Alok Kulkarni is Co-Founder and CEO of Cyara, a customer experience (CX) leader trusted by leading brands around the world.
getty
According to Gartner analysts, "by 2027, chatbots will become the primary customer service channel for roughly a quarter of organizations," which should come as no surprise. There's a tremendous opportunity for automation in the simple, repetitive tasks that take up most of a contact center agent’s time. Sure, this can save organizations money. But more importantly, it can support better customer experiences.
But although there's considerable promise in contact center automation, many of today’s chatbots miss the mark. They still lack the ability to efficiently respond to customer inquiries.
Yes, bots are able to complete thousands of simple tasks more efficiently than humans. But many jobs are still complex and require the intuition, common sense and empathetic service only a person can provide. That’s why, from a customer experience (CX) perspective, machines should support and complement agents—not replace them. It’s also why humans (and even AI itself) should continuously check in on those bots to ensure they’re serving the customer as intended.
In 2022, the respondents of a TechSee survey indicated that "43% switched products or canceled a contract due to poor customer service." That’s 4% higher than in 2019. Given the mixed experiences people have with chatbots, the stakes are too high to risk living on AI alone or to assume your AI is behaving.
When To Consider A Chatbot
That said, if you’re experiencing any of the following, it could be time to deploy a chatbot to assist human agents:
1. Your agents can’t keep up.
All retailers know there are high-demand periods such as Black Friday and Cyber Monday when contact centers are flooded with calls. Agents are often overwhelmed by the spikes in traffic as they try to help as many customers as they can. Meanwhile, wait times for agents get increasingly longer. This is a key moment when chatbots and interactive voice response can help.
But for it to work, humans must remain part of the equation. Most agents will tell you AI helps them solve customer queries faster. AI not only takes more inbound calls off their plates, but it also provides agents with the customer background they need to resolve more challenging issues.
So, the potential benefits of chatbots aren’t limited to what they can provide customers. They also strengthen the human agent’s ability to serve by cutting down high call volumes and the resulting stress. Many companies provide chatbots as a means of empowering agents to access the information they need to help a customer.
The problem is when companies deploy chatbots to shield agents from customer interactions or rid the contact center of humans altogether. Customers catch on, and then they stop being customers.
2. Your customers never sleep.
Customers are often distributed across many different time zones. Regardless of location, they want fast, efficient service all hours of the day.
Millennials and Gen-Z are accustomed to the gratification of immediate issue resolution. But they’re not alone. Recent survey results show that 88% of consumers consider a company’s CX to be as important as its products and services.
Chatbots not only never sleep but also can possess multilingual capabilities that allow them to communicate with customers worldwide. But having these multilingual chatbots available 24/7 isn’t enough. Customers still want the option to escalate their issues to a human. If this option isn’t offered around the clock, a mechanism needs to be in place that lets the customer know a live agent will reach out to them during normal business hours.
But there’s more. You can’t deploy a chatbot into the wild like the Night’s Watch in Game of Thrones and expect the Wall to hold. Like Jon Snow, that bot must be thoroughly trained and continuously tested. This helps ensure that your chatbot doesn't fail to be helpful. If there’s anything getting in the way of that—whether it’s caused by the bot’s training or network performance—by checking in on the bot regularly, you can resolve issues before your customers experience them.
Finally, you’ll want to provide easy access to your chatbot via all the channels your customers will go to find it. These include your website, mobile app and social media platforms.
3. You aspire to go global.
I mentioned the in-language support offered by many of today’s chatbots. The more sophisticated ones can support a wide range of dialects, phrasings and even slang. This is valuable “worldliness” for companies that want to serve people regardless of location.
Thanks to conversational AI and natural language processing, organizations can tap into the massive volumes of data necessary to train their bots. But speaking in someone’s language doesn’t mean “speaking their language.” And how can someone who only speaks French know their bot is able to connect with those who speak Hindi anyway?
This is where AI can help keep AI in line. Core to any chatbot is the natural language understanding (NLU) engine behind it. Here’s where it’s essential that your training data is pristine—something analytics should ensure. They can also help ensure your chatbot provides a fully inclusive experience, regardless of the customer’s background. Such AI-powered evaluations can even help your chatbot to work around bad grammar, shorthand and misspellings.
Deploy Your Chatbot, But Do It Right
For many organizations heading into a possible recession, automating the contact center makes sense. Competing in a global, 24/7 economy means looking into AI-powered technology to serve customers at scale.
But chatbots have a way of turning on the company by alienating customers—unless you do it right. Remember these simple rules:
1. No one can serve by chatbot alone. Automate simple tasks, and make it easy to escalate to an agent when needed.
2. Invest in upfront training of your chatbot to ensure the user experience is uniformly positive. Harnessing data from real-world customer-agent interactions helps ensure the bot is trained well.
3. Test the chatbot from multiple angles including NLU, regression testing, multichannel performance and load/stress testing to ensure the chatbot can handle high volumes.
Following these best practices, instead of adding to customers' frustration, your chatbots should alleviate the load on your busy customer support team and could even help drive revenue for the business.
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?
