How AI Will Make Content Marketing Better (And Why a Robot Probably Won’t Steal Your Job) With Jeff Coyle From MarketMuse [AMP 226]
The blog post headline analyzer will score your overall headline quality and rate its ability to result in social shares, increased traffic, and SEO value.Test every headline before you publish. Try the Headline Analyzer »
Generative Pre-trained Transformer (GPT-3) and artificial intelligence (AI) first made their way into the marketing industry last year. There’s been a lot of talk about whether or not robots will take over content marketers’ jobs. Probably not, as long as they’re doing quality work. Machines like a degree of certainty.
Today’s guest is Jeff Coyle, co-founder and chief strategy officer at MarketMuse, an AI-driven content planning and strategy platform. He knows how AI is and will affect content marketing by making our jobs easier and our work better. What do content marketers need to know to prepare for the inevitable future?
Some of the highlights of the show include:
- Horizon Scanning: What are we going to be doing a year from now?
- MarketMuse: Sets the standards for content quality by investing in clients
- Hype Cycle: AI’s influence on writers’ jobs isn’t going away, but ramping up
- MarketMuse’s Mission: Improve content quality on the Internet
- Content: AI is making writers more efficient and giving more time to be creative
- Cost of Content: Two-order disconnect to value and revenue
- AI: Takes guesswork out of things that drive anxiety, figures out quick wins
- Stage One: Experts have blind spots and forget stuff
- Stage Two: Create content briefs
- Stage Three: Insight is personalized prioritization of content
- AI Branches: Natural language processing provides insights and editing lens
- Competitive Analysis: Publish page on concept, don’t copy or make good enough
- What comprehensiveness means? Creating dice-roll content by copying others
- Wisdom: Knowing what was regurgitated, commonly used, and SME aspects
- Correlation Hangover: Susceptible to competitive risk
- Content Efficiency: Do you know it? Document processes, post-publish tuning
Ben: Hi Jeff, how’s it going this morning?
Jeff: Hey. Thanks for having me. It’s great to see you.
Ben: Likewise. I’ve been familiar with MarketMuse for a long while, a super interesting company, doing really cool stuff. I’m really excited to bring you to the show. But for our listeners who may not be as familiar, would you mind introducing yourself and explaining what you do at MarketMuse?
Jeff: Sure. My name is Jeff Coyle. I’m the co-founder and chief strategy officer for MarketMuse. My current role has been a lot of things. Obviously, when you start a new company and you’re going to be focusing on a market that you know doesn’t exist as it was, in 2014, 2015 when we got started with MarketMuse, we’re doing everything. But right now, my goal and my role as a chief strategy officer are focused on horizon scanning—what are we going to be doing a year from now, what markets are we going to be pushing into, where should our innovation investment be.
A lot of the things that I’m focused on right now are data offerings, natural language processing, and generation innovations. We build all of our own stuff. We don’t rely on external partners. Those are major investments that we have to plan out pretty far in the future. I’m also doing things like this. I’m on a podcast.
We have our own content strategy webinar series. As I probably mentioned, my main focus at MarketMuse, which we are a content intelligence platform that really sets the standards for content quality—what is content quality, how do we build high-quality content, how do we build the right strategy for creation and updating—is to really focus on the client and focus on our audience, even if they’re not clients. How can they ensure that they’re getting the most out of their investments in content? That’s what I talked about really every day. It’s making sure that those numbers are going up into the right for everybody.
Ben: Very cool. Great stuff. What we’re going to be talking about is the future of content marketing and AI. It’s been a hot topic in marketing circles for quite some time now. You and your company are situated at an interesting time and place with as far as that intersection of worlds goes. It seems like the hype has died down or the outright fear has died down a little bit but is AI coming for writers’ jobs? Are junior copywriters going to get wiped out by a robot in the next five years?
