AI & Conversion: Stop Using AI To Find Efficiencies - Instead Find Deficiencies With Andy Crestodina

Andy Crestodina is the Co-founder and Chief Marketing Officer at Orbit Media Studios, an award-winning digital agency. With 23 years of experience in analytics, SEO, content strategy, and website optimization, he is regarded as a leading expert in the marketing industry. Andy has written over 500 articles on content strategy, SEO, AI, social media, and analytics and is the author of Content Chemistry.

Here’s a glimpse of what you’ll learn:

[2:29] How to utilize AI for CRO
[8:15] Training AI models to generate customer personas
[15:20] How AI can identify deficiencies in PDPs
[16:49] Andy Crestodina shares examples of various AI prompts for website redesigns
[25:37] AI-driven Google Analytics 4 analysis reports
[33:19] Why Andy collects screenshots of Google search results
[37:31] The future of AI in internet searches
[45:12] A deep dive into Andy’s book Content Chemistry
[48:29] Best practices for leveraging GA4, Google Search Console, or ChatGPT to increase search rankings
[52:56] Andy reveals the results of his AI survey and AI’s potential beyond e-commerce
[1:00:37] How to generate revenue without purchasing ads
[1:09:01] Andy’s transition from teacher to marketer

In this episode…

Marketers and e-commerce business owners leverage AI mainly for content creation purposes — something customers can accomplish easily themselves. How can you differentiate your brand with innovative, AI-driven content?

Rather than developing a prompt to write articles or PDPs, marketer and AI prompt engineer Andy Crestodina recommends having AI analyze a webpage, PDP, or other body of work to identify missing components. This may involve creating prompts detailing whether a PDP meets industry best practices, identifying how to enhance a web page to drive conversions, or determining whether the content meets audience expectations. To take it a step further, you can train the AI to generate ideal customer personas that can influence CTAs, landing pages, and other consumer-facing content.

Tune in to this episode of the Up Arrow Podcast as William Harris chats with Andy Crestodina, the Co-founder and Chief Marketing Officer at Orbit Media Studios, about revolutionizing AI for CRO. Andy talks about increasing search rankings with AI-driven analytics platforms, how AI will alter internet searches, and AI’s potential beyond e-commerce content creation.

Resources mentioned in this episode

Quotable Moments

"[Using AI for content creation] is absolutely the opposite of the things that we should be doing."
"A fantastic use of AI is to look at a body of work and tell you what's missing."
"AI is a skills leveler. If you're Nancy, I can make you a little bit smarter."
"The biggest marketing question in the future will be how to train AI chatbots to recommend your content."
"The best time to plant a tree is 20 years ago. The second best time is today."

Action Steps

Use AI for gap analysis: Start leveraging AI to identify deficiencies in your content rather than just creating efficiencies. This approach allows you to understand what your content is missing, making it more unique and valuable, as AI can analyze vast amounts of data to pinpoint gaps that human intuition might overlook.
Implement persona-based AI training: Create detailed personas and train your AI tools to understand and respond from the perspective of these personas. By doing so, your content and marketing strategies will be more aligned with the specific emotional triggers and informational needs of your target audience, enhancing engagement and conversion rates.
Optimize product pages with AI: Use AI to conduct audits of your product detail pages, assessing them against a set of best practices. This method can quickly reveal areas for improvement, such as missing trust indicators or unclear shipping information, helping to improve the user experience and increase conversions.
Enhance content with behavioral psychology: Integrate cognitive biases and behavioral psychology principles into your marketing copy using AI to suggest improvements. This strategy can make your content more persuasive and emotionally impactful, differentiating it from generic AI-generated material.
Analyze and adjust based on data trends: Utilize AI to analyze Google Analytics and other data sources to identify trends and patterns, such as peak purchase times or low-conversion landing pages. This insight allows you to adjust marketing strategies dynamically, optimizing budget allocation and improving response times during peak purchasing periods.

Sponsor for this episode

This episode is brought to you by Elumynt. Elumynt is a performance-driven e-commerce marketing agency focused on finding the best opportunities for you to grow and scale your business.

Our paid search, social, and programmatic services have proven to increase traffic and ROAS, allowing you to make more money efficiently.

To learn more, visit www.elumynt.com.

Episode Transcript

Intro 0:03

Welcome to the Up Arrow Podcast with William Harris, featuring top business leaders, sharing strategies and resources to get to the next level. Now let's get started with the show. Hey everyone.

William Harris 0:16

I'm William Harris. I'm the founder and CEO of Elumynt and the host of the Up Arrow Podcast, where I feature the best minds in e-commerce to help you scale from 10 million to million to 100 million and beyond. As you up arrow your business and your personal life, I am excited about the guest joining me today, Andy Crestodina. Andy is the Co-founder and CMO of Orbit Media, an award winning 55 person digital agency in Chicago. The business has grown through a 100% content marketing approach Orbit has generated 1000s of leads over 24 years without any advertising. We're going to be talking today about AI for conversion. Why you should stop using AI to find efficiencies and instead use it to find deficiencies. Andy, excited to have you here today.

Andy Crestodina 0:57

I'm glad to be here, and I love this topic. William, this will be fun.

William Harris 1:01

Yeah, it's one of my favorites. I've chatted with a lot of really fun AI people over the last couple of months here. So there's a lot that I think we're going to get into. I do want to give a quick shout out to the people who put us in touch. Nancy Harut was a recent person on the podcast here, and she's the one who's like, if you haven't had Andy On, you need to have Andy. And I'm like, she's like, do you know? And I was like, I do know. Andy, I've actually been a contributor to your some of your content marketing surveys over the years. And I think the first person who put us in touch would have been maybe, like, 10 years ago. I think was Jason Quay, so thank you, Jason for putting us in touch all those years ago.

Andy Crestodina 1:34

Wow. Yeah, long time friend of Jason's and Nancy. I just love her brain. I've got, you can't quite see it. Her book is right behind me.

William Harris 1:45

I have her book right here as well, so I haven't taken it off of my desk yet.

Andy Crestodina 1:53

We both have her book within arm's reach. That says

William Harris 1:55

a lot. That's that says a lot for that book. We there's another book that we're gonna talk about here in a little bit too. So I'll bring that up before I do I want to announce our sponsor. This episode is brought to you by Elumynt. Elumynt is an award winning advertising agency optimizing e-commerce campaigns around profit. In fact, we've helped 13 of our customers get acquired, with the largest one selling for nearly 800,000,001 that ipoed. You can learn more on our website@Elumynt.com which is spelled E, l, u, m, y, N t.com, that said all of the good stuff, like I said, we're talking about AI for conversion. Why we need to stop using it to find efficiencies and start using it to find deficiencies. And so I'm going to start there with a question, which is, what is what is wrong with the way that we're currently using AI for CRO

Andy Crestodina 2:39

marketers are using AI in ways that their audience could also use AI, sure, which is a problem, because if you're you, if you're just, you know, I'm going to write something. I need a topic. I need a draft. Write me an article about how to improve conversion rates for e-commerce, product detail pages, your your potential client, your audience, your prospect, could just write that same prompt. So it's absolutely the opposite of the things that we should be doing. A lot of marketers are doing the exact thing you shouldn't do, which is race to the middle and write really kind of shallow, undifferentiated stuff. If AI can write it in 10 seconds with you offering just one simple prompt, your audience can get that same article in 10 seconds with virtually the same prompt. That's the last thing you should write.

William Harris 3:28

That's really good. So how do we fix it? Like, what are the ways that we should be using AI as marketers?

Andy Crestodina 3:36

Well, you sort of teased it at the front. I would say that a fantastic use of AI is to look at a body of work, maybe it's one URL, or maybe it's the whole internet, and tell you what's missing. That's actually something that's very, very difficult for the human brain to do. Like, if you and I just pull up a product detail page and we say, like, what's this missing? What What could what would make this a higher converting page? What key information isn't on this product detail page. Now you maybe we've done decades of e-commerce, and you and I could kind of pull it off, or do do good enough, but most people like that's really hard to do, right? It's kind of hard to say what's not there, AI. It's very easy for AI. So you can give it a product detail page and a quite detailed prompt we can share and say, like, to what extent does this product detail page meet or not meet these best practices? To what extent does it meet or not meet the expectations of my persona or my audience? You've got to give it your persona. You know, does this address all the possible objections of an e-commerce shopper, it'll tell you, and then also, in a content strategy way, if you ask it like, what are the most important topics to this persona again, give it your persona first that are not covered on the big blogs. Yeah, that is magical, right? Because it'll tell you what's not out there. So E. Instantly. So you could have it find deficiencies at a micro level, just a single page, or at a macro level, this whole corner of the internet that is a way to make the thing that you're making better, instead of just making more kind of medium quality stuff faster.

