Tidio’s chief customer officer provides his top tips for AI adoption
He looks back on it as his worst day as a customer support agent, but Marius Laza also remembers being proud of himself.
Long before he became chief customer officer for San Francisco-based customer service platform provider Tidio, Laza was busy working amid the holiday season answering customer questions at an electronics company. Over the course of 12 hours, he addressed approximately 200 questions – which were all basically the same question about a shipping issue.
Tidio
“Now that I think back at it, 90 per cent of those responses could have been automated, if not 95 per cent,” Laza told 360 Magazine. “If I had AI back then, I could have handled maybe 1,000 questions – where AI dealt with 95 per cent and I just oversaw the AI.”
This is the scenario Tidio is offering with the decision last month to expand its Lyro AI solution to email. The idea is to keep email interactions separate from live chat conversations while organizing them into into streamlined, manageable tickets directly within teams’ inboxes.
“I think there is space for escalations. There is space for additional responsibilities, and AI also needs some supervision as well,” Laza said. “That’s what I hope the support role will become.”
Laza offered some other advice for CX leaders who are working to bring AI further into their service and support function:
Customers may ask AI for help because your FAQs have failed them
Tidio works with a lot of online stores, and many of them have the same troubleshooting and support questions. In fact, Laza estimates that about 70 per cent of the questions Tidio’s customer success and support team gets are repetitive.
The traditional solution for that is publishing a page with answers to frequently-asked questions (FAQs). However Laza said he and his team’s research revealed much of what’s on those FAQ pages isn’t addressing the areas where customers actually need help.
“We found out that it solves less than 10 per cent of questions in most cases,” he said. As a result, Tidio’s AI solution was developed based on a sort of 80/20 rule, where 20 per cent of the knowledge and training built into the AI can solve at least 80 per cent of customer issues.
Expectations around AI are skyrocketing
With earlier waves of innovation, it could take a lot for vendors to build a business case with their customers can get support for deploying new technology. Laza said AI has been a different story entirely.
One of the biggest areas of concern, he said, is what setting up AI takes in terms of time and resources.
“They want to know, do you need an extra person? Is there some extra role on your team that you need to manage the transition?” he said, adding that Tidio has run demos where its solution has been deployed in about 60 seconds, though a few days or weeks is probably more realistic.
Laza said brands also wonder whether AI can completely automate 100 per cent of their customer service operation.
“The answer is no,” he said. “We’re seeing 80 per cent as a good baseline. We have clients at 90 per cent or over, and the average number, I think, right now, is about 68 per cent (of their support function automated).”
A lot can depend on what it will take for an organization to maintain the knowledge base that AI draws from, he added. If a brand has new pricing on its website or made changes to terms and conditions, any AI solution needs to work from content that’s accurate and up to date.
Focus on the role of navigation in successful digital CX
While a lot of the industry conversations about AI have explored what it will mean for headcount within contact centers, Laza suggested thinking about the technology as a replacement for areas like web site search bars or product description pages. In many cases customers may not be looking to speak to a human so much as find their way to the right information.
“They just want a quick answer about shipping without reading the entire shipping article. They just want to know the status of their delivery,” he said. “People are now sometimes asking, ‘Can I just speak to the chat bot please?’ Because they don’t want to deal with the wait times and for people to sift through the support data.”
Call deflection and handle time is just a starting point
The most common metrics for evaluating the use of AI in customer service have often involved some low-hanging fruit, but Laza encouraged his peers to be ambitious in how they use the technology to create bottom-line value.
“AI can help save a support agent’s time, but because the technology can deal with as many customers as possible, you can also handle spikes of activity, and even make money,” he said. “We have clients that use it for lead generation in sales, in some cases 80 per cent of their sales now happen without human intervention. AI talks to the person, and then that person makes a purchase,” he said. “To me, when you tie it to revenue, that’s when it really becomes interesting.”
