Enable Two-Way Conversations With Your Website Audience Using Natural Language Processing

Video Transcript (Edited for Clarity)

Prema Srinivasan: Hi everyone. Welcome to another very exciting episode of Public Interface by Velir. I’m Prema Srinivasan and joining me today is Mark Stiles. Mark has been developing applications for machine learning for the past five years. His expertise is in Sitecore web development. He's a seven-time Sitecore technical MVP and has been a featured speaker many times at both the Sitecore Symposium and the Sitecore User Group Conference in Europe. He's been developing the video series Catching Exceptions about developers in his community as well as this series Public Interface. Today we're going to be discussing the application of natural language processing to content marketing. Before we jump in, these are the types of topics we frequently discuss on this channel so make sure to subscribe so you don't miss out on future discussions.

[Public Interface Intro]

Velir Presents

Public Interface

Two-Way Conversations

Prema: Hi Mark, thank you for joining me today.

Mark Stiles: Prema, thank you for having me.

Prema: To start off can you tell us a little bit about what natural language processing is?

Mark: Natural language processing is all about taking language—a statement, sentence, or a phrase and breaking it down into its constituent parts the parts of speech—nouns, verbs, adverbs, and adjectives.

Prema: As a lifelong writer and someone who has worked in content marketing for many years, I'm really passionate about the power of the written word to inform and educate people, but one of the biggest challenges is understanding what people really want to know. Is this something that natural language processing can help us with?

Mark: I think the most powerful thing that we can accomplish with this is enabling two-way conversations with all your audiences. Right now, we do a little bit of this. We offer them forms or search and we allow them to say things but not necessarily in a natural way. So if we are able to give them the ability to ask the question in the way that they want, we can pull out context—we can pull out all kinds of information and get a more accurate result for them.

Prema: What are some of the benefits to companies of implementing this?

Mark: There's a lot we can do if you’re say, a products company, and you're selling something. Somebody's going to come to your site and potentially say “I want to buy this product with these options,” and instead of giving them search results that they have to click through go to details switch the link, and do all these steps before they get it into their cart—you can pull out all their information from their phrase immediately off of the button to add it to their cart directly. It's going to help with brand loyalty and customer satisfaction.

Prema: What are some of the driving factors that are converging to make this possible?

Mark: Most importantly, I think what's tying it all together is that we have these really large software companies who have the money to invest into it and they are building these teams, they're building these products, and then they're offering them to everyone and anyone at a very low cost. So for somebody who can only pay a certain amount to be able to take something off the shelf versus hiring an entire team of data scientists, this is a really good cost-effective way to get really powerful algorithms at your disposal.

Prema: what digital challenges does this address?

Mark: A lot of it is just in the fact that we are not really adapting as well as we can to our customers. We build these websites and we put our information out there, we try to track some analytics and we still need a lot of analytics data to figure that out and somebody to analyze it. But on the other hand, if we were able to ask our audiences explicitly, we can get a lot better information—a lot more accurate information—by just asking them directly.

Prema: What's involved in getting this launch-ready for a site?

Mark: The challenge here is really starting to figure out what things matter to your business and building language around them. And being able to write a lot of different verbose phrases to train this algorithm. That's the challenge. And what happens is over the first few months you're going to start to see people ask for it in slightly different ways but eventually it tapers off because there's only so many ways people are gonna ask. Just kind of managing that flow and categorizing the phrases into the specific pockets that's where the work is.

Prema: That's really great information, Mark! Thank you so much for taking the time to speak with me today!

Mark: Absolutely, thank you.

Prema: That wraps up today's episode of Public Interface. Thank you for watching! Join us next time as we continue to explore other exciting topics from our work here at Velir. In the meantime, follow us on Twitter (@Velir), subscribe to our YouTube channel, or visit us online at velir.com.

[Outro Music and Closing Credits]

Special Guest: Mark Stiles
Hosted By: Prema Srinivasan
Produced By: Velir
Directed By: Mark Stiles

Velir
Connected Digital Experiences
velir.com/pi
[email protected]
617.491.6900