Getting AI-Powered Customer Service Right
Artificial intelligence (AI) in customer service isn’t a new concept. Harnessing the true power of AI has been an ongoing conversation in the CX world for years, but it’s a notoriously hard thing to get right.
Customer-bot interactions can swing wildly from absolutely incredible to brand-damaging, depending on how AI was implemented and how successfully it learned from prior interactions. Today, the average consumer expects personalized digital experiences as a bare minimum—experiences in which brands learn their preferences based on past behavior and use this information to tailor future interactions to their tastes.
There’s a perception that using AI in CX makes your interactions less personalized and effective, but that’s not necessarily the case. Here are three sure-fire tips on seamlessly integrating AI into your customer service operations, and how these processes can both delight your customers and alleviate your agents’ workloads.
Start by automating the right kind of queries.
When people think about AI bots, they assume that these bots answer basic questions like, “What’s your refund policy?” or “What are your hours today?” Anything more user-specific than that (like, “When will my order get here?”) is considered better answered by human agents. However, this perception is incorrect – today, bots can answer these questions.
How? They can actively pull customer data from CRM platforms, and provide granular visibility into user queries by integrating with order management systems and helpdesks. This way, when a customer asks for an ETA on their order, the bot simply taps into the order management system’s database and gives the consumer a precise answer. Bots can also be programmed to learn from past interactions, but they can’t do this on their own. You have to put in the effort to program use-cases and commonly-asked questions specific to your business. Without that work upfront, customers can easily get frustrated by bots when they:
- Ask customers to repeat and rephrase their queries multiple times before providing solutions
- Have incomplete decision trees, leaving customers in the dark with a lack of resolution.
- Refuse to hand the customer over to an agent, OR hand the customer over to an agent sans any context of the bot’s conversation, forcing both the agent and the now-impatient customer to start from square one.
Teach your bot to study successful agent-customer interactions to improve accuracy.
When it comes to learning to deliver delightful customer support, who better to study than the rockstars you hired for your team?
The more your bot learns from your experts, the smarter it gets. By studying successful agent-customer interactions from your top performers, your AI bot can respond with greater accuracy. Leaving your bot to its own devices—educated, yes, but still going solo—can be nerve-wracking, even to the most confident of CX teams. In this case, setting up a deflection team to monitor, handhold, and enhance your bot’s initial performance can be of huge benefit to your business. Think of your deflection team like your bot’s cheering squad.
PhonePe, India’s largest payments app, successfully automated up to 80% of their support inquiries with AI bots alone—no human interaction required. They process over 1.5 billion transactions per month, with contact volumes doubling every 3-5 months. AI bots have proven immensely valuable for them.
PhonePe’s bot is able to answer granular questions like “What’s my balance?” thanks to a deep level of integration with other systems that house PhonePe’s customer data. The team eventually automated 850 decision items and dramatically improved their CSAT scores. PhonePe is the perfect example of bots working seamlessly with a team of live agents. Remember, it’s AI plus humans, not AI vs. humans.
Don’t forget to utilize agent-assistant bots, too
Customer-facing AI bots are all the rage and get lots of attention, but there’s a solid place in the CX world for agent-facing bots. Everyone dreams of having the more mundane and administrative parts of their jobs automated away so they can spend more time doing what they love.
Capabilities of agent-facing bots, including ticket classification, automatic ticket routing, and relevant article suggestions for next steps can be a huge timesaver for your agents. By routing specialized tickets to the right agents—who have the knowledge to assist customers with their queries—time delays and internal customer call transfers can be avoided, saving agents around 1.2 hours a day and positively impacting CSAT scores.
And when 1 in 3 customer service leaders believe that building dedicated digital platforms to help customer-facing teams work better is a top digital investment priority, it’s hard not to get excited about agent-facing bots teaming up with customer-facing bots.
AI-powered service is a great way to start building a digital-first service strategy.
AI and bots are slated to be a top CX priority this year, alongside omnichannel experiences, mobile-first conversations, and high-speed service. Read about these up-and-coming trends in the Future of CX:2022 report to make informed decisions on your upcoming CX investments!