Jeff: The hype died down from the first wave, and then the second wave. But frankly, there will be another hype cycle, and then there will be another hype cycle. The various aspects of artificial intelligence that influence writers’ jobs aren’t going away. It’s only going to ramp up. But is it coming for their jobs? Not really.
What I like to say is one of the missions of MarketMuse was to improve the content quality on the web. We always say we want to raise the bar for content quality, and what it means to write high-quality expert content. But the flip side of that is we want to rid the world of bad content. If you’re a content writer and your job depends on you writing massive quantities of low-quality content, yeah, your job’s at risk. Probably it was at risk about nine months ago. If you look at the standard writing network industry, the bar is already slowly rising.
You see a lot of the writing networks going heavy into managed services, heavy into client relationship building, and a lot less of that clearinghouse level, order with the word, get a page, the experience is becoming less and less. That’s going to start to fade away.
If you’re heavily reliant on one day writing an article about the Yeti blue microphone, and the other day writing an article about the Maine Coon kitty cat, and you don’t actually know anything about it, that’s going to make your life a little bit difficult because that’s going to start to fade away.
But overall, what it’s doing is it’s making great writers or even junior copywriters, as you mentioned, much more efficient, and giving them more time to be creative, use their subject matter expertise, or use their research skills that are external, and a lot less of the more painful aspects of content creation.
I see this as a renaissance time for writers because we’re going to be able to get more out the door. It’s going to be able to perform better. They would be more marketable because their content is going to be less painfully rigid and generic. It’s going to be able to be tuned by the fact that they were able to spend more time being creative as a result of artificial intelligence doing some of the heavy lifting, painful tasks.
It’s going to require everyone to be more creative. It’s going to raise the bar for meets-the-minimum for content. But in my perspective, if you’re a writer and that doesn’t sound like a wonderful situation, maybe you shouldn’t be a writer.
Ben: It’s hard to argue against that, but it really sounds like this is going to improve everyone’s experience on the internet, whether they happen to be aware of why their experience is improving or not.
Jeff: That’s a clear thing. One of the things that I see most is content already in many businesses has a two-order disconnect to value and revenue. Some people don’t have the real value of content. They don’t understand the true cost of content. They don’t understand how it connects directly to revenue. Maybe it’s the early stage of the funnel. What compounds with that is content efficiency rates. Most businesses are so low in their content efficiency meaning, how many articles am I publishing or how many update motions, or updating and expanding motions am I doing every month, quarter, or year, and versus how many of those are actually successful to hit their KPIs.
If you’re going out and you’re writing 10 articles or 100 articles. Let’s say 100 articles, and only 10 of them are successful. That also creates the scenario where writers are in constant pain because 9/10 of their work fails. That’s a common situation. You’re spending all this time, and then it’s just like this poof of smoke in the air that doesn’t yield any benefits. Thinking critically about if a writer could write 100 articles and have 40 to 50 of them have a chance of being successful, and how that changes their life, it makes their life great. It makes instant gratification.
It’s like, I’m confident in what I’m building. That’s really another big wing of what artificial intelligence is doing for writers. It’s making them write a lot fewer things that don’t have a chance of succeeding. Where does that come from? It comes from people relying on brainstorming, or subjectivity, or other things to guide their creation when it really should be experts and data that should drive what we write.
Ben: What it’s going to allow writers to do is peer into the black box of what is it that people want, what is it that search engines are going to reward which should follow what people are looking for if the search engine is doing its job. It sounds to me like you’re going to take a lot of that guesswork out that gives a lot of marketers a lot of anxiety.
Jeff: That’s exactly right. I want to steal that line because it’s perfect. It’s taking the guesswork out of the things that drive anxiety. Spending a great deal of time right now on a content item without that predictability, without that confidence is scary. If I spend 10 days working on a piece and then post it, and it’s like a poof of smoke in the air, my job is probably at risk. That can happen less.
You then end up making excuses for that, we weren’t really focused on direct traffic for this one. This was more of a support content item. You don’t have to do that post-publish. You want to be able to declare that pre-publish and say, hey, this is actually a support vehicle. This is supporting our main pillar page. This is supporting one of our core landing pages that we feel is a little bit at risk because of our competitor.