William Harris 5:16

So it's similar to a gap analysis. I use Ahrefs and have the gap analysis like, how would you compare and contrast what Ahrefs is doing for keyword gap analysis versus, you know, from what you just talked about,

Andy Crestodina 5:31

actually, a lot of what you can do with AI is sort of replacement for those tools, sure, unfortunately, for those developers, and I love Ahrefs, so lots of the prompts that I use are sort of ways to get output from the software. It's like the ultimate software tool, like it's just with the super detailed prompt. They joke. English is the hot new programming language, because you're like writing it feels like you're writing code, almost like if we share, if we get to it. You know that prompt for auditing product, video, pages, it's like a piece of software. It runs like a piece of software, and it breaks you a you know, the output as a report, even a chart, for example. But the other difference is significant, which is, if you train the AI on who your target audience is. So here's an example. Every year give us a presentation at a conference of college bookstores, independent college bookstores, just fun, cool, weird, niche. And they're, they're very independent people, right? They're kind of rebels like they want to. They're, they're fighting against the man. They don't want to be Amazon or Barnes and Noble. They want to be like the you know, they're speaking of the tribe, and they're part of the tribe. They're the people who go to that college. So if you talk to it. Talk to the A about your persona. What is this college sophomore? Love, hate, fear, hope for. Where do they insecure about? What do they what do they want most out of their day, or out of this backpack, or whatever? The thing is, it'll talk to you from the perspective of your audience. You can train it on your persona. Ahrefs isn't really doing that. It doesn't know you're not first. I mean, this is a these are all two step processes. But step one is in AI is to give it lots of information about the information needs, the objections that must be addressed, the emotional triggers of your target, and then you're talking to the thing that you made. It's like building a synthetic member of your of your target audience, sure, and then doing gap analysis way better, I think, because it'll tell you, Oh, this person loves the sports team they care about. You know, the, you know how many days to the weekend you know that they they love the, you know, the wolf mascot, whatever so if you, after you teach it that, and you have it write a new draft product detail for, you know, a product description, it'll be hilarious, it'll be fun, it'll connect. It'll be emotionally connected. But other tools don't do that as well, because they don't know the audience, and they can't really emulate or simulate being a member of your audience.

William Harris 7:59

So let's, let's talk practically. You mentioned that we could almost look at how we do this. How are you training the AI to be this synthetic member or this synthetic audience person? Well, I have

Andy Crestodina 8:15

very detailed battle tested ideal client profiles, and I keep them in a folder on my desktop, and when I'm talking to the AI about something that I'm working on, by the way, this could be a public facing article, landing page, call to action, whatever, or it could be an internal document like this might be my, you know, a contact calendar, you know, or a positioning statement, or whatever, like Mark, you know, strategic assets. But I had that folder and I grabbed that ICP. I actually have two ICPs, because we offer two things. We do, web development and also website optimization. So the different information needs. Some people just want SEO, some people want a new website. They're kind of like different concerns you have to address. But I've found that the best results are when I just make one big doc and I just put it all in there. It doesn't matter that much. I've had clients share with me their personas, and it's like a PowerPoint with 12 pages, three pages for each of their four personas. That's fine. Just give it the whole thing that's working for me. If you don't have that, then I have or we could share. Sounds kind of boring. I could read it, but a persona prompt little cheesy. It's not the it's not super rigorous. But you know, in this like little paragraph, you give it, it will output a persona that will be mostly correct. You have to fix it. Don't trust it. Tell it what it missed. But at very least, you know, if I'm preparing for a meeting and I want to impress a prospect, I will take a good 20 minutes to prepare for that and do my usual old school, you know, check out the rankings. Find low hanging fruit key phrases, you know, make a list of likely, you know, edits or hypotheses for improving traffic. But now I take an extra minute and I. Making a bunch of assumptions. You know, I generate a quick persona. I start giving it pages. I talked to it about the, you know, the differentiation of this brand. I give it samples of the copywriting, and then I have IT audit or make recommendations to improve conversion rates. So now, in less than half an hour prior to a meeting, I can go to a meeting with a pretty good idea of how I could improve both the cheese and mouse trap, the traffic and conversions for that for that client. And then everyone got value, you know, they they might hire us, they might not, but, but they can leave the meeting with six ways that they could get better results, like, you know, by Monday.

William Harris 10:37

Yeah, that's good. So you know when you're training, let's say that you you're going to talk now to this, this synthetic member. Are you basically telling ChatGPT or something like that? It's like, Hey, your your bill, your Alexa, your whatever. Here's the persona. Like, I want you to respond to me as if you are this person now, but this is who your identity is, and then you start asking it questions. Or what does that look like?

Andy Crestodina 11:04

Well, to be efficient, I have like a like a shared prompt library, which I'll probably publish here eventually, and one of the one of the folders in it, or one of the documents in it is like the persona prompts. So I give it the persona prompt, and then it generates a persona, and then I'll have to go and look closely and correct it, or tell it to D prior. Tell it to de prioritize something or add something. The subsequent prompts are also coming from the shared prompt library. I'm just copying and pasting stuff and it they just reference the persona, you know, based on the information needs of this persona. Take the following article and write six draft headlines and four draft social posts or whatever. That's not actually my favorite use case. I don't know why I thought of that, but the idea is that these prompts all say based on the perspective of the persona. My personas and my ICPs are named Alex and Alexa, or Alex and Alexis, and so when I'm doing it for myself, I'll just say, you know, what does Alex think of this? But no, you could 100% say, Okay, with this is a sales training now. You are the persona. Talk to me from the personas point of view. Interview me as if you're considering my services. I'm going to respond back and afterwards, give me a score card to show me the extent to which, you know, I triggered, I leveraged cognitive biases, or, you know, addressed your top, top, you know, underlying concerns, so you could actually, you know, in a training tool context, just tell it to take on that role. But my prompts are all sort of pre written, so I just reference the persona instead.

William Harris 12:41

So I do the same for writing copy a lot of times for ads where I'm basically saying it's like, okay, referencing this persona now that I've, you know, supplied you with whether they whether Chat GPT generated, or I did, or client did, or whatever. But okay, you have this persona. I want you to write me, you know, 15 different headlines using cognitive behavioral therapy, which I've trained it on, you know, stuff that I've learned from, you know, Nancy, or using, yeah, mind states, or using whatever it's like, different approaches. And so I've got got some really fun names for all the different copy, ad copy bots that I've that I've created and written, but exactly to your point where it's like, Okay, now, referencing what you know now about this, give me this, and the output is significantly better than, like you said, if you just came to it and just say, like, hey, write, write 10 headlines. Well, it's not going to understand it well

Andy Crestodina 13:34

enough. Yeah, it's and knowing people like Nancy or I once sat in a session of hers and just took a bunch of notes, I My hands hurt after I leave Nancy sessions. I ended up just turning all of those things that she mentioned, all of those cognitive triggers, and putting them in a prompt and say, look for opportunities to make this ad landing page, product detail page, home page, offer whatever the thing you're making social posts. Look for opportunities to have that asset incorporate some of these behavioral psychology phenomena. And it comes it's pretty funny so to take so it's like the research shows this. AI is a skills leveler. If you're Nancy, AI can make you a little bit smarter, but you already know your stuff. If you're Andy, right, and you have not spent a decade researching cognitive biases and marketing, then AI is a, you know, a big lift. It makes me much smarter, makes it can make me closer to Nancy faster. So in that close, I'm still we will never be right? That's fine, yeah, but it will give us some ideas, as if Kenneth she was sitting on our

William Harris 14:54

shoulder. So you also talked about finding what's missing on like a product line. Page. And we're kind of getting back to that almost a little bit too. What does that prompt look like? And we can share a screen too. I know that there was like, almost like, a tool or a prompt or something like that that you had, and maybe it wasn't for this one. It was another one that you had. But you know, how are you, how are you finding deficiencies then, and how are you asking it to find what's missing?

