How Can You Balance AI and Human Touch in Your Content Strategy? (With Andy Crestodina)

Scott: Hello, and welcome to the show today. You will hear the advice of Andy Crested Dena co founder and CMO at orbit media. Andy has been at the forefront of content innovation for some time. And today. He'll share how to optimize AI and your content strategy without losing that human element that sets your business apart from the others.

My name is Scott Murray, and this is the content brief where I bring you the key points and takeaways from conversations with today's experts and humanized and innovative content. Every time you hear this sound, that's a key takeaway from today's episode. Let's get started.

Thanks for joining me today. Andy Crestedina is the Chief Marketing Officer and Co Founder at Orbit Media Studios, an award winning digital marketing and web design agency. He's also been recognized as a top influencer and marketing expert by SEMRush, Forbes, Entrepreneur, and more. Andy is also the author of the popular book, Content Chemistry, the illustrated handbook for content marketing, which has just completed its latest update.

If you know Andy's work, you've undoubtedly trusted his well researched, And insightful content surrounding blogs, SEO, analytics, content strategy, and website optimization. But like a lot of digital marketing experts, Andy has turned his attention to AI as well. However, one of the specific things that I love about Andy's insights on AI and content marketing is there's always that balance between What AI is good for and what it's not good for

Andy: probably the two biggest mistakes with AI are one Dismissing it too quickly or trusting it too much.

The right perspective of course would be that this is a tool we can use it It can help some of what it gives us is good like any tool The more of what it gives us is not good like any tool, but I trust myself. I trust you I trust the humans. I know to be experts in these topics more than I just do I trust an AI model,

Scott: you know, as with anything in content and content advice today, there's a lot of overlap and there can be so much overlap.

You can still find yourself. Looking for deeper questions and maybe even answers on how you should just do something, well, you know, you can read all the philosophical stuff all day long, but at the end of the day, you might still be looking for, okay, well, what, what can I use this for now? Without ignoring it completely and how can I use it in a way that's going to benefit me now without overusing it?

So amongst all that generalized conversation about AI is good versus AI is bad Marketers may just find themselves wanting to know How do I use it effectively? And Andy has worked really hard to find

Andy: those answers. I try to compartmentalize in my mind that when I have these conversations about AI, there's really two conversations.

One is, how can this help? Can it help me? What prompts might be useful? Uh, is there a, is there an applicable use case? Which parts of my job could be done a little better or a little faster, more comprehensively? That's the practical side. The other side is accuracy, bias, plagiarism, copyright, labor market impact.

So I find it unhelpful to mix those much. I prefer the conversation and I get more value from the conversation with my peers about Which of these use cases is the most appropriate for an AI? Which prompts might get me this result? What uploads from which tools I have? What data could I give the AI to get, to get a better, uh, to get a better result or do a quick analysis?

These are very useful. They can make a big difference in, in my job. But, uh, I'm also not really using it for its classic use case, which is writing. I'm not a huge fan of that because I think that many of the things that are digital content best practices, AI really is not very useful for.

Scott: Yeah. AI should not be replacing your writers.

Instead, you should evolve and improve the skills of your writers to write less like the internet, which is where AI pulls its writing style from. And write more like people, humans can write for humans better than AI can write for humans. Also, Andy has conducted a really great blog survey over the last 10 years.

And in it, marketers actually share what's working for them right now. And let me just say, there are components to a modern blog that AI can help you with, like outlines and research. And there are plenty of things humans can do. Still need to contribute and control, like

Andy: language. With or without the tool, there's some language that just doesn't connect.

It feels watered down. It's undifferentiated. Uh, you can train, or try to train, ChatGPT to not do those things. If you have a ChatGPT Plus account, click on your, your face in the top right and then go to Customize GPT. There's two boxes in there. Uh, one of them says, Custom instructions, what do you want ChatGPT to know to give you better responses?

The other is, how would you like it to respond? I put these words in. These are my avoid words. Unlock, unleash, synergy, uncover, furthermore, garner the days, in light of, in the sea of, digital landscape, utilize, delve, showcase, leveraging, in the midst of, game changer, treasure trove. So yeah, you can train it on that and if you have it right drafts, you know, maybe it'll use fewer AI typical words.