That’s what AI can do. It can figure out those quick wins versus those infrastructure pieces, versus those risk avoidance pieces. It can actually help to define why you’re writing this. There’s a lot less with playing the results. Don’t play results with content.
Ben: If we’re being honest, we’ve all been there. If you’re a marketer, you understand. We’re fortunate here at CoSchedule. It’s been a little bit less of an issue. It’s understood here that you’re not going to hit a home run every time you step up to the plate. Nobody’s batting 1000, but the closer you can get, the better. Wouldn’t that be great?
Jeff: That’s exactly what we’re focused on. It’s really getting that predictive lens. We are getting in the predictive lens. We’re moving people from 10% to 40%-50% on those numbers. To me, that just makes me feel so good because I know how painful it is. I’ve written pieces or had the idea for a piece that then got published. It’s a beautiful piece, but it’s nothing. Nothing happens. That hurts.
Knowing that ahead of time that this is a hashing piece versus this is focused on performance, can make a big impact on the team morale. Team morale is so key to content work. No one knows that more than CoSchedule. You’ve got to be operating like a machine and when there’s this fear of failure, machines don’t work with the feel of fear of failure. It just doesn’t work.
Ben: No, I think machines like a degree of certainty.
Ben: If you just feel like you’re taking shots in the dark, it’s hard for people to feel that their work matters or to tie it to results, or in the long term, to keep their jobs if they can’t prove value. It’s all very important stuff.
So far, we’ve talked a lot about some very forward-looking, future-thinking stuff about what AI is going to mean for content marketing, whether people are aware of it or not right now like AI is with us right now in this space, with the things that you’re doing at MarketMuse. Whether using MarketMuse or using other solutions that might be out there—I don’t even know if there are other solutions that are similar, to be honest—what can AI do to help with content planning right now? What is within marketers’ grasps right now to use AI to start taking the guesswork out of constant planning, and to really go into each piece they publish with confidence that it’s going to produce results?
Jeff: People are getting more comfortable with the concept of building content briefs. I think 4-5 years ago when we launched our first content brief solution, not everybody had adopted them. That’s the story of the journey of how AI can help content planning and content strategy, and actually, the writers. Nobody was used to these things. The first time you used a voice search, your Amazon Echo, and you asked a question and it wasn’t terribly successful, you’re like, this is terrible. Now, it’s pretty wonderful.
The same thing happened the first time you used the spell checker. Now, you’re using Grammarly or Hemingway. It’s native to your experience. That’s happening slowly with various workflows. The first one of how AI can help with content planning is giving you insights on ways to make this page at the page level better.
We do that by looking and saying, if I were a subject matter expert, I truly knew this concept, I was the deepest domain expert on this, what are the things that I would naturally have included in this piece? Providing that as a heads-up display inline is the most helpful workflow that feels natural to a writer actually giving them the same type of experience they might expect with Grammarly, but with concepts that if woven in elegantly, will allow them to exhibit subject matter expertise.
It will allow them to avoid blind spots. That’s the primary use case that everyone can quickly adopt. Give me some tips so that I don’t forget any of the concepts that an expert would write about because we’ve even done this at scale with experts and experts have blind spots, too. They just forget stuff whatever the source material is. That’s stage one.
Stage two is helping you to create content briefs—building those outlines, knowing what concepts to cover, internal and external links, questions to answer, giving you a research context that you can construct your own content brief. That is a manual process that takes you on average 12 hours for someone to build one. Even though they might say it’s 3, it’s about 12 if you look at all the feedback loops and such. That’s another one.
The third one, I’ll mention without even looking at generation as a concept which we can get into a little bit later, is the insight that is personalized for you. The big gap in the game right now is that sources provide third-party data that doesn’t impact you. It’s not talking from the lens of your seat. It’s talking from their seat and saying, here’s how powerful the competition is. Here’s how many links they have. It is not evaluating that in terms of what you’ve done, your content inventory, how much content you publish, what subject matter expertise you have, and on what topics.
What AI can help with now is to personalize the prioritization of content to say, you’re already an expert about social media planning. Based on the information I know about where you have authority, here are the areas that are opportunities. Here’s your competitive advantage. Here’s the stuff that maybe if you want to get into it then, you’re going to have to invest a lot to build a foundation.
It’s building that predictability, building that understanding and expectations about how much is needed to succeed. If you think about that as far as strategy, what should I build? What should I update? In the middle, what is that execution phase, that briefing? Then actually executing, giving insights while I’m inline. All of those pieces of the content creation updating game can be influenced by AI to speed up that process and improve your hit rate.
Ben: One takeaway that I think is important to focus on here is that AI is going to raise the bar for content marketing, and that is if the promises of AI are constructively delivered upon. Eventually, lazy applications of the skyscraper technique and other content writing shortcuts are going to stop working as well as they once did. A lot of that stuff is already starting to deliver diminishing returns.
But if that news happens to give you some amount of anxiety, or maybe even if it makes you want to celebrate it because, good riddance, we’re going to have tools that are going to make it more possible for marketers to more easily create content that stands out and provides better experiences for consumers. If you’re worried about the future, those worries might not be as necessary as some may have once feared. If you’re optimistic, it really sounds like that optimism may be well warranted. Now, back to Jeff.
Once a team has their planning operations together. They’ve incorporated AI into their content planning. They’ve gotten comfortable with that process to the point where it becomes like spell check in Word, you don’t think about it, which is a fantastic long-term vision.
The first time someone says spell check, that probably seemed just nuts. When it comes down to the actual nuts and bolts of constant creation though, and you touch on this a little bit, what are some specific things that AI can help a writer do that might one day feel as natural to them as spell check, or as a built-in thesaurus, or as any of these other word processing things that seem so ordinary to us now?
Jeff: Yeah. It’s all the fields or the branches, they call AI’s branches. It’s all in the branches of natural language processing, which is the basic processing of natural language as it sounds obviously. The insights that can glean. Some of the applications that you’ve mentioned, to become resident or native are giving insights as I’m authoring. Being able to give me insights as I’m authoring to say, hey, these are some other angles you could go down, roads you could go down with this piece. Here are some sections that you might want to consider adding. Here are some intent profiles that you didn’t address. Do any of them fit with your editorial perspective? It’s providing insights there. It’s providing an editing lens.
The other one that is maybe forgotten about is the competitive analysis dynamic. Why would I ever publish a page about a concept that isn’t equal to or better than my competitors on any topic from the standpoint of quality and comprehensiveness? Why would I consciously put something out that’s weak? That question is in my mind to ask your content team.
Why would we ever put something out with the ability for us to know that it meets versus what’s out there? Many people have just been happy with good enough. Good enough, first of all, is not good enough. You don’t have to live with that. It’s out there. It’s so easy to access data that would allow you to say, the most comprehensive, highest quality piece is written by CoSchedule. Let’s at least make sure we knock out a page that’s better than theirs, every time we publish on this concept.
That’s a possibility today, and that’s something that can really move […] It’s not that you’re copying, that’s the key. Don’t copy. Copying is terrible not just because it’s not a good idea, but also because some of these larger entities and publishers are playing against different rules. If you go out and copy Amazon or one of these big box publishers, you’re going to be sorely mistaken in the outcomes. That’s not how you win. You win by building higher quality, more expert, focused content that appeals to all stages of the buy-cycle.
Artificial intelligence can help guide you to get to that reality with every point that you publish. That’s where we’re seeing teams level up quickly. They’re saying, hey, we’ve got new standards now. We have AI in-house. We’re never publishing anything that doesn’t at least hit these bars. That’s something that everybody can do literally today is set the bar higher with everything that they publish.
