For customers seeking self-service support on a brand’s website, chatbots are increasingly becoming the director of first impressions. But when chatbots aren’t designed and/or deployed properly, the optimism the customer starts with in getting an issue resolved can quickly turn to frustration. In fact, 45% of customers say their interactions with chatbots have frustrated them.

If your organization’s CSAT scores have been burned by your current chatbot offering, you’re not alone. Many business and contact center leaders who were told to quickly implement AI to solve a certain problem—maybe handle more customers more quickly while reducing costs or headcount—have found out the hard way that throwing AI at the problem can easily make things worse.

In most cases, taking a step back before implementing AI in the contact center will help organizations move forward faster and with better results.

Take a different approach to contact center chatbot implementations

Even when they are viewed as value centers, contact centers are still focused on saving money. Cutting staff has typically been a method for cutting costs, and today, the view is not very different, with AI being seen as a means to an end. In the contact center, Gartner predicts that conversational AI will reduce labor costs by $80 billion in 2026. But while most IT directors and CFOs are currently focused on reducing costs when it comes to AI implementations for the contact center, the last thing the customer wants is no human agent to talk to when they experience self-service friction. While 89% of consumers say they appreciate customer service chatbots for quick answers, 55% say that being able to get to a human agent, and particularly one with context of what they typed to the chatbot, is one of the most important aspects for them in a support experience.

Save money (and contact center agent jobs, too)

A recent NPR article featured a story on an administrative software company that implemented a chatbot to save both time and money by answering repetitive customer questions, but to also assist new and underperforming agents with the responses that created the highest customer satisfaction. Long story short: no agent jobs were harmed in the making of this AI success story, but this implementation led to a 14% increase in agent productivity, higher CSAT ratings, customers being nicer if and when they needed to talk with a live agent, and new contact center agents and low performers rapidly improving.

We also recently spoke directly to a global vice president of support with a similar approach to AI and automation. The organization saved more than one million dollars a year while improving customer satisfaction by 20%. He advised, “you can’t go at AI and automation projects with making cutting costs and cutting jobs your number one goal. If you do, you’re going to cut customer satisfaction, too. But if you approach AI and automation with a holistic look and what the customer needs, and how it can compliment people and processes, you can very likely have it all.”

Consider these three things first when implementing AI in the contact center

Taking a step back, what should organizations consider before implementing AI in the contact center?

  1. Listen to what the customer needs. With the recent fervor around AI, contact center and IT leaders need to remember first to focus on the customer, and then the technology. Ask “what do my customers actually need,” not just "what can this technology do."
  2. Define how AI can compliment both people and processes. Think about how conversational AI and Generative AI can compliment current successful processes, versus replacing them or people.
  3. Understand what insights are needed to be effective and improve. Methods and metrics matter. Take time to truly understand the customer and pain points in their journey and what kinds of insights and measurements you will need to allow your organization to be effective and to understand what’s working / not working.

Want to hear more about chatbot implementations for customer service? Watch the on-demand webinar: Not your average chatbot: self-service reimagined.