Balancing AI and Human Expertise in the Era of Google's Helpful Content Update

Hello, everyone. Today we have with us Jeff Coyle, Co-Founder and Chief Strategy Officer at MarketMuse. Marketmuse is the industry-leading technology and methodology for content planning and evaluation where semantic relevance.

Jeff, Welcome to the show.

Thanks for having me. I’m so excited to talk to you today.

Same here. Jeff, I know Everybody must be asking you this these days after AI and you are in the content world. But still, the first question I would like to ask you would be AI content versus human versus hybrid. What are your thoughts?

So many of you users may not know this or may know this. Markham uses an AI-first company. We’ve been around for eight to nine years. We have two patents in artificial intelligence from nearly 10 years ago. We’ve been working in the field personally for over 10 years. My background is in computer science and I built vertical search engine platforms, ad servers, and all of the above. And the question of AI versus pure AI content, human versus hybrid, frankly, not having hybrid content hasn’t been part of my life for more than a decade. It’s always been assisted. It’s to what parts of the process that it assists. That’s where the real success stories come and the failures expose themselves. A quick example would be, if you’re not using artificial intelligence to augment your research processes or your planning and prioritization or to personalize the actual plans you’re distributing to your writing resources, you’re missing a huge part of the boat because you need to look at your success rates with the content you’re creating. If you’re publishing 100 articles and only five of them are successful, you have problems earlier in the process. It’s not about the drafts.

You’re picking the wrong horses. Your batting average is too low. I would think about it as the entire process. It would be irresponsible to not use artificial intelligence at all those stages to generate the strategic needs so that you hit more home runs, hit more big hits, and put up more sixes, depending on whoever you’re listening to, whoever’s listening to the show. That has always been part of my life. Then we were the first technology that built a content brief with artificial intelligence. And that was a huge innovation. We used to make them manually five or six years ago. We built a large language model technology and put it into the market long, long ago when people were not ready for it. And it worked well. The only problem was it took about five days to create a page and it cost like $30 of compute time. Now that same quality can be built with pennies, fractions of pennies. So that changes. That’s where the answer to your question comes from. It comes from the fact that we’re at an infinite scale. So when we go to Infinite Scale, Microeconomics 101, all of the things that happen before that product is made and after are the things that start to matter most.

So the hybrid now is you have to have human expertise in the loop because of infinite scale. So hybrid is the future. It is the past, but it’s a different hybrid. It’s a hybrid where you understand what commodities are and you understand what special, unique, differentiated value you can bring, and that makes it to the page and everything else. The world of mediocre content died. That dog is dead. Sorry. It’s not coming back. It’s going to get worse. You’re going to have to create higher-quality content. You have to understand the context. You have to understand intent at a level you never thought you’d ever need to. You can’t just get by with mediocre content.

Jeff is also a co-founder of Silver Bluff Brewing Company. Jeff, how did these two different business niches happen? Beer Brewing, and content.

Yeah, I am. I’m a co-founder of one of Georgia’s fastest-growing breweries Silver Bluff Brewing Company. Among other accolades, we just won the 2023 Beer of the Year from Brewskies, which is a nation. And we’ve won a World Beer Cup, a Silver Medal, and a US Open Silver for our Mexican Lager flagship. You can buy it online, I’ll give you the link for your listeners, I think 44 states can ship. And then we’re also served all over Georgia. How did you get into that, Jeff? Oh, my gosh. It was always a plan for my brother and me to do that, and my brother and me to do this after we retired. My two passions are online marketing, search engines, usability, and beer and everything about it. I was a home brewer. My brother was as well. We won a few competitions, and there was a law that changed in Georgia, and it made it easy to open a brewery. Whereas before it was the 48th worst state to open up a jury. There was so much red tape. And in 2016, they changed the law. We both thought, Oh, my gosh, somebody is going to jump in on this.

Because where we built it, there was nobody there because it was really hard to build it. A couple who are also our co-founders came to town. They moved into town and they were like, Has anyone ever thought of opening a lawyer here? And they asked three or four people, and all three of them were, for instance, people that played soccer with me and my brother. And they were like, You need to go talk to Kevin and Jeff Coyle. And after the third person, Chris and Ali, Moline, and my co-founders, I finally called my brother and he’s like, Oh, okay, cool. And then two days later, we incorporated. And a year and a half later, we had a construction project and we opened in 2020, right in the middle of the pandemic.

Nice.

As you said, you were the industry’s first artificial intelligence-driven software. I’m sure there must have been some unique challenges. If we talk about AI now and everybody says, Yes, let’s do it. Let’s go for it. But when you were doing it like industries first, I’m sure there were a lot of opportunities, but what were some of the unique challenges you faced in the early days of developing MarketMuse?