Ben: Something I’m curious about is the manual kind of cheesy way of making sure you have the best thing out there. Just go look at what other people have done and just mash it all together. Rewrite it in your own words. How does AI help marketers go beyond doing that? That’s a common misconception about what comprehensiveness means when all it’s doing is just recycling what already exists, whether what already exists was ever accurate in the first place. How does AI help you get past that? How does it help you? The real question I have in my mind, can AI help bring in insights that you maybe wouldn’t be able to easily stumble upon without it?
Jeff: That’s a dream question for me. It really differentiates how not to do this. This creates what I like to call dice-roll content. The hit rate on content that you do by copying others is as likely to perform as if you didn’t. That’s what we’ve seen time and time again. The example I’ll give is if you could truly know what an expert knew on the topic, someone who lived their life evaluating this and knowing everything there is to know and you were able to cross-reference that against maybe those top pages, that’s when you find true wisdom.
Wisdom is knowing what was regurgitated, and knowing what is commonly used, but also knowing what aspects a subject matter expert would know that aren’t. That’s their differentiated value. As Garrett would say, that’s your blue ocean. It’s your differentiated value. It’s your distinguished value. It’s something that if you had spent 20 years in the game, you’d know you should have mentioned it. But the publishers who wrote the top ten ways to XYZ it or my guides to X because they’re on popular platforms, they didn’t cover that.
By the way, if you go copy those people, you’re sorely mistaken. That is one of the biggest pitfalls, and one of the biggest misconceptions of search engine optimization. It’s that you should go in and copy your top two competitors, top three results, and make your page out there. It doesn’t work. It’s a dice roll. It’s not predicted. Understanding where your differentiating value is is the key to making that successful and also adding your creativity especially if it’s something you’re passionate about. I’d much rather see sections of that page focused on that than focused on copying. It just doesn’t work in the long-term.
Ben: It sounds like the vision is that AI is going to help marketers and writers go deeper on a topic rather than just wider. If you are pretty much just taking things one step above scraping everything that exists and then just changing it enough so that Copyscape or whatever won’t ding you for plagiarism, it’s pretty clear to see how AI can help break writers of that tendency or maybe eliminate the temptation to do that. It’s time and fear that leads people to do those types of things.
Jeff: Exactly. When you do that, you are very at risk. If you go look at the history of folks preaching those workflows and go look at the sites that they originally posted about those workflows, what’s their turnout? We called that the correlation hangover and it’s what it is. It’s a correlation hangover. It doesn’t work. It’s very susceptible to competitive risk. That’s something you do not want for your business.
If it’s a website that you’re building, and if it were to crash tomorrow, you’d like to throw it away and go build another one. Cool. Maybe that’s okay. If you’re working for a business or something that you are focused on long-term growth, why put yourself in the danger zone if you don’t have to? I see it time and time again.
I get folks walking in the door crying about it, honestly in literal tears because they did that. It worked for a while, and they were the hero. Now they’re the zero. For one reason or another, if then you’re a victim to a competitive threat or otherwise, you don’t know what to do next because you’re out of ideas, because you weren’t thinking critically to start. You were just copying and when that happens, that’s when the bad stuff goes down.
From a pitfall perspective, that’s something like, use this for good. Don’t use it for bad. Don’t pepper in words that aren’t naturally there. Don’t completely cost out the creative part of your brain. I don’t even know if it’s on what side. But it’s possible that it’s the creative part of your brain because I told you so. You got to use this as an augmentation. This is your accelerator. This isn’t your replacement for thinking because it doesn’t work. If a social media platform told you over and over again to post that midnight on Tuesdays and it doesn’t work, you’re just like, whoa, wait for a second. Don’t just rely on that data. You got to think critically. Is this the right thing that I should be doing for my website?
Ben: There are limits to how much guidance aggregated data can provide for you. We like to have those things as starting points but you’ve got to think for yourself.
