Wayfair’s customer experience analysis may soon run on AI

Late last year, Neville Clemens, associate director of customer experience at Wayfair, took to LinkedIn to crowdsource ideas on leveraging the home-goods retailer’s vast troves of unstructured data about the customer life cycle.

“Every other week I hear about AI taking over the world in some newsfeed. Somehow I can’t reconcile that narrative with what seems to be the current state of our ability to deeply understand customer interactions at scale,” he wrote.

“But curious to know if there are truly high-quality solutions that can do this…and by high-quality I mean that the results would be comparable to having an army of humans do the same task manually.”

Fast forward five months, and Clemens, who oversees the customer experience post order (like shipping and delivery) is in the midst of getting proofs of concept from third-party providers who claim to do just that: analyze customer interactions across channels using natural language processing, and identify emerging themes in real time.

Leaking nuance: Clemens is primarily looking for a solution that will enable Wayfair’s customer experience teams to interpret immense amounts of unwieldy call center transcripts.

“One of the primary means of us understanding the customer experience is when [they] call in,” Clemens explained, adding that between chat, email, text, and calls, Wayfair handles “millions” of contacts.

At the end of a conversation, agents choose a tag for the interaction from a dropdown list. But according to Clemens, that’s where a lot of key information slips through the cracks.

“A lot of the nuance of that conversation is lost when you’re choosing one reason or two reasons [for calling in]” he explained.

“That’s essentially where we were struggling, because…[if] there were 500,000 calls last year related to ‘address issues,’ I don’t know what that means,” Clemens said. “Does that mean that they tried to call in because they wanted to correct an address? Or did they call in because it got delivered to the wrong address?”

Predicting future problems

One of the companies Clemens is working with is CX forecasting platform Artifact, whose customers include everyone from DTC food brands like Siete to air-freshener makers like Pura.

“We see companies being less satisfied with being able to tell data stories about what’s happened in the past,” Artifact CEO and co-founder Nate Sanders said. “We’re seeing a lot of teams want to act right now, and set strategies right now for the customer experience.”

A lot of analytical tools available today are able to analyze past interactions, Sanders explained. Predictive analytics, combined with natural language processing, can enable companies to understand trends in real time, and get predictions about themes or issues trending upward, he said.

According to Sanders, retailers like Wayfair are great use cases for this high-tech approach to consumer experience because of the sheer volume of customer interactions retailers field.

“They’re paying very close attention to the ‘why’ behind consumer behavior,” he added. “Those types of businesses are very forward-thinking about technologies like Artifact and others right now to explore and understand the customer experience.”

Large language models (like those created by OpenAI and others) are ultimately what makes a product like Artifact possible, Sanders added. “Large language models have opened doors that were previously scientifically impossible to open,” he said.

Ready for change: Which provider Wayfair chooses will ultimately come down to the actionable insights they’re able to provide, said Clemens.

“What would be most interesting to us is stuff that we don’t already know…that was not on our radar,” he explained. And that’s a possibility that has garnered interest from other departments.

“Everyone would welcome a sharper understanding of what the customer experience is really like and to be able to get a little bit more surgical in figuring out what the big opportunities are,” Clemens said.

But he’s not set on a particular outcome, and says he’s staying open-minded about the future of AI in the world of customer experience.

“Organizations should have…more of an experimentation mindset at this point rather than crystalizing…[what] the solution should look like,” he said. “Because it’s quite likely that in six months, the solutions out there will change.”