Andy Crestodina 15:20

Yeah, so the classic prompt would have it would tell give it role in skills. I kind of keep that simple. You know, you are an e-commerce conversion expert, skilled and maximizing the percentage of visitors who add a product to cart. It's like those two things like, that's really prompt engineering, 101, but then after that, I'm going to say, you know, the best product detail pages is shared the following common traits, and then it's like a list. It's like a, like a mini blog post of best practices. Uh, another shout out Justin Rondeau, uh, he and I trade notes and call each other like neighbors bar and a cup of sugar. Hey, do you have a prompt for that? Yeah, sure. Here, try this one. So Justin Rondeau, who's like, a master of digital, and he headed, he was head of growth, I think, for digitalmarketer.com he's done 1000s of AB tests in his career. He's like, Yeah, yeah, no, I got a lot about product detail pages. Here's my 180 page PowerPoint. Wow. Thank you. Justin perfect. So I sat down one morning and went through the whole thing and converted by hand, manually, very time intensive, his 180 page PowerPoint into a prompt, and I am happy to read it to you. You're

William Harris 16:38

limit. There's a limit, though, too, isn't there? Like you can't, you can't copy the whole thing over using the API, okay, yep, no, no, it's not. I mean,

Andy Crestodina 16:49

I could have given it the whole thing and had it write the prompt. I probably could have, but I wanted to learn myself, so I decided, there we go. Okay, sharing screen. So here's my shared prompt library, not fancy at all. These are, you know, prompts for all different use cases. Here's the persona prompt. You're a job title, company, size, industry, geography, with certain roles, skills, responsibility you need help with, challenge, problem, task, you're considering a product or service, right what Elumynt does, or what Orbit does, list their hopes and dreams, their fears and concerns, their emotional triggers and their decision criteria for hiring or even just contacting a potential company like mine. So after I'd much rather have the detailed ICP that you've kind of vetted, because I've had 1000s of conversations with prospects ChatGPT is not so this is a cheese ball. Way to do it. And no matter what it gives me, I'm going to iterate and tell it to fix it, take that out, put this in. You'll see right away, you don't trust anything AI gives you. Never, right? Always look at it critically. Okay? So it has that. So then I'm back in the prompt library. Then I'm just popping open my e-commerce prompt folder, document. You are a this one might be, I don't think this one references the persona, but, but here you get the whole thing. You are an e-commerce expert, skilled in maximizing sales on e-commerce websites. You evaluate product copy, UX, trust, conversion optimization of e-commerce websites. Here's a 15 point checklist of best practices for high converting, highly engaging e-commerce, product detail pages, the copy, it's descriptive, differentiated benefit, focused use of personal pronouns, guarantees, money back. Can it be returned? Is it in stock? Shipping, explained, images. It's high quality. Multiple pictures, zoom. Feature, the product in use, shows the content, shows the the product size. You know, that's a common failure of e-commerce websites. You can't tell how big something is. It's got a product. Video, badges, trust, sales, payments, new product, sale, low stock, sustainability, varieties, options, reviews, ratings, star, star. Reviews, calls to action. So the way to use this is, I'll just go to some random what was I looking at the other day? A led Frisbee. Here we go. Actually, this product is a problem. I couldn't even see it. Where is this? The something's wrong with that page? LED jump rope, $300 it's got a video. Okay? It's next step. I'm going to do a full page screenshot of this product. Detail page. I am not interested. Yep, that might not be working based on some crazy, weird JavaScript that this page has on it. I am not trusting that product. Let's try another one. I'm going to try just once more if this doesn't work, but you'll get the idea I'm. Um, so what I want this to do? I wanted, I want a a full page screenshot, because I wanted to evaluate the this page is so tall they can't capture it. We should pick a different product. I was looking at a commode. We have some people celebrating their 10 year anniversaries, and so I was going to come looking for different gifts to buy for them here. How about a giant cement ceramic grill? Love it. Okay, we're gonna, we're gonna, we're gonna cook some stuff. William, you're ready $2,000 grill.

William Harris 20:36

I'm always ready to try again.

Andy Crestodina 20:39

Okay, so I'm taking a full page screenshot using a browser plugin called go full page. There's another one called Awesome Screenshot. They are the same. A few issues with this screenshot. I see some collapsed elements, some accordion content, the FAQ. Stuff is all closed down. I probably should take a expand these first filters, ratings, lots of reviews. Looks like there's a carousel. That's gonna be a bit of a problem. We'll just keep it as is. That's fine, close enough. So now I'm gonna go to ChatGBT. I'm gonna copy and paste this e-commerce prompt. You're an e-commerce expert, skilled at blah, blah, blah, blah, blah. I'm giving you a detailed I'm just going to take the first part of the prompt, paste it in, and then this bottom part, I'm going to skip right to the second part, rate, the extent to which this page does or does not meet the the best practices. Paste that in, grab our detail. Grab the full page screenshot. It uploads up. Upload, okay, ready?

William Harris 21:45

This is brilliant. I I'm absolutely mesmerized by how you've done this. I

Andy Crestodina 21:51

This is, it's fast, right? It's very fast, yeah. I and it makes, there's some risk that we become a bit lazy, because I didn't really even look at the product detail page that closely. But in so. Ai not to make things, but to audit things. Ai not to be efficient, but to find deficiencies. This is a chart showing the extent to which that page did or did not meet the best practices in my prompt look. It's a color coded chart. It loves the high quality images. It says that we missed the opportunities to add any kind of trust seals or badges. Not too bad. It gave us a three out of five on money back, guaranteed. Can it be returned? Is it in stock? Probably could do better at that. Sure so. And then down below here, it's starting to we get the details of the audit right, like it's got areas for improvement. It could be more direct and engaging using personal pronouns. It has limited visibility of trust indicators. There's no certifications or badges or of any kind shipping while shipping. And so info is present, clear details on cost, timelines, international options would enhance transparency. That is often true. So this is obviously a, you know, a well made page from a mature e-commerce program. But hey, that was a point of view, and the cost to us was basically zero, and the time we invested in it was under two minutes.

William Harris 23:22

Yeah, it's gonna make a wild difference, right? Like, if you're just getting started in I think about AI a lot this way, where it's like, no matter where you're at, like you said, it's a skills leveler. It can't make us a Nancy, it can't make us an Orbit Media, but it can get us to the point where we can maybe optimize our page well enough that we can afford to go with an Orbit Media Right? Like there are, there are things that you could do to say, okay, great. Get to that point where you're actually generating enough income for your business, where you could say, Okay, now I'm ready to take to that next level where it's not me running it through ChatGPT, but I'm actually working with professionals who could take this a step further.

Andy Crestodina 24:00

Yeah, and I'm a content marketer, and I want to be useful internet citizen and help people, whoever, whoever's out there. And, you know, only a sub, a tiny subset of all possible web products are a fit for us. I mean, my most experienced clients have have gone through maybe, maybe five or six website redesigns in their career, and that's a lot, but people on my team have done hundreds. I've been part of the planning process for like, 500 or more website projects you don't need, though. Have done 500 I mean, to to identify the missed opportunities on a URL, any URL, it's just an amazing teaching tool. I love it. I think it's great for this. I'm excited by it. I'm happy to share that. Like I said, I'm luckily going to publish my entire prompt library. Why not? It really doesn't cost me anything. It makes other people smarter. It can help us in our marketing by having a popular post that day or being get top of mind and word of mouth and those other things that are, by the way, ever more important. The AI era, if you want to differentiate yourself, now more than ever, you have to do something bigger and more interesting and more helpful than your competitors. So publishing every prompt, all my best prompts, iterated, you know, in their most recent form, is good for the reader and probably good for us too. Yeah,

William Harris 25:19

I like that. What other what other ways can we use AI to enhance our e-commerce websites you'd mentioned? Thank you. Pages to me before. You know, maybe we want to go down that path. But what are some other areas we can use it to find if efficiencies or deficiencies?

Andy Crestodina 25:37

Yeah, there's, I like to create reports in GA4, and upload those to AI and have it do some analysis based on that. So one of them is, if you go to, you know, you don't even have to use the conversion or key event metric in GA4, if you just go to, thank you. Page visits, go to the Pages, report, filter, search or filter so that every column is just a thank you. Know, it's just the thank you page visits. And then add as a secondary dimension, got to click a tiny blue plus date, and it shows you the date and hour when every one of those thank you page visits happened. Export that, upload it to AI with a prompt that basically says, like, do it a now, you know, draw me a heat map matrix showing the time of day and day of week when people bought my products. And it will draw a chart showing the time of day and day of week. And if you see that people on Mondays tend not to buy. Crank down your ad budget on that day. Maybe, you know, like where everyone's you know, you know, contacting us on Thursdays. Ramp up your sales response times on Thursdays. You know, like this. I've never seen that chart before, and I never had any value in analytics before. By doing date, you know, by adding a secondary dimension, which just puts another column on the report for date is now suddenly useful, because you can export GA4 reports and import them into AI, and just like, you know, do for me a, you know, day of week analysis or month of year analysis, you know, it can and then whatever It gives you, you can just ask it for ideas. Okay, based on that data, what, how could I adjust my program? What should I be looking for? So that's an opportunity, I think, for anybody is to just go farther with their GA4 data by exporting it and importing it into ChatGPT,

William Harris 27:38

big fan of that. We use Power BI at Elumynt, and it's so robust and powerful, and I'm waiting for them to really get ChatGPT fully enabled in there. They started doing a little bit of stuff there, but it'll get a lot better there. But I still find myself going to my super metrics stuff, and using super metrics, getting all my data from j4 and Facebook ads and Google ads and Shopify, bringing it all in real quick and then sending all that data real quickly over into ChatGPT, because it can connect to my Google Sheets and asking it a whole bunch of really interesting queries about the data. It can do a bunch of merges very easily and to your point date is literally one of my favorite things, and at least for me, because I feel like I can visually see the trends better. I like tables and tables that, you know, we highlight, if there are, you know, things that are changing where it's like, this is going up, this is going down, highlight this green, highlight this red, those are all fine, and it's helpful. And I like that because I like numbers. But there's just something about seeing, you know, a line chart over the last 14 days, you're like, oh, that's when this took place. Or if you're looking at this sometimes, week by week, you might see that maybe your cost per visit is twice what it used to be. But you find out when you look at a line chart, it was just because of one day. There was one day I just had this astronomically high thing. And so you can see that when you break it down by date, and you miss it if you're breaking it down, you know, some other ways you're looking at it more in aggregate.

Andy Crestodina 29:06

Yeah, that's, I think, the main, the main reason for the line chart is to show you the spike, the blackout, you know, any weird anomaly in the data, because, you know, you could be comparing to, I've done this before. Line charts kind of visually noisy, and it's annoying to me, and it's annoying to me. And sometimes I'll customize how important GA4 to remove it. But then I'm looking, I'm comparing two URLs, and one of them is like, way out performing the other. But then I realized, like, Oh, I think the second one wasn't even live in the first half of this date range. So you got a line chart to see. Like, Oh, okay, well, there's zeros for the first two weeks in this date range, but William, part of what you just said is that like You're like you're the Nancy in this situation, because you're talking about Super metrics and Power BI and all those things, I am not. I've never I was so confused that I downloaded Tableau. I don't do any of those. I'm basically, I mean, I've done two. Years of analytics, but I'm kind of a hacker with it still, and I'm use it for very specific kind of use cases. So for me, and people like me, the same way that you and I can be a bit more like Nancy and leveraging psychology, you know, I could be a bit more like you who has, you know, these power tools, and I can just give, it's like, Okay, what about, you know, mobile versus desktop for this type of conversion. And I'm gonna make a quick exploration. I'm gonna export that traffic source and landing page by mobile versus desktop. Interesting one. Yeah, that's actually one of my favorites. It's a, you know, when you combine those three, but give that to ChatGPT, and it'll, it'll give you ideas. That one, sometimes it jumps out at you even before you do that. I do this for friends. I'm not a PP. I've never done any paid as you mentioned at the top, but I love to perform, do analysis on landing page performance. And if you make a GA4 exploration that shows landing pages in in the on the left column and sessions, engagement, conversion rates as the values of the right columns, but and filter the whole thing only to show paid you know, medium exactly matches CPC, and then add another column in the column. Thing can add a third column. It's like three dimensions in one report. It's 3d literally three dimensions in analytics for device category. And in seconds, people's mouths drop because they realize I'm buying a lot of mobile visitors to a certain landing page with a 0% conversion rate you'd look at landing page performance for paid by device type, by device category, and and I, I see that's like, you know, Andy, I'll call you right back. Like they, they turn off their they leave the meeting, and then they go turn off ads from that, you know, from mobile to that page, and save like 50 grand in like two minutes, all you have to do is to find that answer. So that's a really I did a piece about that we could share, just like the step by step, how to create a GA4 exploration that shows landing page performance for paid sessions, engagement and key events per device category, and it's very likely that everyone listening who has a paid program is buying mobile traffic to a page with 0% conversions. Sure,

William Harris 32:30

any and all of these things that you're mentioning to these tools, these resources, blog posts, et cetera, we'll try to link in the show notes as well. We try to be very good about that, so anybody who's listening is like, oh, I want that. I'm trying to Google it. I can't find what I'm looking for, whatever. Let me know if something's missing, and we'll try and get it in there, but try to make sure we have that in there. The $2,000

Andy Crestodina 32:48

grill. We'll share that

William Harris 32:51

$300 jump rope. We'll link

Andy Crestodina 32:55

to that in the show notes. Anybody that needs a $300 programmable led jump rope, you can't take a screenshot of it, but your kids will love it,

William Harris 33:02

yeah. What about you? You mentioned to me when we were talking before that you you have an interesting hobby of collecting Google screenshots of Google screenshots of Google search results. Why? What? How is this helping anything? Oh, man,

Andy Crestodina 33:19

it's very it's important, because when I'm invited to testify in front of Congress about Google's monopolistic power, I'm going to pull those out and show you. Know, there's a name for this. Nancy would know. It's called a creeping normality or landscape. Amnesia. It's basically when things change very slowly over time. People don't really notice them. It's like the boiling frog, you know? So I Yeah, we did SEO back when there were 10 blue links and search results were 10 blue links, you know? And now it's like a TV, slash magazine, slash Yellow Pages, right? It's like, not anything like what it was, and you would never really notice. I mean, sometimes day to day, like the ads got a little bigger, or the video carousel changed. So by taking full page screenshots of search results pages for 10 plus years, I have this post. I do, I update it once a year, where you can see, side by side the changes over time in Google search results, and just how ranking number one it, it's it's actually sad that SEOs are so still so focused on rankings, because ranking is not position. Ranking number one is sometimes like 800 pixels from the top of the viewport, right ranking number one is sort of like ranking them where number eight was before, because there's maps and people also ask and featured snippets and a knowledge panel and AI overviews. So it's. It's obvious looking when you see it all side by side in the series of screenshots, why click through rates from search have gone down. And Rand Fishkin talks about the no click reality. And Ellie Patel from verge talks about Google zero. There may be a day when almost no traffic. Now, my perspective on that is that really, there still is a click through rate for certain key phrases, the bottom of funnel key phrases, the people who are looking for a service. You know, who need help. You know, behavioral psychology consultant, marketing consultant. I don't know Nancy yet, but I'm trying to find someone like Nancy because I need help here. I'm going to click. I want to read about it. I want to look at options. I want to investigate. I want to read about pages. I want to see people's pictures. I want to read their bios. I want to seek testimonials. Or, you know, is this person legit? So I believe that there will still be click through clicks from for bottom of funnel key phrases, but the little the hobby or that, that little, you know, that long term project of collecting screenshots of search results pages is really interesting, and it's one of those things that just keeps getting more interesting

William Harris 36:15

over time. Yeah, I think, to your point, one of the things we were talking about is that there's a significant change in the way that people are finding information, and let's say, not just finding information, but finding products as well. I'm curious a little bit more about your thoughts about what this looks like in the future, and part of this because if I remember, you had a you had a blog post, something along these lines, if I remember correctly. But so I was, let me back up a bit. I was talking with one of the other guys on here, Britain lad, and he was mentioning this idea of in the future, it will be absolutely archaic, in his mind, for you to actually go to a website because to buy a product, at least for instance, because you would just simply say, like, hey, I need a new pair of shoes. And you ask your AI agent, which has been trained on all the things that you like and the price ranges you like in the stores that you like, and it just says, Great, here's the best deal for this shoe, which I think you'll like. You want me to go ahead and buy it? And you're like, Yeah, done. The AI agent will be doing the shopping for you, more or less, right? What do you think about where AI is, I don't know, kind of stealing or moving things away from some of these searches?

