How UNC Health Makes Site Search More Google-Like

// By Jane Weber Brubaker //

Can AI make Find A Doctor functionality and site search work better for consumers?

Think back to the earliest days of online search, when keywords first came to be. A typical call to action — in a TV spot or a print ad — would be something like this: Visit us at www.pups.com or type in AOL keyword “WOOF.”

How far we’ve come! Today we can type long, complex phrases into a search box and find pretty much exactly what we’re looking for on Google. Machine learning is running in the background, getting smarter and smarter.

What if health system site searches could be that good? With a form of artificial intelligence, natural language processing, it’s becoming a reality.

Carrie Liken, head of industry for healthcare, Yext

Carrie Liken is head of industry for healthcare at Yext, a technology platform that drives search experiences on sites — via find a doctor, site search, app search, and more — and off-site, via listings. She says, “In the process of developing the listings business, we had connected via API to every single one of the different publishers that we power this information for. We realized we’ve built a relational database.” With this realization, a lightbulb went on. If they were able to power external search experiences on Google, they could build Google-like search experiences for customers, too.

“One of the really attractive things about Yext was this idea of a knowledge brain of some sort of natural language processing behind searches that starts to mimic more of a human interaction,” says Matt Schwabel, executive director of marketing technology at UNC Health.

UNC Health’s former system was very much keyword based and if you used the wrong keywords, you might not get any results, a frustrating experience for site visitors.

“The ultimate goal is to be relevant and personal without being creepy. That’s a fine line to balance,” Schwabel says. Read on to learn how UNC Health is approaching the challenges and opportunities of using AI-based solutions.

Log in below to access this content: