AI Search: The Next Big Thing In Customer Support

AI Customer Support

As many employees and leaders in Customer Support already know, there are pros and cons to every service model.

Build out a team of agents, and you’ll give your customers the impression of being “high-touch.” But high-touch can mean high-cost: according to Harvard Business Review, the average cost of a live support interaction (phone, email, chat, etc.) for a B2B company is $13.

Does every problem need an agent? Gartner reports that 40% of live support interactions could be resolved in self-service channels. Naturally, there are some customers with particular needs who require the human touch, but what’s the ROI on live agent support for customers with simple requests?

The other end of the spectrum is scaled support: providing service via an automated phone system, chatbots, FAQs, etc. The approach here — to provide modern solutions for savvy customers in cost-effective ways — makes sense. The problem? Execution.

According to Gartner, 70% of a business’s customers use self-service tools, but only 9% end up fully resolving their issues. Why is it that saying — well, more like yelling — “agent” into your phone repeatedly or desperately hitting “0” on the keypad is the standard? It’s not because customers actually want to get redirected five times before speaking to the right agent. It’s because many of these self-service options simply don’t work. And customers know it.

Whether dealing with a person who isn’t much help or an automated system that leads nowhere, bad customer service can be a deal breaker. In fact, Forbes says that poor customer support is costing businesses $75 billion a year.

But all hope isn’t lost.

Dimension Data says that 73% of customers prefer to use a company’s website for support. And according to Parature, 84% of customers want to resolve their own issue using search, before raising a support ticket or calling customer service.

So the $75 billion question is: does your website have the search chops to scale your service model, without compromising quality of support?

Transforming your website into a support hero

Search on a business website isn’t a new thing. In fact, it’s quite old — outdated, in fact. Millions of businesses are using keyword search to “power” their websites. The problem is that the technology behind keyword search hasn’t evolved since 1999. Today, when you type a question into a search box, you get back a list of hyperlinks that contain the keyword, but lack the context. In other words, the search bars on most business websites don’t know what you really mean.

But consumer search engines — think Google, Bing, DuckDuckGo — are a different story. Over the last 20 years, Google has led the charge on making consumer search more robust. Instead of only using a single algorithm to scan for keywords and return blue links which may or may not be relevant, Google has incorporated natural language processing (NLP) to understand questions and intent to deliver actual answers.

So the key is to ensure that the search powering your website is modern, based in AI, and thus designed to enhance the customer experience — not take away from it like keyword search. Features like NLP and extractive QA (an algorithm that allows your website to deliver featured snippets of information from long-form and unstructured data like product manuals, tutorials, or ebooks) ensure the rich content about your business is accessible for your customers to actually self-serve when they have an issue.

This benefit applies to your own support agents as well. Beyond affecting the bottom line, having information spread across a variety of systems affects productivity — 36% of a typical knowledge worker’s day is spent looking for and consolidating information. By equipping agents with AI search, they can deliver the most helpful information in a timely manner, providing a better experience for both themselves and the customers they’re serving.

As a last line of defense, this modern search experience can also be embedded as a customer fills out a support ticket. In real time, AI-powered search can turn a customer’s ticket description into a live search query intended to deliver helpful information, meaning customers no longer need to submit a ticket because they got the answer they needed through search.

Expanding the reach of your content

Going back to our two key stats, we’ve addressed the 73% of customers preferring to use a company’s website for support. As for the 84% who want to use search, your website search is a critical piece of that, but not the only one.

Any discussion about search has to include third-party engines, i.e., primarily Google. You can pay Google for higher SEO relevance, but 1) get in line and  2) you’re playing in the game — the AdWords engine —  that created other problems to begin with.

This isn’t meant to be a deep dive on search engine intricacies, so the bottom line is: branded search on Google isn’t helping your support-seeking customers. It’s serving them ads about your company and even competitors. There’s nothing more frustrating for customers than to be sold to, before getting their issue resolved.

With AI-powered search for support, you can use your search-optimized landing pages and synced FAQs to influence rankings and relevance on Google and other search engines, and even increase the chances of appearing as a featured snippet. When your business controls the messaging, your customers have a better chance at getting helpful results, instead of a sea of ads.

To learn more about how AI search like Yext’s Support Answers can help scale your service model and improve quality of support, visit:

About the Author

Marc Ferrentino, Chief Strategy Officer, YextMarc Ferrentino serves as Chief Strategy Officer of Yext. He is responsible for product management, user experience, product marketing, corporate strategy, customer insights, platform developer program, and verticals.

Prior to joining Yext, Marc was the Founder and CEO of Nomi Technologies, which was acquired by Brickstream in 2014. Marc has worked as a senior executive for high growth tech companies throughout his career, including serving as the Chief Technical Officer of SaaS at BMC Software, and Chief Technical Architect for

Marc holds a B.S. in Electrical Engineering from the University of Michigan and has participated in the M.A. of Statistics program at Columbia University.

Leave a Comment