Andy Crestodina 37:31

Yeah, there's quite a few predictions about that. The famous early one from Gartner is that global search volume will drop by 25% that prediction you just said it was like dropped by 100% because nobody will go to, you know, you don't have to use the internet the same way as a user. I think that AI is a fantastic experience and a great way to use the internet, because it's super clean, it's fast, it's it's more lightweight, there's visual noise, it's more private. It's, it's just a lovely, I mean, as it's basically it ingested the internet, and it summarizes the internet, and it gives you the answer you wanted for buying products. There are. There was a great post on CXL once years ago who said that they said there are two kinds of product detail pages. There's the spec driven buyer who needs details about the, you know, the performance of the thing, or the the data behind it. Then there's the visual product buyer, style esthetics, like, how would it look on me? And there's really two kinds of product detail pages we should be making depending on what what our information needs are of our visitor. Is it highly visual or highly data driven, the visual thing. AI might do that well, but whatever your amazing product pictures are of, you know, the girl wearing the Halloween costume, your website will still be the way that you put that picture on the internet, right? I think that the value of websites may change and that they become more important as feeds into other systems, like AI, kind of like Google products, right? Like your website is not just a user's experience, and we all added schema to all of our pages for this reason Exactly. It's because we know that our websites are being crawled and that the data is scraped and that the data needs to be reformatted and other users experiences, so the value of sites will change. The AI bot will be one of your very important visitors. The way that Google bot is an important visitor for your site. Now you'll be doing lots of micro formatting and schema and tagging. You'll be thinking about, is this thing optimized? Is my content going to be easy to be reused in other places, but you'll still need to have a home for your you know, a digital property, which is how you have the best control of your brand and all of your information. So I think that people will continue to build out super detailed websites, knowing that the site itself is not necessarily the user's experience, but. Yeah, but that also, I don't know. I feel like there's stories and influencers and brand will still matter. I really don't imagine a day when people basically stop using brow, stop browsing completely, that that prediction you said is like someone saying humans will no longer browse. Sure, I just don't, I don't know. I mean, am radio is still around. It's an absurd things don't really go away completely. I'm not saying that websites would be like am radio, but it's a the value of them is changing, but people still do want to click land, read, feel, hear what that place is all about, you know. And, you know, actually have a connection in that and a digital experience.

William Harris 40:48

Well, in the experience, I think, is the key word, you know, taking that a step further, I'd say that it's just the experience just changes. It becomes more immersive. I can't remember the name of the one that Walmart launched earlier this year. I didn't really like it, but I appreciated what they were doing, where it was, like, this very different, you know, VR shopping experience, where it's like, you're, you're kind of like, there in this world. And I like, I like that. I do appreciate where that's going. Some of the stuff that they're doing right now on, I've got the apple vision Pro, and some of the stuff they're doing within that kind of context, I think is very immersive. And so I can imagine where, you know, the AI glasses become better, and what was it meta came out with Orion, or whatever, things like that will become actual, practical and real, and those experiences will still be there. We want to experience the products. We want to experience what it's like to shop. And like, to your point, like I even like to experience shopping. I still go to the store sometimes because I just genuinely enjoy the experience of being at the store. Those experiences will just change and modify a little bit, and you have to at least make sure that what you're doing is is set up well for the AI bots to be able to find, access, crawl, understand, want to be a part of what your website is doing as well.

Andy Crestodina 42:06

Yeah. And then as marketers, we have to figure out how to train the agents to recommend us and recommend our brand and our content. And that is the big unknown, William it is. I had a call just before this talking to this guy? No, he's doing research on this. And almost the only way to try to understand, like, why AI recommends a certain thing, or what AI's sources are, is to ask the AI itself, which feels a bit meta. I'm not sure that how reliable that is. Like, why did you recommend that brand? What were your knowledge sources? Why did you choose those knowledge sources? So the biggest marketing question of the future will be how to train AI chat bots to recommend your content, your insights, your brand, your people. And it just seems so unstable, like, does the user have memory turned on? You know, in ChatGPT, sure. Do they have memory turned on? What words did they use? You know, is this, and it's weird to me, because it looks like it's ripe for spam, depending on how this goes, because some of the things that it does will be like, you know, what are the best x in this category? And it just finds one list post and just regurgitates a single list post. So is the game just going to be like, do PR, you know, listicles as guest posts. PR style, like, listicles like, it's kind of keyword cramming, and I'm a little worried. I don't know what let's get.

William Harris 43:37

I think there will be an element of spam that happens the same way that when Google first launched the ability to spam, it was very high, and then it gets tighter. And I think that you're completely right, like when this does get more broadly adopted, the Spam is going to be high. There are going to be people who figure out ways to hack the system, and then it will get smarter and it will learn over time. But, yeah, we're gonna go through a period of what the heck is this doing?

Andy Crestodina 44:07

I'm really disappointed, because if you ask it like, you know, you know Chicago, you know who are the best plumbers in Chicago, it's mostly just grabbing from like, one or two directory sites, or, you know, a giant list post that is not it. I mean, it's a great way to get answers, but to get recommendations for a product, like a trusted product, I don't think there's no it doesn't have there was a lot of rigor behind it's not like it Google does 55 things, right, like they, you know your different keyword sets and categories, and these are your money or your life pages. And you know these things, you know these keywords. It's looking for trust signals. And you know, I can tell this person here has, you know, just information intent. So go ahead and give it simple answers. It's, it's just all it's all fun, I guess. And you know, we it's. Because I love working in a very dynamic industry and in a category where things are unclear, I'm fine. I'm a content marketer, so I'm clearly fine with ambiguities and indirect benefits. So yeah, it's all TBD.

William Harris 45:12

Yeah. Speaking of content marketing, that is a good time for me to bring up you. You actually the book that you sent me here, that I've read. I have the autographed version, by the way, I just need to brag about that real quick here. But it's content chemistry, the illustrated handbook for content marketing. And I've read a good chunk of this. I've had it for maybe a little bit of time here. And I think one of the things that I appreciated the most about your book, unlike a lot of other books, is, I wouldn't consider myself necessarily a visual person. I am, I am an auditory person. But this book is so visual, and so I highlighted one of the pages that I really like here that I want to show it off where it's like, if you are a visual person and you're trying to learn content marketing, I mean, this goes through all the stuff, but like, you have actual, like screenshots and visuals and call outs. This is it reads almost more like, let's say, I hesitate to say textbook, because it has a negative connotation, but it reads like a very good textbook of showing you how and why and all of the different things that go in with this. And like I said, this is way beyond content marketing. When I first got it, I thought it was gonna be a cop, but it's you go into all kinds of levels of marketing and how they bridge back into content marketing, which I really appreciated. Wow.

Andy Crestodina 46:32

Thank you, William. It's a you're holding version seven, so we've worked hard at it and rewritten it many, many times. That one is the fresh one you've got. It has the AI prompts and has the GA4 reports. But the the original kind of hypothesis was, let's take our our best, most detailed, most helpful articles, and repurpose them in this format where they all make sense, like goes in order, maybe a good time to point out the inherent weirdness of blogs, which is the reverse chronological pile of stuff, like, there's not, they're not very organized. Blogs are sort of, you know, come out of this format where things were date driven, and it doesn't really make sense, you know, just like these, it's just a bunch of stuff. The book, it's got a table of contents. Is an outline. It's a guide. It takes you through these steps. You know, you can open it to, you know, you can jump to a certain section, and, you know, everything in there is kind of prioritized and flows, but the visuals came the legacy of this book is that much of that content was published first on the web, and the web is an insanely competitive blood sport. That's the thunder dome of content competition. It's very you have to be very visual, sure. So those got carried over into the into the book, and it make it's it's a fun read. The downsides, doesn't work as an audio book. Can't do it and it doesn't and even an e book doesn't really translate. Yeah,

William Harris 47:56

yeah. One of the things you talked about in there, and it's also on your website. Here too is that I really appreciated the seven steps to rank higher using GA for Google Search Console and ChatGPT. We chatted about this a little bit by using some of, you know, GA4 and stuff like that. But without going through the entire spiel of this, give me a little snippet of you know, what are some of the best ways to use j for Google Search Console, ChatGPT, to find better rankings?

Andy Crestodina 48:29

Oh, there's two fantastic use cases. And again, I sort of feel bad, because I have friends who have seo software companies, and this is ChatGPT sort of doing what an SEO software piece can do. One of them is to site wide. It's like a site wide thing. It's almost like a scale SEO at scale, if you go, if you export from Google Search Console, the a report that has all of your pages, their average position and their click through rates, you can, if you give that to ChatGPT, it will tell you which of your pages is lower than expected click through rates based on its average position. That page should rank should have a higher click through rate based on its average position. But what it doesn't have is the title tags of each of those URLs. The title tags are all in GA4, so it's a it's a two step process. If you export from GA4 the pages report with the primary dimension set to title, that's every title tag of everything on your site, secondary dimension of page path that's every URL of everything on your website. Upload that to ChatGPT with the Google Search Console report, and it will tell you it'll look at the title tags for all of your pages that have lower than expected click through rates based on their search position, and then recommend higher click through title tags. What like the first time I saw that I'm like, every one of these suggestions is good. These are all good ideas. Like. Yeah, like, I missed that opportunity to trigger some, you know, an emotion, or that title tag is, like, just a little bit too keywordy. There's no real hook, you know. So just as it, just as a an idea, ChatGPT is a way to do it. You could do it manually. Larry Kim did this years ago without any AI at all. But find your the URLs in your website that have a lower than expected click through rate based upon their average position as in their rankings, and then go look at those title tags and ask, could I make that title tag more compelling, more interesting, more unexpected? That's just a solid exercise. Again, William, I'm just mostly using chat GP to do things I have always done. Yeah, the other method is to just do it for a specific page. You just export from Search Console all the queries for a single URL with the average position for each and then you upload all that to ChatGPT with the article and say, recommend on page. SEO copy edits to 10 sentences on this page to better incorporate the phrases for which it ranks but does not yet rank high, so you give it all, so it has all of the queries and how high you rank, and it has the content, the page itself, full page screenshot or copy paste the text. It's going to come back and just do some basic copy edits to better incorporate those phrases that maybe weren't even on the page at all. Super reliable SEO win, very likely to work well so that that's a you know better title takes for higher click through rates or better on page SEO body copy edits for higher rankings. And really, all you're doing is giving it the data and a detailed prompt, and it will come back and make the recommendations for you. One bonus trick there, if you, if you add to the prompt, highlight the changes. When it recommends them, it'll highlight the changes. It'll bold them for you. So it's easier to scan through and just say that one's dumb, that one's great, that one's that one's good, and then make decisions about which of those edits that you agree with and would like to change, and then pop up in WordPress update save.

William Harris 52:11

This is literally shaving up to be probably one of the most practically helpful episodes I think I've ever recorded. This is a lot of fun, great. I want to stick with AI for one more question on this then. And I've had the chance of interviewing some other just fantastic AI minds. If you haven't checked out the episode with Noah Kravitz, that's actually Nvidia's AI podcast host, and so he's interviewing, you know, Jim fan and all of the best AI minds like these absolutely brilliant AI creators, if you would. But you recently interviewed 1000 people about AI doctors, lawyers, etc, like, are we ready for this? What did you find out from the results of that

Andy Crestodina 52:56

survey? That was a it was a collaboration with question Pro. They have, they build panels in minutes, like they can, they can get you huge amounts of answers very quickly, because they have, you know that their tool is partly like pulling together people from the general population. So collaborating them is fun. It's like, what, what Mark, what, what data can we get from the general population? That would be interesting, because you're not talking to marketers, you're not talking to brands, you're talking to brands, you're talking to just people. So the question was, is about AI readiness? It's like, Do you want an AI to fly your airplane? You know, certain people want that. Some don't. That's a funny one, because AI because airplanes are largely flown by autopilot. You know, do you want an AI account. Do you want there were four levels of answers for each question accounting? I want a 100% human accountant that uses no AI at all. I want a human that checks their work with AI. I want an AI that humans check, or I want a pure AI to do my accounting. So it's a way. So all these questions and all these categories were a way to find out people's general readiness to accept AI and their services, and to what extent the outcome for marketers is really like, at what point should we start putting AI into our messaging? We build websites, AI is going to write the first draft of all your copy. Wait, wait, what I'm going to work with humans, you know, maybe there's not AI readiness in my category. Sure, that's just an example. You know, so, but, but there's a point depending on the service, you know, where you know you're, you know, it's like Wealthfront. Like, some people don't want a human financial planner. They want an app. They trust the app. So it's a it's fascinating, like, whatever you feel now, it's likely in five years, you'll feel differently about an AI doctor, an AI accountant, an AI attorney, like all of those things.

William Harris 54:51

Yeah, I want to run with this for a little bit, because I think it's fun. It's something I've thought a lot about as well the future, let's say AI and robotics and. General, because I'm gonna, I'm gonna lump them together for a couple of reasons, because I think a eyes can be incorporated into the robotics. And already is more and more that AI Doctor, let's say the AI surgeon, etc. Let's say, even say, I don't know. 50 years from now, I could be wrong a little bit on the numbers, but if I remember correctly. The article that I read suggests that the human hand has 22 degrees of freedom, and the Tesla robot has 19 degrees of freedom. I could be off by a little bit, but it's very, very close to, you know, the dexterity of the human hand. Then I'd say, 50 years from now, it will probably surpass the dexterity of the human hand. I think it would be very unreal, for unrealistic for not to you'll be able to train that doctor to perform. Let's just say, you know, it could have run through simulations of 10,000 open heart cases, more than any open heart surgeon could have ever possibly performed in their lifetime with higher dexterity. I think it's almost inevitable. I think the thing that will hold us back in all of these areas, the same things can be true for a plumber as well, right? Like the plum it can it could have seen so much more about plumbing than any master plumber ever could an electrician, etc. The thing that'll hold us back likely will be regulation, because we have a lot of regulation in all of those areas anyways, and it's very hard to maybe shift blame. If your human heart surgeon makes a mistake and it was an understandable mistake, then it's just simply that it's an understandable mistake. If it was an egregious error, they were drunk while they were doing it, then, you know, we very easily, we can, we cast the blame on them, if it's the robot, I think it gets a little bit trickier.

Andy Crestodina 56:48

So interesting. I read a piece once, like the limit of the there's like a practical limit to the Height of Buildings, super tall buildings. One of the limits is the human ear. Like elevators just can't go that fast. Like people get uncomfortable. It's not, it's not a good experience to go in a insanely. They can make them faster, but people don't like that. So I think human readiness, yeah, we only want certain levels of technology in our lives. And there's already been massive disruptions in the last. You know, you could, I could count five huge technical disruptions in the last 25 years. Big ones. Very big So, yeah, what do people today? If you saw a car without a human in it, driving down the street, you might be a bit alarmed, right? Like, that looks kind of scary. Like, wait, what's going on? There's no person in that car. 10 years from now, you might see a car with a human driver and think, Whoa. I hope that person's paying attention, not distracted by their phone. Like, that's a little bit scary. That's a human driving that car. Totally so how it's a sociological, psychological change that has to happen before we're ready for more of these technologies in our lives. And we should also just be open to asking. I'm reading currently the Yuval Harari book called Nexus. Wow. Really good. You highly recommend it. You've all heard, I mean, started that one that's on my list. So good, wow. I mean, he breaks it down. He's a he's alarmed. He's basically sure he believes. There's a lot of people who believe in kind of a techno utopia, and his the goal of his book is to give the CounterPoint. He's purposely being provocative and a bit of a doomsayer. And he acknowledges that, but yeah, we should have the conversation. Do we want you know these things in our lives? Is it a better accountant because they make you know their their accuracy is higher? Or do you want your radiologist to a human who's seen a 2000 heart scans, or a robot that's seen 2 billion heart scans? You know which of these, which of these? You know, healthcare providers? I almost said doctors. I don't know if they're both doctors. Which of these service providers for healthcare are going to give you a better outcome, and then, then what's left, right? Let's say we accept this, and we and many of these services are augmented by or taken over by algorithms and bots. What's left? My friend Joe Martin, I doubt if you'd met him, he said, you know, the last job will be to just make people feel special, the thing that humans can do right the bedside manner. That I don't know if we're going to like the bot for that or never, say never. Maybe we will. I love this conversation, William, it's all TBD, but I think for many services, people will happily accept a cheaper, faster, less error prone, higher quality service provider. Have you seen the robot sushi waiter?

