John McAlpin:“LaMBDA is a machine learning technology language. They’re trying to train computers to understand language a little bit better. It’s to be used for chatbots and AI, so that we can reduce the need for people to have conversations with customers.”
Read the Transcript
Announcer: Welcome to the Ignite Podcast. The only healthcare marketing podcast that digs into the digital strategies and tactics that help you accelerate growth. Each week, Cardinal’s experts explore innovative ways to build your digital presence and attract more patients. Buckle up for another episode of Ignite.
Alex Membrillo: What’s up everybody. I’m Alex Membrillo still CEO of Cardinal Digital Marketing somehow. Today, you guys are in for a treat. We’ve got John McAlpin, the SEO director, here at Cardinal and he’s a master of all things technical, wizardry, innovation, and trends in SEO. We’re going to be learning a lot from him. Google recently dropped a big change on us at the usual user conference talking about conversational search. Let’s get into it. John, welcome to Ignite.
John McAlpin: Thanks for having me back.
Alex: At the I/O conference they foreshadowed that search will become more conversational and less direct query-based. SEOs love going crazy about this kind of stuff. Is it something we should be concerned with but what the heck is it to start with? How do we explain this in layman’s terms?
John: Search has been going more conversational since it began. Back in the late ’90s, early 2000s, we were all searching with not real language, we were searching with grocery store near me or grocery store Dallas, Atlanta. It was very chopped up together and they’ve been working overtime on developing algorithm we can ask more long-tail questions. Now, it’s getting more conversational whereby that means you can literally have a conversation with the search question, “Okay, where do I go to find the best grapefruit near me?” and it will pull up maybe some health food stores or natural stores or organic stores or something like that and it will become a lot more natural.
Alex: That’s the basic thing like you say, “Siri, I’m not feeling well,” and she’ll recommend urgent care instead of saying urgent care 30319 or whatever it is.
Alex: There are two different conversation technologies. Tell me about this Language Model for Dialogue Applications, LaMBDA. What is this and what is LaMBDA?
John: LaMBDA is a machine learning technology language. They’re basically what they are doing is they’re trying to train computers to understand language a little bit better. It’s to be used a lot for chatbots and AI so that we can reduce the need for people to have more conversations with customers when we can’t field a lot of the customer questions before actually coming into your of care operation basically helping create more qualified leads down the road. Now, this is all developing technology, it’s not rolled out or anything.
They showed some really great examples of what it can do how powerful it is. Two examples they showed was one, they trained a computer model to learn everything about Pluto, everything that we know about Pluto, and then they had a conversation with Pluto asking all about like what’s the temperature like? What’s it like being in this planet? How far out are you? Who are your neighbors? They even trained a computer model to learn everything about a paper airplane and they had a conversation with the paper airplane. I think what’s the best type of paper airplane to make? How do I make one? How do you make one fly the farthest? How does it get to fly real straight what’s it like? What’s it like to fly? It would explain everything on what it’s like to be a paper airplane. It’s a really powerful technology that they’re developing and we may see it roll out over the next few years and a lot of different variations.
Alex: What does that help when you ask about an entity we don’t have to do in such a proscribed manner and to say, “Tell me about this,” or, “Tell me about that,” or, “I’m wondering about and because it understands the subject in detail. Is Google going to have to train their algorithms on every single noun? Like every–
John: This may not necessarily be used in the traditional search that we’re looking at. This is more in technology of question and answering. If you have a Chatbot let’s say on your website and you train the Chatbot, “These are the main types of questions you’re getting and here are the answers.” In this conversational technology will help maybe guide the user to these preset answers a little bit easier because you won’t have to type in a specific set of questions. Nowadays it’s okay if they ask these questions, give this answer. We’ll now be able to pair maybe a different question to the right answer a little bit easier.
Alex: What is this Multitask Unified Model, MUM? What is it? Is this different?
John: This is very similar to what we talked about last year with BERT, which is actually, it’s again, not rolled out yet, but they’re working on implementing it into the algorithm. It’s another AI basically, help them understand questions a little bit better and string together more complex questions. We saw this a lot with entities where instead of crawling, and I’m trying to understand the whole page content, it’s pulling out individual entities and stringing them together by relationships, entities being any kind of noun really. A person, place, or thing. There’s a lot of technology going behind MUM and it’s really leading the end to allow users to speak more conversationally and more long-tail queries and help them better understand these new long-tail queries that are coming up because when Google started rolling out in this past year is a new feature in search where if your question isn’t answered, it’ll actually say, “Hey, we couldn’t find a good result for your question.” You might want to try rephrasing here, an article about search operators. You can search more effectively. They want to reduce that effect and serve better results. And so that’s why they’re investing more into these conversational search technologies.
Alex: Do you think the future of search is conversational?
John: Absolutely. It’s been going that way. That’s where Google’s headed. They’ve been hinting and talking about it forever.
Alex: What do marketers need to do to prepare for that? Do we stop keyword stuff and what are we doing?
John: Keywords will always be important. There’s been a lot of talk about the decrease in the importance of keywords, but without keywords, we don’t know what people are asking or searching for. Without the right kind of content, we’re not going to know. Now, how you use keywords will definitely change. It has been changing. It used to be like you said, we used to have the stuff keywords everywhere to make sure it was put a certain amount of times and every variation was there, but now we can use synonyms a lot more and not sensitive, but there are other variations of keywords, similar keywords and speak, maybe even more technically. Especially in healthcare, we’ve had to use common names for certain procedures or medications, and we may be able to actually use the more formal names and so that Google will understand that they’re the same thing. Before we had to dumb down a lot of healthcare information for users and we still do to a certain extent, but now we can come across as more authoritative with proper language and we don’t need to always find the substitute keywords just because of volume.
Alex: How does this impact UX? How do you look at pages differently? You got to find the right bounds of content or conversion-friendly stuff, the right design, like, how are you looking at the way things are going as it relates to the design of a page the experience?
John: I mean, design a page and content are going to go hand in hand. It all depends on your audience because since Google is trying to start the best results for that audience, based on how they form their questions, we needed more of our websites into that. If we’re targeting providers, if for referrals, if we’re targeting patients for direct acquisition, we need to make sure we serve the user journey to what their audience profile is.
Alex: UX always in– John, you remember the good old days and like, “Oh wait, it didn’t– UX wasn’t a thing. Every website was bad.” So we just had to book your words on a page and rank and– [crosstalk] what?
John: Flash was still allowed?
Alex: A lot? That’s right. I don’t know. I forgot about flash. It’s throwback Thursday. We’re recording this on Wednesday, but whatever. That’s good, any final thoughts on conversational search that you would want people to know before everyone freaks out and goes and changes everything?
John: Of course. You’re going to get a lot of articles, how to optimize for Lambda and MUM, and how to optimize this. Take a step back. Remember, think about your audience. Always write for your audience. Try to satisfy every– try to think about every barrier to entry they have every question they can have and try to provide useful information back to them. You always remember that and try to achieve those goals first, you will wait. Remember, it’s not about optimizing for where Google is now, because last year everyone was talking about BERT and now talking about MUM and Lambda. Think about where they’re going and trying to optimize for where they’re going. Google has always been heading towards conversational search. Don’t try to optimize your pages for short keywords and quick action, super long, just for the sake of length. Optimize for your users and you will still win.
John: See you next time.
Announcer: Thanks for listening to this episode of Ignite. Interested in keeping up with the latest trends in healthcare marketing? Subscribe to our podcast and leave a rating and review. For more healthcare marketing tips, visit our blog at cardinaldigitalmarketing.com.
[00:10:23] [END OF AUDIO]