This article was originally published in the Union Leader.
For those of us paying attention in the customer service space, there are new developments in artificial intelligence integrations every week. These developments promise to do everything your regular customer service team can and should be doing, improving your customer satisfaction scores, reducing your overall costs, and, perhaps, tucking you into bed at night with a sweet bedtime story.
As with all things in the tech space, however, bigger promises come with larger costs. And while large organizations may be able to experiment with new technologies, start-ups should be more hesitant to throw money at unproven technologies.
That’s not to say all AI systems should be off the table. Organizations looking to bolster their customer service departments should be investing in AI systems. But new technologies and systems are constantly being developed, and navigating the industry at the onset can be costly to an organization.
Before you commit to a new AI project for your customer service team, here are a few helpful steps to consider:
First, understand your organization’s goals for the customer service team. AI systems should directly impact reportable targets for your team. Are you focused on deflections so you can reduce headcount, or are you focused on customer satisfaction improvement?
There are dozens of different KPIs (Key Performance Indicators) your team can use to measure your team’s overall success. Which ones are you looking to bolster with your AI systems? Be candid in your discussions with your team about your goals and defining what a successful AI system will look like for your overall performance. Just remember: the more KPIs you expect an AI system to help, the more complex it will need to be by necessity.
The temptation here will always be to focus on deflections, and with good reason. It’s basic math: the more emails and chats you can deflect before they reach your customer service team, the fewer agents you have to make available to otherwise answer those inquiries. These chatbots can – and should – be able to handle the low-hanging fruit within your inquiries, allowing your agents to focus on more difficult tasks.
Improving customer satisfaction is a more nebulous proposition. Certainly by deflecting contacts with automated systems you can create an overall better customer experience, but that’s not always the full equation. For example, customer service departments that have ongoing relationships with their customers can benefit from AI systems that can create customer summaries based on call history. This lets your reps have a quick, paragraph summary of your customer’s history when they call, instead of requiring them to read through dozens of call notes, thus improving accuracy and customer satisfaction.
Second, know that most AI systems are only as good as the information you can input into them. An AI system will not be able to magically answer your customer’s questions if you do not have good data to feed it. This is part of the “generative AI” buzzword that gets all of the attention: if you don’t have a robust FAQ or Help page to help train your AI system, your AI system will make more mistakes when answering customer inquiries, leading to greater dissatisfaction overall. Take time to build those before you invest in AI, and then the AI system integration will run much smoother.
Some systems can go back and read your historical contacts and learn from those. Of course, that usually costs some additional bandwidth for whatever AI system you choose, so expect that discussion. Historical information isn’t always a good thing, however. How often do you have updates to your product? It can become tricky if, for example, your product didn’t have a highly-demanded feature a few months ago, and suddenly, it does. That’s a lot of AI training to overcome for a significant product update, and can lead to confusion for your customers, your customer service team, and the system.
Along with that AI training is the understanding that training your AI system is not a “set it and forget it” process. As your company grows and develops new products and features, you will need to dedicate resources to consistently improve and update the AI training. Failure to do so can cause added customer confusion, particularly when they want to know about those new features.
Finally, know your customer base. While many tech-savvy Millennials and Gen-Z professionals are happy to embrace new AI technologies, not every customer base is as open to adopting new technologies. Work with your marketing and customer experience teams and understand the demographics they are targeting to better understand your customers and their tolerance for new technology.
It’s a brave new world in the customer service field; one where new promises are being made and new innovations are constantly under development. CS managers and directors shouldn’t shy away from this new world, but neither should they fully embrace it without a solid roadmap.
Ken Easthouse is the co-host of “How Can I Help You?”, a podcast focused on the customer service industry, and has more than 20 years leadership experience within the customer service industry. He lives in Manchester with his two cats and his crippling student debt.