William Harris 59:38

No,

Andy Crestodina 59:38

it's adorable, like this rolling cart that comes over and it's got like, little animated blinking eyes. Oh, nice. It's fun.

William Harris 59:46

Yeah, we want that. I wanted to give a shout out as well to Orbit Media a little bit specifically, because I think there's just some fun things I want to learn from you, and just leadership and what. You've been able to accomplish that transcends, let's say, even if somebody is an e-commerce owner, they can learn from your leadership and what you've done, you guys have generated, you know, like, over 8 million in revenue now. But you said you've never bought an ad. And I think that's interesting, because that's what we do, is advertising. But I would say for those who are looking at this saying, Okay, we're scaling up on ads. We're doing everything we can there, but we're maybe missing something else. What's the something else they're missing that clearly you have found a significant win from that they should start considering? Well, that's there's a

Andy Crestodina 1:00:37

human side tying back to this entire conversation to content into marketing, where there's a story or a point of view that can make a brand and a brand's message very memorable, that really doesn't have much to do with advertising. You can come out for something and against something. You can it's like, you know, when Nike signed Kaepernick, Colin Kaepernick as a, I don't know anything about sports, but, you know, remember this, like, big story in the NFL, you know, he kind of, like, has quiet, sort of peaceful protest. He takes a knee during the national anthem as a message against police brutality. Was that an ad? I don't know. Was it influencer marketing? Like, when that? I mean, the sales went up. Their market cap went up. So I think that there are human things that happen in the micro scale, like what we do. It's like just having a point of view, having conversations, networking, going to events, giving a little presentation, hosting an event, you know, collaborating with other people, like we're doing now. You know, none of these things have anything to do with advertising, but the impact for the listener, for the attendee or the reader, is sometimes high enough that they think of that thing weeks and months later. They talk about it for several days. That doesn't ever really happen with ads, right? That's just not very common. People don't really share ads or subscribe to get more ads. You know, search for ads, sure, but content marketing is but, yeah, there are a few. There are a few, but content marketing. Those are the content marketing channels, right? Search and social. Search and social and email so that. So if the goal, if, if, if it benefits, if you're a very transactional brand, you don't really, you know, you make led jump ropes, and people search for an LED jump rope and they they click and they land and they buy. But other things, the more consultative things where trust is higher, I'm gonna spend $150,000 on a website, you know, oh, that company explained once in that interview, like, how to find, like, which landing pages of 0% conversion. I remember I ran that GA4 report, and that guy helped me right, that that was a, that was a cool trick that, you know, that I talked about, or someone told me about her. You know, people keep telling me, I should, you know, reach out to that guy. That's something that is a bit special about content and when possible, like true thought leadership and influencer collaboration and publishing original research and just being out there. I mean, it's, it's not, there's no out of pocket spend, but it takes a huge amount of effort over the long run. But that's we're one of many examples, actually, yeah, 8 billion in revenue, 55 employees. And I write an article every two weeks, I give presentations, I record videos sometimes, and my goal is just to maintain high activity levels and be the most helpful, useful, you know, best resource that I can possibly be to anyone that wants to pay attention for four minutes. You know, here's the thing I made. Maybe it helps. And then, you know, long run, it really does work. It's amazing. It's, it's a giant act of faith. But it, it's, another way to go to the market. Yeah.

William Harris 1:04:02

I mean, to your point, that's kind of what the Lego Movie was. I refer to it as an ad as well, but the Lego was, yeah, it was content marketing for Lego. And I saw people that it was also an ad, right? It was an ad for going to buy Legos in the idea that, but it was an ad that you were willing to pay to sit through and watch. So it's like, people do like ads. We like seeing things that are, you know, encouraging us to buy things. You just have to make the right types of ads that people enjoy watching, not interruptive and you're excited about I thought

Andy Crestodina 1:04:35

of that movie today. I don't know. I haven't seen it in years. Yeah, like the Will Ferrell at the end, the micromanager and like the it's so clever. That was so, so good. Yeah, that's a, that's another extraordinary case study.

William Harris 1:04:51

Yeah. The other thing that I really appreciated that you told me is that you're not the CEO. You. Of the company you founded, and there's a good reason for why.

Andy Crestodina 1:05:04

Yeah, I'm bad at that. I'm no good at that job. William, sure. I'm a, I grew up working in restaurants. I'm kind of like a people pleaser. I'm a service industry kid who grew up, you know, from way back. I'm a kind of a teacher. The CEO is a hard job, and I respect you and all the founders and non founders who step into that role. I mean, that's in those famous cases all over big tech, you know, has lots of case studies and famous stories about that. But no, I think CEO is a difficult job. It's not necessarily a fun job. It's not a job I'm well suited for. This company is kind of designed so that I don't have a lot of direct reports. And I don't. I'm not. It would be. It's disruptive when I show up in a meeting. Give people things to do, you know, like, how, how does it? They all have jobs. I mean, they're Orbit is a very well run business with careful management and directors and accountability and utilization and capacity and reporting and to predictably succeed on web development projects which are pretty complex, right? There's lots of requirements. It has to it needs a lot of structure, and I'm a bit too chaotic to be a good COG in that machine.

William Harris 1:06:28

It's fair to recognize, I appreciate that, that it's like recognizing where your skill sets are and aren't and calling it out that it's like it is a job. It should have a job description, and if that's not the job description that you want to have, he said, find the person that that is what they are excited about doing, because there are people that like that. I have a great COO president of my company, or whatever. That is not my thing. It is not my calling. It is not something that I find joy in. It robs me and sucks everything that I could possibly think is good out of it, but I am glad that there are people who find joy and can do that very well.

Andy Crestodina 1:07:06

You don't find joy in that either. That's hard. It's, yeah, it's, I think that there are startup people who just assume that their business will grow and scale and they're gonna crush their goals, and they're gonna run the whole place. Like, is that really what you want to do anyway? Like, is it, I mean, you our CEO, reads every agreement. I mean, runs the finances, like these. The ultimate accountability kind of person, you know, I just, I like to, you know, if I'm on a call with someone and they want to do any of the things that we talked about, I can just start doing it. Sure. Hey, there's a problem in, you know, these things aren't set up yet. In GA4, I'm going to fix it, you know. So then, if I go do everything right away when I see it and want to take when I tackle it and get an outcome immediately, let's fix this page. Let's publish that post. Let's record that video. Let's, let's, let's, let's finish setting up GA for that doesn't scale at all. That's not a process. There's not a plan B. You can't grow that way, like, that's just like, you know, just like the low Ranger, you know, trying to get outcomes and moving fast. That's not a business, really. You know, businesses have structure. They have, you know, plans and backup plans to predictably succeed in a project every time. I mean, you have to, you need a lot of a lot of pieces in place, and it's not always good to just immediately jump in and start doing everything right away. That's not, yeah, that's not how a company works, at least that's what they tell me, yeah.

