Sitel Group - October 8, 2018 - 184
When the 2018 European Contact Centre and Customer Service Awards get underway in London’s fashionable Battersea Park on November 27, TSC will compete alongside Booking.com, Marks & Spencer and Capgemini in the Best Implementation of Artificial Intelligence (AI) in Customer Service category.
The implementation in question is for a personalized chatbot developed for Compagnie des Alpes to accompany visitors to, around, and from one of the most popular tourist attractions in France – the Parc Astérix theme park.
“We successfully negotiated the judging panel to make the final cut on October 4 and the nomination gives the project an extra layer of validation,” begins Geoffrey Boulakia, General Manager EMEA for TSC of the award nomination. “It is proof that beyond the buzzwords and the hype, AI is a powerful tool that brings real benefits to consumers of all ages, now. But, only if you are prepared to take a truly methodical approach.”
Chatbots are in vogue at the moment and interest among brands is growing.
“This is partly due to Messenger, WhatsApp and WeChat,” explains Boulakia. “These messaging platforms support chatbots and their popularity is still growing. A chatbot on Facebook’s Messenger platform could potentially address 2 billion people.”
That’s what makes the Parc Astérix chatbot even more special. At a time when TSC is receiving a large number of requests to build chatbots, this project in partnership with Compagnie des Alpes, started with a question.
“The client wanted to know how it could reach and help as many visitors and potential visitors to its theme park as possible and it wanted to use AI to do so,” says Boulakia. “As we began working together and developed the use case and parameters, the answer became a chatbot.”
And it’s a chatbot that took nine months to come to life. All existing AI solutions had to be benchmarked and the client’s existing IT ecosystem had to be audited to understand what solution would work the best with Parc Astérix’s existing set up.
“You have to be pragmatic and technologically agnostic – otherwise you won’t select the right tools for the job – the tools fitting both the needs and existing infrastructure,” Boulakia warns.
If executed correctly, a chatbot offers an organization an always-on, always-there customer assistant capable of responding to a host of frequently asked questions. It can deflect low-value calls that otherwise have to be dealt with by agents – allowing them to focus their talents on the customer engagements that really matter. Further, the technology supporting them is beginning to mature at a time when people have never been more open to self-service.
According to Dimension Data research, when a customer has a question or issue with a company, 73 percent will start with an online search, compared with just 13 percent that immediately reach the phone and call the firm’s contact center. Indeed, a recent Nuance study found that 67 percent of consumers prefer a self-service option to having to speak to a contact center agent, hence 75 percent of consumers viewing self-service options as the most convenient way of resolving an issue, as long as those options really work.
“That’s the crux of the matter. A chatbot needs data and a chatbot should never be a one-shot project,” says Boulakia. “Like any customer experience channel, it’s an ongoing investment and something that must be constantly refined in line with consumer expectations.”
Even though chatbots are gaining in both use and acceptance – there are currently 300,000 in operation just on Messenger – there is still a perception gap.
“AI is happening now but machines learning by themselves only happens in Hollywood,” Boulakia points out. “For a customer service chatbot to really deliver, it needs to be trained and calibrated by customer service agents. Many brands don’t ‘get’ this. With this chatbot for the Parc Astérix, we didn’t simply use existing data to build the chatbot. We conducted field work, interviewing contact center agents to get qualitative insights that would really help us understand what questions required answers and the many different formulations of each question.”
“Once launched, you need to continue to collect data to take the chatbot to the next level and you must make certain each iteration is noticeably better than the previous version,” continues Boulakia.
There is concern that like with other disruptive technologies or tools, companies believe they can just buy a product and in Boulakia’s words, “wait for the miracle to happen.” But nothing could be further from the truth.
What’s more, even companies like TSC, who have successfully delivered countless chatbots, must take each new project as a unique set of circumstances and a unique use case.
“Never forget that it’s a new channel and it will be used differently. People were surprised by this chatbot so they asked it surprising questions,” Boulakia explains. “We believed the initial launch version would be able to handle 50 percent of queries but because of this novelty factor, it was closer to 30 percent. Now it’s surpassing 50 percent and it will continue to get better, but no customer service chatbot will ever achieve 100 percent – that’s why you have to give customers a human alternative.”
If this Parc Astérix chatbot scoops the prestigious award, it would be a fitting conclusion to a very demanding and challenging CX project, but even if it doesn’t, the chatbot’s performance figures are reason to celebrate. It is currently serving around 10,000 people, handling 16,000 different questions and one-in-two users are highly impressed with it.