Conversational AI: Search, Chat, and Great Customer Experiences

The potential of conversational AI has been a hot topic for several years, but the reality of AI's impact has recently become clear to the masses — thanks in large part to ChatGPT. Recent advances in large language models (LLMs) have made things that once bordered on science fiction possible, and these fundamental technical breakthroughs have made conversational AI more capable and useful.

Businesses are paying attention: many are shifting budgets to invest more heavily in AI. AI-related titles within organizations are becoming commonplace. And as we wrote recently, there is no digital experience today that will not eventually be converted to or augmented by conversational AI.

But what does it really mean for businesses to have AI-powered interactions with customers? And what technologies should companies focus on in order to create meaningful — rather than simply "novel" — conversational experiences?

For the bulk of the digital era, businesses have "talked" to their customers in two primary ways.

1. Two-way human interactions: customers could have conversations in-person or over-the-phone with store associates, service reps, and the like.

2. One-way digital interactions: customers could sign up to receive email communications for a business, look up information on a website, or consult a wide variety of third-party digital platforms (Yelp, Facebook, Google, etc).

Now, that's changing.

With conversational AI, the opportunity for brands to speak to consumers or their employees will move beyond the indirect world of updating third-party platforms, to a more direct, two-way interaction. But in order for this to happen — i.e. for conversational AI to be able to answer business-related questions — it needs to be able to draw on all of a business' facts and information in a centralized, organized place.

In other words, businesses will need their own knowledge graph — which can then be combined with a conversational AI interface. The knowledge graph serves as a platform that is designed to "feed" answers to AI systems that can actually "converse" with customers — while still allowing businesses to maintain human control and oversight over exactly where each answer originated.

When we talk about "AI systems" here, we're talking about the wide variety of AI-powered search experiences, voice assistants, and chat interfaces that consumers interact with every day — with more emerging all the time.

That's a lot of information. But for now, it's helpful to focus on two critical technologies that businesses should prioritize in order to deliver great conversational experiences: chat and search.

At their core, chat and search both enable businesses to provide answers to customers who are looking for information in an interactive way.

But they each excel in different areas — and they can best be used as compliments to each other. In fact, chat needs to rely on a search algorithm in order to find relevant, up-to-date information. This is why, for example, Bing's chatbot can answer questions about current events while ChatGPT only knows about the world up until 2021.

Search strengths:

• Search is better for browsing experiences. For example, you wouldn't want to use chat as the primary interface to your eCommerce site or your store locator. Those are visual browsing experiences, ideal for search.

• Search makes it easier to merchandise/hard-code results based on business logic. This is quite difficult to do in chat — though not impossible.

• Search can have UI widgets like facets, filters, sorting, and maps that help the user navigate content.

Chat strengths:

• The user inherently expects chat to be a slower, more conversational experience. A longer anticipated response time means you can use bigger, more powerful models to generate more detailed answers.

• Chat can take the history of the conversation into account. This means users can ask follow-up questions and dig deep into certain topics. (E.g. chat could help answer a customer's question about a return they made last month, or a purchase they inquired about previously.)

• Chat is capable of synthesizing multiple search results to provide an answer. For example, someone could ask "What's the difference between the US Return Policy and the EU Return Policy?", and chat could tell them the answer by looking at two different documents.

Additionally, many digital experiences in the future are going to blur the line between search and chat, combining the best aspects of both. Many search engines like Bing are already announcing new features that combine search and chat into a single interface, allowing users to have a free-flowing conversation about search results.

At Yext, we've focused on search experiences for a long time, and you can see that through Yext Search. But conversational AI isn't only about search — and it's with that in mind that we've created Yext Chat, a new product that will empower organizations to create conversational experiences with cutting-edge AI.

We've written before that the "right" technology to focus on between search and chat is… both. That's truer than ever: there are massive benefits to each that businesses need to start incorporating into their strategy.

Businesses can't afford to lose any opportunities to drive engagement based on poor experiences, and it's important that they leverage AI to communicate with customers the way they want. (Here's a fun stat: 74% of users say they prefer chat while looking for answers to simple questions.)

With Yext, any business can immediately hit the ground running with out-of-the-box AI in Yext Chat, Yext Search, and various other aspects of our platform. We'll continue to innovate on our products as new and better language models emerge, but Yext Content will always give customers the freedom to use their content with any model they want, in any way they want, regardless of what lies ahead.

Ready to explore? Learn more about AI and the future of digital experiences with Yext's new virtual event series, and explore Yext Chat at yext.com/platform/chat.