Oh, my gosh, so much. MarketMuse tends to beat the market by two or three years with our ideas, which isn’t a great thing. You end up being almost like a change agent or a market change agent. That is not a way to be financially successful. It’s very difficult, especially when you’re talking about technology catching up. We were always pushing somebody to believe that something was a value before it was common. And that’s how product market maturity works. You have unknown introduction concepts, you have a growth phase, sorry, you have maturity and saturation. We, time and time again, came in at the intro to growth very early, and that’s very difficult and expensive too. But when you win big maturity, everybody believes something is valuable and a market starts to form, and competition starts to come. And then when something becomes ubiquitous, you have saturation. And to break out of saturation, you have to break out with differentiated values. And that’s hard. So that’s when prices go to zero. So for us, it was this constant cycle of bringing something to market that nobody had thought of yet. And then us having that early advantage and then the market waved over us because we didn’t have enough money to be the billion-dollar company.

So that’s been our cycle. It’s to say, where are we uniquely differentiated right now? Where do we think we’ll be six months from now? And what do we think we’ll get the saturation? And that’s been the persistent problem. Market confusion is the biggest problem. So your competitors try to confuse the market to suit their needs. So in search engine optimization, that’s often confused by the promise or the simplicity or the snake oil. Just do this. Do A and you get B. It’s that easy. And so what that did was for our target audience, content strategists and optimization professionals, we were the people who were recommending that you eat a balanced meal, eat your vegetables, high quality, comprehensiveness. It’s hard. It’s like the person selling you a weight loss program that requires a ton of exercise. But sometimes it seems a lot easier to just eat what you want and have a magic pill that makes you lose weight. And in our industry, magic pills are everywhere. Everyone will sell you a magic pill. I’ve optimized some websites for magic pills, frankly. And so the biggest difficulty for us is beating the market maturity curve, and then our industry tends to skew towards people selling magic pills.

News offers advanced AI and NLP features. How do you strike a balance between making the platform accessible to users with varying levels of technical expertise? Because like you said before the interview as well, we were talking like you have big firms that understand even the technical side of it, and then you have these small players that don’t understand the technical side.

Gosh, it’s so hard. Just divorcing it from the product-market maturity curve. These solutions aren’t easy to use. None of them are. Pick a platform. Most enterprise SEO platforms or product-led growth-based SEO platforms are insanely complicated. They typically average over 40 to 50 report-style views, and their grids and data. Oh, gosh. So what do people do? They buy this thing, and they end up using one report, that’s it, or two, that’s it. And so what we’ve done is we’ve analyzed usage patterns and we’ve focused on workflows. And so our entry tier right now is the profile of I have a couple of pages and I optimize them. I need to optimize them, and I write a handful of pages every month. And I value the fact that I want them to perform and I want to make them high quality. Then we bring in research and advanced topic-based research, cluster analysis, and competitive analysis. That’s that next-tier person who gets that they need to know and differentiate from their clients. But the content strategy side, picking the right winners, that’s the more advanced user right now. What we want to do, and this is what we’re doing right now, is we’re bringing the real tough stuff, planning the strategy, the competitive research, and we’re going to bring it to somebody who has had no experience working with AI to do that.

We think that the next big thing for market news is bringing content strategy to the masses. And we’re doing some cool stuff. The things that I’m working on right now are the coolest things I’ve ever built, by far. It’s wild. The technology has caught up to where if you know how to do this stuff manually, the advantage you have is dramatic. That’s where it separates market news because a lot of people jumping in the ring right now, their experience with AI is they use ChatGPT, that’s it, or they were able to get an API. They don’t have any experience doing this the hard way. I carry a book with me about how to build content inventories manually just to remind me how it already used to be, how to build knowledge graphs. Knowledge graphs are hard to make manual, topic models. And you hear these folks on Twitter or X Twitter, whatever you want to call it, and they’re telling you how easy it is to build topic models. No, it’s not. What they’re building isn’t a really good topic model. A lot of times it’s like substring match searches, which is a nerdy way of saying it, like keyword variants.

They’re not building your topic model. They’re just giving you lists of words. What they don’t realize is that when you’re using market news, what we’ve done is we’ve taken a three-dimensional complex model and we’ve turned it into a word list so that it’s easy for you to digest. But it isn’t a word list. We transform it into a word list because that’s easy to digest. How everybody else came about it, you know what they do? They make a word list and they give you a word list and they try to say that it’s complicated, but it’s not. They’re just telling you how to copy your competition because they didn’t do it the hard way. And that’s the difference. That’s the market news difference.

With all of this AI content coming in, like you said, like people using ChatGPT and so much AI content out there, how do you think it is affecting the E-A-A-T?