Jeff: We absolutely do. We have a generation platform called MarketMuse First Draft. We called it that specifically because we want to make sure people didn’t take it, and post it. This is literally your first draft. I love it when people will take their content brief and look at that raw material and then turn it into something beautiful and say, hey, that would have taken me 25 hours before I did it in five, and it’s crushing. This was my scaffolding. It’s not my replacement. That’s the dream scenario for me and that raises the bar. You’re never going to put out a 400-word puff piece if you’ve got 3000 words and inspirational source material to then construct from. It changes the game.
Ben: Absolutely. Take one step back just briefly on to kind of the skyscraper technique-type of stuff. We hear it at CoSchedule. We preached that gospel for a time, a lot of companies did around 2015, 2016, 2017. The core principle that underpins it is not wrong, but the execution, 90% of the time is awful because it misses the point.
If you haven’t heard that term before, the skyscraper technique was you look at the top 10 results for a keyword, make a list of all the information that’s on those pages, and then go higher. Maybe you either write it in-house, or you hire somebody out, and you just say, hey, spin all this stuff into a new piece. It becomes a race to the bottom pretty quickly when everybody is doing that against each other. What got missed was there wasn’t a strong enough push, or maybe the voice that was crying out for us wasn’t loud enough or got drowned out. It doesn’t work without original insight. It doesn’t work without a fresh angle. It doesn’t work without going deeper. People stopped at just knowing what had already been created. They didn’t think about what was missing.
Jeff: You’re right. It is still a great reference resource process. The two wings of this are—getting too deep into it—first of all, if it’s multiple meanings and you do this, you put yourself at a dramatic risk. If it’s multiple intents, you could potentially create content that doesn’t make any sense that is unoriginal. That’s what we were seeing happen a lot. The other piece is, I like to use that you just haven’t earned it yet. With some of the people that were doing this in a way that was unoriginal, I can go write the best definition in the world for a topic. But if I haven’t built any other content across the buying cycle on that topic, my definition isn’t going to rank well even if I write the best definition in the world. That’s the greatest aggregation of what is the term search results page. It doesn’t work that way. You have to have the infrastructure and the foundation.
The touts in the space that we’re going for that already had the infrastructure. They already had coverage across the buying cycle. When they publish and take this technique in hand, they already had that […]. You would have a much harder time building from scratch. You would have to be doing heavy link hustling. You’d have to be doing heavy outreach hustling. You have to be doing all that stuff which we then know that outcome has a much shorter shelf life and a quicker hangover versus if you truly were building the foundation like the people who are instructing you on that. They have the reel. You have this one moment in time correlation.
It still works well if you’ve got great infrastructure and you add value. Adding value, as you’ve mentioned, I love the term, is information gain. If you provide information gain in addition to differentiated value, that’s when the magic happens. It’s still a magical process. You just have to add to it. You got to make sure you have coverage at other stages of the buying cycle, too. You have to make sure that fresh angle and really look at the content if you’re emulating. How much is being published on that site, on what topics, at what cadence? What’s their off-page link velocity? What are the things that contributed to this page doing well? If you know that, that’s your spirit guide, frankly. Don’t just do it blindly because you can run into so many problems. I see it constantly. It hurts me on the inside.
Ben: Same. Jeff, this has been great. Before I let you go, is there anything else that you want to get off your chest on this topic or anything else that you think would be important to leave our listeners with as a parting thought?
Jeff: One thing I like to say is content efficiency—do you know it? How much content are we creating or updating and how often does it achieve its goals? If you don’t have that metric internally, go get it. The other one is to go into your content planning platform, whether it’s a spreadsheet, whether it’s at CoSchedule, whatever. Document your processes generally—research, planning and prioritization, briefing, writing, editing. Write down each one of those things in that life cycle. Don’t forget post-publish tuning, more integration. Write down all of them and make a little box above them. Am I using artificial intelligence to improve this part of the process? Go across that list and see how many checkboxes you have. That’s your maturity model.
March 16, 2021