Flight Centre finds the ‘why’ with natural language processing capabilities

Flight Centre is now able to determine the ‘why’ behind customer insights following its AI and natural language processing implementations.

Ashley Taylor, global voice of the customer (VOC) program specialist at Flight Centre said that without a VOC program, customer issues might be missed.

Taylor told a recent Sydney Qualtrics X4 crowd it saw a “60 percent reduction month on month” in people discussing a particular problem after finding a “particular process defect.”

“What we were seeing was that people that paid for a specific upgrade on specific airlines, through specific systems, weren't receiving what they were doing.”

Taylor added “with no VOC program at all, you wouldn't know this was a problem. You'd just have customers that tried, you'd never hear from them again.”

After “receiving tens of comments about this particular issue through this particular channel per month” Taylor said this could translate to “a million-dollar problem”.

He said, “You would have known there was a group of customers, you would have known that they're at risk because they've told you they're the detractors, but unless you've gone into a particularly strong deep dive, which can be quite labour intensive, been able to know why that was the case”.

“With using the AI component, it's the next level of applying AI smarts to it. We're able to do something about it.”

Following the discovery, Taylor said under the new implementations Flight Centre could create a workflow and “pretty much resolved and sorted out in May.”

“That's a million-dollar problem solved,” he said.

In the talk, Taylor also said while the AI usage is “not as cool” as other use cases the travel company is “using some of the exact same principles, which is around natural language.”

He said that “knowing how our customers feel is good, but knowing what to do about that is better.

“The specific talking point to that is using AI and natural language understanding to categorise millions of sentences, hundreds of thousands of different comments from lots of different channels”.

He added the travel company pulls customer channels together from a range of places such as social media and surveys which provides a “good read on what it is we need to do in a way that's actionable”.

As the company has a footprint across online, retail and app and in different countries, Taylor said Covid-19 “gave the business a chance to pause and reset” and focus “towards globalising that … so that the local nuances remain.”

“As part of that, we decided we were going to roll out or develop a global VOC program,” he said.

Commencing in June 2022 “we started with a blank canvas” and opted to implement Qualtrics in a “sort of garden variety foundational program where we are sending them to solicited requests out to customers in a structured way.”

However, the company began using AI and natural language processing to gather “solicited and unsolicited” customer insights and pull “information from all our socials and review sites.”

“Instead of having one narrow prism, we've got a wide variety of signals and the types of feedback you get from the unsolicited compared to when it's solicited are quite different and you get a different view from that.

“It's also structured versus unstructured.  I've read somewhere recently that 80 percent of data in the next year or two that we see in our space is going to be unstructured,” Taylor said.

“That's really where the gold is. The verbatims tell you what you need to do. Score's great, but if you want to change it, you've got to understand what you need to do.

“That's have a mechanism that helps you get a lot of unstructured data together and make sense of it, humans can’t do that”.

Taylor said the company surveyed roughly two million people in the last year “feeding through operational data” plus ‘pulling through Genesys calls” and incorporating “emotional intensity” with software, Sprinklr.

“We have our own Flight Centre centric model. Probably took about 150 hours to develop … it's very specific to our needs”.

“This all flows into, into this one layer. Many sources coming into one place, not just survey data -natural language processing, the AI component enriching that and giving us insights and helping us to categorise what's going on, which allows us to do two things - business optimisation and frontline response.”