Is it affecting the E-E-A-T? It added the E. That’s the first effect. It added that first E. And that’s dead serious, right? Google works backward here, man. Trust. How do you discern trust? We have to figure out an automated way to calculate authoritativeness. How do you calculate authoritativeness? It’s through link modeling. Link modeling of what? Link modeling of page and site section level, topic level, and link models. This is getting a little complex, but I like to describe it this way. Do you know why? Because then you’re like, okay, cool. What if we can manipulate authoritativeness and link matrices? What’s next? Expertise. Cool. Can’t think that. So my content has to be good and the links have to tell the story that I am an expert. But what if we figure out how to manipulate that? What comes next? Experience. Does the query deserve experience? The thing that Google cracks always slaps at is figuring out intent. They know how to tell the story of intent, and they tell it through search features on the page. They test it with user patterns. I’m allowed to talk about it now because it’s public knowledge. A great deal of the way Google works is through the way users use Google and Click patterns.

They have a clicks team. Go look it up. Now that it’s public knowledge, it’s so great. The name of the Google algorithm is public knowledge. I’m not going to say it here because I don’t want to get shot in the street. But finally, I’m allowed to say these words out loud. But how does that affect it? It affects it in two ways. It makes it so that they have to keep building in the other direction. And they’re going to make mistakes. They’re going to be heavy-handed with those things. But what it does is it forces you to make a decision. Are you authentic or are you not? And, if you have a prisoner or gambler dilemma at one point in your website’s decision, you’re going to have to make the decision. Am I authentic or am I not authentic? When I decide not to be authentic, I have to recognize that if I get caught, my entire business is at risk. So if I say that I went to the park yesterday and I played with my dog, he’s sitting right here, and I didn’t, I’m deciding to not be authentic.

So if I say I went to the park yesterday and I played with this dog toy and that link links to an Amazon affiliate link and I get paid four bucks if someone buys that dog toy, and then I do that every day, 50 times. At some point, my authenticity may catch up with me. And if I was lying that whole time, bye-bye, gone. So how does this work? It forces every business to branch either authentic or non-authentic. And mark my words, 11/16/2024, there will be very few inauthentic websites getting traffic.

That was a lovely way of explaining it. I agree with you because with all these updates and algorithms and all of that coming in, it’s very hard to not be authentic. Being authentic is the right way to go about it. Talking about updates, how do you adjust your keyword research strategy for the Google helpful content update?

Great question. I have a visual aid, I could hold up, but I’m not going to. First of all, I have to understand the helpful content update. The helpful content update is a site-level weight. So the easiest way to think about it is a hot air balloon. All right, What is the helpful content update? If you do not provide helpful content updates, helpful content at the site level, it will act as a sandbag hanging from your whole site. So every article will be impacted, the good ones and the bad ones. But it will only drag you down some unknown weight. Is it a big weight? Whoa, crud, we’re going to go everything’s going down. But it might only get you a little. So you have good content, it may be that it went down a little bit or maybe it didn’t even impact it all that much because that weight wasn’t that big. But overall, you took a ding and it’s going to be difficult to bring. So how do you assess? Now you’ve got to make sure your mirror is very clear. So you wipe off the mirror and you look at your entire site and you say, Where have I put out the garbage?

How do you define garbage? In authentic content that was made, you have to assess how much of that content exists. And you got to come up with a plan. That plan isn’t, don’t listen to the snake oil, bleed it all. Not a good plan. Terrible idea. That’s like removing all of your teeth because you have one cavity. You have to figure it out. I have lots of analogies, as you can probably tell. Figuring out ways that you can make your content better, improve it, consolidate it, repurpose, expand, and then setting forth with and this is hard if you’ve created what they call editorial debt. If your entire strategy has been to build low-quality content and you’ve been impacted by this, I’m very sorry to tell you you have to update every page. And I’m sorry that’s going to be expensive. And a lot of business owners don’t want to deal with that. And that’s the thing. So first is to recognize it’s a site-level weight. It is not a boost. Maybe you haven’t been impacted by it. Other people in your world have. I’ve seen people saying, oh, I went up with the content update.

I’m like, no, you haven’t. Two of your main competitors went down. That doesn’t mean you went, Dude, come on. You don’t understand what you’re talking about. Identifying that, looking at yourself squarely in the eye, figure out a way to get something that can analyze your entire site like MarketMuse, or Bing. Not Bing, just MarketMuse. I said Bing as a Romani pia for a successful bell ringing. But no, find a solution that can analyze your entire sight. Get screaming frog or SiteBold. Have a way of crawling your site, building points of reference week over week, month over month. You know what it looks like to go through your site. Then decide what’s good and what’s bad. Don’t do anything brash and delete. Without a great plan. Start to consolidate. Get yourself on a healthy diet for the future. Don’t double your velocity and do not increase your traffic. Don’t delete everything because you know what? You delete everything. Traffic might go up. It might not. You might as well roll or die. That’s the correlation mentality that you want to avoid. You only see examples of pruning when it works. Have you ever seen people tweeting about the fact that they deleted half their site?

