change.
Good: Overall Best Practices
At the first level, you introduce a change, test it against the current standard and determine which is more effective. Then you apply whatever performs best across the entire customer population. So overall, your customer experience solutions perform better, but they don’t create the personalized experiences today’s customers really want.
Better: Individual Customer History
A better solution incorporates your newly found best practices but then looks at individual customer data. You look beyond how the general population or even a subset of similar customers prefers to be treated. Instead, you look at how the strategy performed during past interactions with a specific customer. If the strategy works best for the overall population but not for this specific individual, then create a more customized solution based on past interactions.
Best: Using Real-Time Data
For best results, you look at not only what happened historically for a single customer but gather real-time data on the customer’s status and what they’re doing right now. That may include using APIs to tap into your CRM systems to find the status of different products and services the customer purchased. Or it may look at communication across all channels. For example, if the person just accessed your website, use that information to figure out what they need when they call or text your contact center. Whatever they looked at online probably holds clues as to what you should recommend next.
Steps to Improving AI Customer Experience
As is so often the case, data is key to making all this happen. So when it comes to testing and applying AI to your customer experience solutions, follow these three data-centric steps.
1. Introduce Change
If you continue to do the same thing you’ve always done, you will never raise the bar. If current practices work for customer group one but not group two, then introduce changes for group two. Continue testing whenever possible or you’ll merely maintain the same level of performance you’ve seen historically.
2. Learn from External Sources
Your customers don’t live in a vacuum. They do business with other companies who gather data as well. Collect that data when possible, but more importantly, make sure your own data doesn’t exist in siloes. Specific customer interaction data may be stuck in a single channel. Put it all together to create a batch of information that benefits all your systems.
3. Gather Real-Time Data
Don’t stop with learning what customers have done in the past. Look at all the historical batch information you’ve collected and determine how to start collecting it in real time. Data is constantly growing, and real-time data enables you to provide a frictionless, personalized customer experience.
Creating Your Own AI Customer Experience
While the benefits of automatic testing and AI customer experience may sound great, the concept can still feel intimidating. If you have more questions or would like to learn more, call Intrado at 800.841.9000 or visit us at www.intrado.com/Customer-Experience. Customer experience is they key brand differentiator today, and AI and testing will be the differentiator of experiences in years to come.