But I think even beyond that, I think one of the things that is important to remember is a great quote from Eugene Schwartz, copywriting legend from the sixties and seventies, uh, he said, great copy isn't written. It's assembled. If you think about the elements that go into a strong piece, they are compelling visuals, AI is not very good for that contributor quotes, AI doesn't have any friends to reach out to strong opinion statements, AI has no opinions or points of view or experience.

compelling It has, uh, lots of formatting. Short paragraphs, bullet lists, bolding, internal links, subheaders. You know, AI doesn't do any of those things very well. I've tried, I've given it like a hundred of my articles and say, I'm trying to write something in a first draft like me. So, I think that the best use right now for AI is to write outlines, and uncover provocative topics, uh, do topic research, find gaps in existing things.

You know, your favorite writers are still writing every word. I think, I think even as it gets much, much better and today's the worst it will ever be, it's going to be better every day. I think that me and you and Mark, and there's most of us are, we'll have no trouble. Separating ourselves from AI, when you read it, you'll be like, it's got me that I can

Scott: tell by the way the mark in that example was Mark Schaefer, who has also spent a lot of time studying AI, while still maintaining that human element in content.

And I think I've talked about this before. He recently was talking about some of the challenges that could arise if. There's A. I. content out there that's disguised to look like a human wrote it, but in actuality, A. I. wrote it, and how that can make your audience feel duped. And the example was, like Milli Vanilli, they're from the 80s.

If you don't know about them, there's this really good documentary on them that just recently came out. They were a band that exploded in fame only to get busted for not being the real singers of their songs. They were lip syncing. And when that happened, obviously there was a lot of outrage from fans.

Understandably so. So the question was, how would an audience feel if a brand tried to fool their audience by implying something was from a person, but it was basically AI pretending to be them? I mean, if you were in their shoes, would you feel duped? Would that frustrate you? I mean, just the fact that they thought you would fall for it?

Look, man, I ain't falling for no banana in my damn pants. You know, I mean, what does that say about the company's attitudes about you? My latest LinkedIn newsletter just talked about how attitudes about audience can impact the success of content. If you took the time to read the content and they implied that we wrote this for you, uh, when in actuality it looks more like we got AI to write this because we didn't want to take the time to fully develop this thing and load it with value for you.

Last year, Mark actually shared this on LinkedIn. In the past week, I was asked to review and promote two different books written entirely by ChatGPT. I declined. Doesn't a book need a soul? And this is kind of what Andy's referring to when he talks about the critical importance of human writers and the relationship they have with an audience.

And by the way, this connection isn't just limited to marketing content or marketing books.

Andy: Uh, let's say AI becomes infinitely better and it is just unbelievable at creating anything you need in any format. I like reading crime fiction. I like detective stories. I like, uh, historic fiction. I like things that are set, that take place in Chicago.

Uh, I like plots with lots of twists, but yet they resolve on a note that's kind of like, um, very satisfying. Now, let's say I write a prompt to say, write me a novel that has all those characteristics.

I've got a novel in front of me now. Let's print it as a paperback. Let's say, you know, the technology is such that I can just do this thing just shows out the next day. Now I've got two books, right? One written by a human, one written by AI. Do I want to read a book where I would literally have no relationship or perspective on the creator?

Or do people who consume content in the background get more value from it when they feel like it's coming from a point of view? This is unknown, uh, but I suspect that in the future, people will still have a preference for that sort of indirect relationship with the creator of the content, the tick tocker person, you know, the YouTuber, the, the author, the teacher, the writer, the questions partly, I think comes down to, to what extent do we feel connected to content because It comes from someone, lots of things that we don't care, right?

The commodity stuff. You know, I need a recipe for a cheese sauce for my nachos. Who cares who wrote that? For that detective story, I'd, I'd rather read the one from a person for that YouTube video for that article, but probably, you know, people will have a preference for consuming content that is just created on demand.