It didn’t work? They don’t tell you those. They’re all of them.

That basically would have been my follow-up question, but you already answered it. Is content pruning a good idea? You already answered-.

Almost always the wrong choice. Just delete in 404, there are very few instances in this entire world where that is. Delete and redirect, it’s usually a crappy solution. And then you get into the more advanced techniques. You have to know who you are to know what to do.

You work on content clusters, and how do you analyze whether your cluster is performing or how do you analyze that?

Great question. The answer to that question is very self-serving. I use a software product called Market. But let’s just think about it fundamentally. If you don’t have Market, first of all, get your checkbook out. No, I’m just kidding. If you don’t have it, how do you do it? Here’s what you don’t do. You don’t report on pages only in isolation. Your pages work together as clusters. Think of them as blobs or organisms, and they all work together. Some of your pages that get no traffic are the most important pages on your website because they are part of that cluster. So you have to report on an individual page basis and you have to report on the cluster. You have to make sure you sell internally to these people who are evaluating you. Hey, we need full funnel coverage. We need full cluster coverage. For the tip of our spear to be successful, you need the foundation of the house. The first rule is don’t just report on one, report on the collection. Recognize that some pages are not going to have direct value and also recognize that, I don’t know how much to say.

But recognize that off-page factors can influence the cluster, not just the page. So do not just buy traffic to one page. Do not just buy links to one page. How much should I whisper this? When you’re analyzing this stuff, you want to make sure you’re building the mass, you’re building the clusters so that they’re there. So then what do you do when you need to build another cluster up? If you’ve already done one well, how do you bridge the two together? Through relatedness and theming, you start to get advanced context windows and context selections. So you don’t just go, hey, I rank well for Golden Ales. And I also now want to rank for Boston Terrier Dogs. You don’t talk about the fact that when you’re walking your Boston Terrier Dog, you carry a Golden Ale. That is not cluster connecting. Cluster connection works at 6 degrees of Kevin Bacon. I’m not going to figure out those bridges to get me to the bus and tear your dog. And guess what? If you try to jump right from the beard to the dog, it’s a really bad signal. It can be really bad for your sight.

I see so much of this where people think again, they look at their competitor, they look at a famous website that’s powerful, and they think they can do what these… Don’t mimic your heroes. Your heroes oftentimes have so much authority on so many different things. They have so much authority that they can get away with some mistakes. They can get away with going from dogs to the beach. Sorry, the beer is to the dog to the beach. And the path here is to not mimic your idols. Do not copy what Amazon does. Don’t copy what your favorite healthcare site does that gets a billion websites in a month because you don’t and you’re not going to get away with it. And it’s going to be even worse. But yeah, don’t copy your idols. That’s the answer.

I know we’re short on time, but before I let you go, I play a quick rapid-fire round of three to five questions. Love this.

All right.

Perfect. How long does it take you to get ready?

In the morning? Yeah. Oh, gosh. Because it involves taking my two sons to school, making breakfast, making lunches, and all of my stuff. Dog, walk in probably about an hour, and 15 minutes.

That’s sweet. Do you make lunches for your kids?

Every day.

Nice. What is the best that you cook?

Italian. I make fresh pasta, pizza is my favorite.

For sure. Nice. Texting or talking. What do you like comfortable more?

Such a hard question. I talk a lot less now. So texting is 90 % of my interactions, and I wish it was talking.

Yeah, I always do the voice convert.

My thing is the group text is awesome. The group text, you can’t bottle that because of a group chat and lives, technical difficulties, and all that garbage. But the group text, my favorite thing, is way better than talking in some cases because you can’t get all those people in the same room. And then I still do it. I still do the manual text, the one-off text. It depends. There are way more people that I would rather talk to verbally. Right.

Our favorite ice cream flavor.

I’m bored with ice cream. Ben and Jerry is a fish food, PH, ISH, although I’m not a huge fan of the bands.

What was your last Google search if you remember?

Content Marketing World BC. I was trying to remember what day it was because I talked to my friend and I needed to look up. I hadn’t talked to her in a while, and I needed to look up what days those were so I could find this document when I found it.

If a movie was made on you, what genre would it be?

It would be a Tragic Hero’s Journey set in the ’80s, the movie for the movie structure. So it would be like a trauma-built, low-budget tragedy.

Jeff, thank you so much for your time. It was fun having you. I had a blast and hopefully, we’ll get you again for another episode.

Awesome. Love the rapid-fire round too. Hopefully, everybody knows what the heck I’m talking about. Or they can look it up in Google too. And it’s been awesome. It’s great to learn about you and the businesses you represent as well. Everybody should check those out while they’re buying my beer and my software. They can also hook you up too. And it’s been a pleasure. Cheers.

Thanks, buddy.