I think the implication though is, is not good for the creators to quickly make more of that stuff because really the, the person who needs that content. We'll just have it created on demand. That basically takes the content marketer out of the game. If this is so fast for you to create an article that checks the boxes mostly in 10 seconds, it's also that fast for your prospective reader to just have that article generated on demand for them.

So the opportunity goes away for the marketer and the creator of undifferentiated content because the person who wants that can get it on demand in 2 seconds. Right after you realize, wow, they can write a pretty good article for me with almost no effort from me. It's the same effort level for the person who needed that article, they're going to stop looking for it.

They're going to start prompting for it, and they're just going to get it when they want it.

Scott: I really love this point by Andy because it adds another layer to some of the bigger problems with AI. Where it's about speed of production, taking the human out of it, things like that. And one of the biggest early warnings experts gave us about AI when it comes to content creation was, you know, as you all are just jumping on this tool, this new technology and trying to get it to do everything from writing and creating content, they said, don't forget your competition has these tools as well.

And if they tell it to do all or most of the writing or content creating, and we know AI likes to use a lot of generic and meaningless words and phrases, then as all that stuff gets produced by you and your competition, the differentiation, the humanized elements, the connection elements are all pretty much lost.

I mean, AI and say, Yay, it can write stuff for us. And then there's just that much more generic content out there that looks like everything else. They just got it out faster. But now, Andy says, if all you're going to do is simply say, write an article about this, or write a blog about this, or write this, and then just post it somewhere, another reason why that's not valuable is because your audience could do that as well.

I mean, think about it. If your audience, whether it's a B2B or B2C audience, They wanted to learn something they could tell chat GPT, Hey, give me a little write up on this product or this challenge or this trend or something like that. And it could give them something. Why would they need that from a company?

And if it's not much different from what they can have ChatGPT do on their own, how does that develop a connection between you and a brand? How does that inspire somebody to trust you? And how does that inspire somebody to want to buy from you? What that blog needs is your stamp, your voice, and then all those other things that Andy talked about, regarding the collaboration and the additional resources and visuals that make a blog worth buying.

Well, not only looking at, but maybe reading. I mean, this is one of the things I really worry about when it comes to AI or really any technology. Brands getting so consumed in what it can do for them, that the people they're supposed to be helping or talking to suddenly become secondary. If they're lucky.

I mean, you've probably already seen this type of behavior at its worst already. Maybe you've seen blogs that were clearly written by AI or some other content. That's why you see more platforms creating warnings, letting people know ahead of time. Yes, this was AI generated or created. I mean, there are those that are so excited about A.

I., they're just looking for something to use so they can say that it's powered by A. I. or A. I. powered, or they may even say they're an A. I. company when at the core of their business, there's something completely different. And then there are those that are just looking for how many places they can make A.

I. do something. And then they announce it like it's practical when it really isn't. For example, there are social media platforms that say, Hey, want to comment on this post? Click this button and, uh, let AI do it for

Andy: you. I'm disgusted that people are using AI to do the most human and social things that humans can do, which is, you know, interact with each other on topics we care about.

If you want AI to write a comment for you, that's point of view. You have a point of view. That's a human thing. AI does not have a point of view. It has no perspective or opinion at all. So how can you outsource your opinions to a machine? Like, it's, I think it's terrible that LinkedIn puts that button there.

I will never click that button. Anytime anyone ever sees a comment from me or a message from me, it's written by me, just for the record. Let's say 10 percent of comments in the future are generated by bots and they're garbage. Those of us that are really talking to each other, the difference may become more pronounced, which is a benefit to humans.

Maybe you're not improving your relationships. You're not networking. You're not, uh, differentiating yourself. I'm sure hilariously today, probably there are, uh, AI generated articles getting AI generated comments and no one's getting any value. I think it's a waste of energy, but no, it makes this conversation maybe stand out even more.

Scott: Okay, so we've established some ways that an over reliance on AI can work against you and even dehumanize you. So what are some of the ways it can help you improve or create content that does connect with the humans in your audience? Well, you can start by telling something like ChatGPT about your audience and your industry.

Andy: First, train it on your persona, or use a detailed persona prompt. After it has your persona, after it has the details, after it knows the hopes and dreams and the fears and concerns and the emotional triggers and the decision criteria for selecting a company like yours, it has all that. Now, everything that you ask for is, will give you a more, Relevant and more targeted response.

So don't expect AI to really give you good results until you have trained it on who you're talking to. That's marketing 101, right? You wouldn't hire a freelance writer and have them just start creating content without having shown them who we target and talked about their information needs. It's very useful to just talk to it about what topics are missing.

What might you find provocative? What are the counter narratives in my industry? What are people talking about in my industry? And what are some things that are likely to trigger a response or a reaction or a point of view? What do people in my industry think is false, but is actually true? It's very hard for humans to identify these things quickly.

AI will give you a list of them. A third of them will be garbage. And the, and a few of those might be gold. There are lots of use cases that really sidestep the concerns about. Quality, uh, you know, efficiency, accuracy, differentiation. If you go to Google analytics and export, uh, from the pages and screens report, the setting title tag as the first column, and then upload that, uh, click share, and then download a CSV and then upload that and ask the AI to do.

Semantic distance analysis on all of your content, inferring topics from the title tags, which it does beautifully in seconds. And then say, uh, which articles have I not written, but almost written, or where are there gaps in my elbow? They say, write your lifetime body of work, my LBOW, and it'll tell you what you didn't publish yet.

So gap analysis, it'll give you six topics and four of them will be boring. And one of them was incorrect. And one of them was great. So go write that article. That entire exercise takes five minutes, probably. Right.

Scott: So apart from sharing insights online, Andy also teaches classes every year. One at Northwest University, which he is teaching now.

AI is a required part of that class. But so is learning the use cases and limitations of the tool.

Andy: I want all the students to learn how to use AI well. To show how the limitations of AI, as an exercise, I ask each of the students to speak a sentence that AI could never write. Sounds strange at first, but actually it's quite easy. I'm going to just say some things now that AI can't do. Literally would never put into a prompt.

If you only like red licorice, not black licorice, you don't actually like licorice at all. Red licorice isn't, isn't licorice flavored. I think that if you put cream in your coffee, you're not drinking coffee anymore. You're trying to make a candy bar. That's gross.

Scott: Hey, wait a minute. I put a lot of stuff in my coffee.

Cream, Splenda, syrup, cinnamon.

Andy: I'll have a half double decaffeinated, half capped, with a twist of lemon.

Scott: I still think it's coffee. I order coffees like that. They say it's coffee. When I told Andy this, he says, that sounds more like a Starbucks dessert than a coffee. Wait, what were we talking about? Oh yeah, content and AI, sorry.

And what he teaches his students.

Andy: Most of the things we say, like, you know, GA4 is better than Universal Analytics. Any time that you express a point of view, or an opinion, or tell a story, You are automatically different from AI. AI can't do those things. It can't help you find some of those topics. You know, give me a list of counter narrative opinions in my industry.

What false things do people think are true? Or, you know, what's likely to trigger a strong emotion or response. It'll give you ideas. Some of those you don't care about. Some of you will. So useful for thought leadership topic research, but taking a stand is something only humans can do. Chatty PT cannot throw a punch.

AI does not care about anything. It's read the internet and it can summarize beautifully, but it doesn't care. So. Uh, just take a stand even a little bit, even on a mundane topic, and you'll automatically be different from AI.

Scott: Finally, now that we know what AI is best for, and what's not the best use cases for it, how do we map all that out?

I asked Andy, what does that pie look like? What percentage of content should be mostly AI, or less AI and pure thought leadership, straight from that human, that author that people are connected to? Thank you. Is it 70 30? 50 50? 92 8?

Andy: Yeah, 220 221, whatever it takes.

Scott: How do we figure that

Andy: out? I love this and I actually was just recently thinking about this.

So I'm going to share some new thoughts and things that I think are some ideas that aren't fully formed yet. First, I would say that commodity content is a really special and uncommon use case. There are moments when you do need content at scale. You've got a thousand page website, none of the images have alt tags.

You want to be accessible, you need to write a thousand alt tags. Humans should not do that. It's not necessary. The AI can do that and there's tools that will do that for you. You've got 500 pages on a blog, all of which are missing meta descriptions. AI can write meta descriptions for 500 blog articles.

It's not really worthwhile for a human to use, uh, you know, their, their life to take that time. Uh, another weird example, I podcasts and each episode has their own, has its own image on this popular sleep podcast. Those images are AI generated. Which is fine because no one cares about that much about the image for a podcast.

I, I hope that Podcaster isn't spending a thousand dollars per episode having a painter make those pictures. Maybe YouTube descriptions. There are some examples of commodity content. Setting that aside, content mix, I'm going to just throw some numbers out there. Let's say a successful blog needs maybe a total of 200 URLs.

You can keep them updated. You can keep recycling them. You can keep building up more and more content, detailed in depth content on those URLs. I don't think most blogs need the thousand URLs. We need a hundred great URLs. So let's take a, let's assume first that the number is finite. Secondly, what should the mix be of those?

I'm just going to give an example. 60 percent how to practical stuff, uh, 20 percent original research, 20 percent strong opinion thought leadership. Yeah, the practical stuff has a bit of thought leadership, and it has a bit of original research in it, referring to those other, other, other articles. Now those are topics.

There's also, you could say, what are the formats? What's the mix of formats? Articles, videos, webinars, podcasts. But generally, it's, it's an interesting content strategy question. What percentage of our content should be things like strong opinion, thought leadership? But I don't know how much, if any, of our real articles should be ever considered a commodity.

That is hopefully the, you know, what we're running far away from.

Scott: So while I hope we've done a good job in helping you kind of visualize the line between AI and humans in the world of content marketing, I hope this was also a reminder of how important it is to not let a great technology take over our work or our lives.

We've been making movies warning us about that danger for years. Watch war games from the eighties where a machine called the whopper. Replaced humans to help the military launch nuclear missiles, when it thought we were under attack. And it almost launched those missiles when it mixed up a simulation with reality.

Shall we play a game? Love to. How about global thermonuclear war? It also didn't care about human casualties, it just wanted to complete the objective and win the war. If you haven't seen it, I highly recommend it. Despite things like large floppy disks, dial up modems, phone booths, and smoking on airplanes, it still holds up.

It's important to remember that while this is a new technology that is a hot topic right now, and it can change a lot of things for content marketers, the one thing that has not changed is the demand for brands to speak to people like people, to form connections, build trust, and show humanity.

Andy: Content should be so specific from your voice that if you covered up the logo of the brand, you could tell what brand it came from.

Also, this highlights The value of doing what we're doing now and anyone who's out there, you know, growing their, uh, speaking from their own point of view, the personal brand becomes more valuable too, because trust becomes more difficult. We will continue to seek out those that we already trust. So personal brands will become more valuable.

I think that for the, the swing toward commodification of content and low, low costs, as everything swings toward this, like, Oh, we can make everything faster and easier and commodity medium quality stuff is easy, you know, There's much more of it. The market also will likely swing back the other way where we prioritize human connections, uh, personal brands, people we trust, strong opinion, thought leadership, the differentiated content.

So ironically, this mega trend is great for humans, great for people that want to differentiate themselves, uh, and, and be vulnerable, be a person, tell stories. Uh, and plant a flag on the topic.

Scott: All right. Let's revisit some key takeaways from today's episode. AI should be used strategically in content creation for practical tasks. Like creating outlines, conducting topic research, identifying content gaps, and optimizing workflow efficiency, human writers are essential for injecting personality, authenticity, and strong opinions into content.

If AI can focus on the practical applications, Human content creators can focus on high level creative tasks that require critical thinking and personal insight. As AI technology becomes more accessible, both competitors and audiences can use it to generate content, making it crucial for brands to differentiate themselves with humanized, unique, and engaging content that stands out.

I will have links to Andy Crestodina on the show notes for this episode, and that will include a link to his company, Orbitmedia. com, and his LinkedIn profile. If I can help you improve your content to better stand out or communicate with others, you can email me at scott at scottmurrayonline. com, or reach out to me on my website, which of course is scottmurrayonline.

com. I'd like to thank Andy Crestodina for being a guest on today's show, and thank you for joining us on the Content Brief.