William Harris 1:08:41

Well, I want to get to know Andy Crestodina A little bit more, and I think that's a good place to kind of transition to this, because you talked about teaching is something that you are very good at, and you enjoy doing. And from what I understand, your early career took you full circle. So tell me about you wanted to be a teacher, but you know, you've come full circle from that. Tell me a little bit about

Andy Crestodina 1:09:01

that. Yeah, I was going to be, I majored in a foreign language and then graduated and realized I didn't really know it well enough, so I had to go to the country and finish learning. So I did a semester in Beijing University. Finally, you know, getting fluent in Mandarin, and my plan was to go back to Iowa, where I went to college. You and I are both long time Midwest and but there were only two colleges. This is the mid 90s. They're only there were only two high schools in Iowa that taught Mandarin. And I thought I did not want to, you know, graduate and travel and then immediately go live in Cedar Rapids, Iowa and teach Mandarin into high school. So well I already been teaching when I was in China. I was taught to ESL. I was like an English teacher. So I came back and got a job, and it was a recruiting job, but the internet was super interesting, and I want to do I wanted to combine creative things and technical things build websites. Want to be a web designer. No one's going to hire me to be a web designer. I have no skills, so it's easy. To start a company than to find a job. That's weird, but it's true, someone hiring you for for a project is taking a smaller risk, and someone hiring you for a full time job. So started a company, actually, I joined with my friend Barret, who'd already been doing it, and that was January of 2000 and then that one fell apart a little bit. Started this company in April of 2001 and then, you know, had to figure out search been an SEO just that long. Had to figure out analytics, been an analytics person just that long. And then when blogging and email and social and influencers and and video all come together, 2007 I jumped in. I'm like, I'm a content marketer now and then realized, sort of ironically, well, I'm actually a teacher after all, because I'm constantly it's like, in two weeks, I'll be at marketing profs in Boston doing a seven hour AI workshop. Wow. Yeah. I mean, I do long. I did last week, I was in San Diego for Content Marketing World. I did a four hour GA4 workshop. I like the long form ones when I when I can have the opportunity, but I teach lots of sessions. I mean, the videos in the book, like you said, textbook. It is like curriculum development. It is like lesson plans. It is about learning outcomes. I actually do teach a class at Northwestern University, a content strategy course. I also teach a class in Spain every year, which is fun. I go to Spain for two weeks a year and teach a master class there. So I love it. I mean, it's, you get to see those light bulb moments and people realize, like, oh, wow, I can totally do that. Or that really worked. Or, you know, this makes sense. You

William Harris 1:11:41

You mentioned Mandarin you and you collect Google screenshots of Google search results. I collect languages, so not very well, but here we go. Ni, Hao, Ma, ah,

Andy Crestodina 1:11:53

well ahead, how you're already gonna go past you? Ma, yeah,

William Harris 1:12:00

we're gonna stop that same fine.

Andy Crestodina 1:12:02

Yeah. Well, me Xiang down the huashan. You that's like, I didn't know you can speak Chinese? Yeah?

William Harris 1:12:08

Yeah, a couple of phrases. That's about it. My brother actually lived in Shanghai for a couple years. I wanted to learn and try, and I'll have a little bit. Did you go? Yeah, I did. I actually, we took the family there. My kids were young. Then, at the time, I want to say it was, like six three and, you know, maybe six months old or something really little, but it was a lot of fun. And lot of walking. We really enjoyed just around seeing everything

Andy Crestodina 1:12:34

that's an amazing place to visit, like vast Right? Like the unit, the world is bigger than it seems. That's

William Harris 1:12:41

the thing. You can't say that you visited China. And just like, cross it off your list. You're like, I saw such a small amount of China, and there's so much more than I want to get back and see sometime, William,

Andy Crestodina 1:12:50

that's true of everything. I mean, I know people who are like, you know, I've been to 50 countries. Well, you know, I've lived in Chicago for 25 years, and I haven't been to, you know, even 1% of the restaurants, and even those restaurants, I haven't tried even 5% of the menu items. I mean, it's, there's no way to come close to anything. So, yep, just, it's a slice of life. Just stay out there and enjoy it when you know, one meal to the next.

William Harris 1:13:19

You we were talking the other day about phrases that we live by, right, like sayings that were like we say over and over again, and the one that you told me that I really appreciated here was the best time to plant a tree is 20 years ago. What do you love about this? Why is that a phrase that you live by? Well,

Andy Crestodina 1:13:38

I think it's a Chinese the way I heard to do as a Chinese proverb, The best time to plant a tree is 20 years ago. The second best time is today. So I'm kind of an environmental alarmist, and I've planted tons of trees in my neighborhood, and I just, I don't really ask permission, I just dig a hole and plant trees using my bike and buying trees at garden shops in this ton, they're all over. And it's amazing to see these little guys start to get big, you know, and to go, like, sit in the shade of a tree you planted. Like, what a what a great day. But it actually has meaning beyond that, I think, obviously not just about trees, because whatever the skill, you know, we've talked about, you know, product detail, page optimization. We've talked about GA for reporting. We've talked about AI for audience research and for audits, all the things if any of those to any listener sounded like something new or something difficult or something interesting. Today is a great day to stretch your legs and experiment with a new channel or try a new skill, or, you know, test out some prompts. It is an amazing time to be doing what we're doing. We're recording this in sort of late 2024 what an amazing. Time to be around and doing what we're all doing. We are on the bridge. We will all look back and say that other stuff was in that or in the early days, and now we're in this new era, and it's plenty of time to become a leading expert in any of these things, you know, if that's a goal of yours. So, yeah, it was, you know, 20 years ago was a great time to start doing digital but second best time is right now. Let's do this. It's totally accessible to anyone who is interested.

William Harris 1:15:29

I think that's brilliantly worded. A lot of people don't realize just how small e-commerce actually still is. The in terms of the amount that e-commerce makes up of total retail sales in the US is only maybe around 17% I think right now, really, yeah, in China, it's like closer to 80, 85% but it's about 17% of sales, of total retail sales. And it's growing, you know, I think I'm going up into the right now, if I, if I go backwards for you, but it's growing like this, exponentially. Um, but it's still so small and so to your point, even get, let's say, getting into AI 20 years from now, like you're going to be light years ahead of anybody in AI. But even getting into e-commerce in general, you're still at the very beginning of where this is going to be 20 years from now. There are so

Andy Crestodina 1:16:19

many, so many niches. It's, it's amazing just how many areas that are still not yet mature, how many categories like, I'm an SEO right? So doing keyword research, there are blue oceans everywhere there are and with AI, it's like a giant newly discovered continent. There are all kinds of opportunities to start a little store to launch a new product. I mentioned about buying these 10 Year gifts. I have a bunch of employees celebrating their 10 year anniversary. So each one, we buy 10 gifts. That's how I came across. It's like led jump ropes and Frisbees. I literally just assumed that there was such a thing as a giant glow in the dark Frisbee. Made it up. Looked for it. Wow. This company looks like, not surprised that this is, you know, that it's right there, you know, there. And there are just entire categories being invented every day. It's a really exciting time. And if you know, you don't want to, you know, jump in with both feet, and you'd like the stability of having a day job, and don't want to, you know, all the chaos of founding a new business, no problem. But in whatever your role is, there's still some little entrepreneurial opportunities for anybody that decides to take a bit of a leadership role on their team, or, like, try these new things with these prompts, you know, or be the one who steps up and finds answers in GA forum, brings them to the meeting, you know, or figures out how to connect it all to Power BI, ask William, he'll help. It's, it's a there are, it's actually almost overwhelming. How many opportunities are in front of each of us at this moment?

William Harris 1:17:59

Yeah, no, that's beautiful. Andy, it has been absolutely amazing talking to you today, learning from you, hearing about you, about who you are, even as a person. And I appreciate that you're planting trees without permission and just getting out there if people wanted to follow you and work with you. What's the best way for them to do that?

Andy Crestodina 1:18:23

LinkedIn is my best network. There's a setting where you can change the blue button to say follow, but you don't need to click the blue Follow button. There's another, there's like, three dots, and then there's many that says, connect. If you want to chat with me, just go to connect and connect with me. Cap and just you could reference this, but I'll accept anyway, orbitmedia.com that's our that's what we do. Orbitmedia.com I write an article once every two weeks, so that's the newsletter there. Other than that, I don't know. I'm at events. I'm in Chicago. If anyone's in Chicago and wants to connect, there's like, a monthly happy hour. I'll invite you too. I like the offline. I like the face to face. That's always fun. But yeah, I think LinkedIn is a great place to start. And I appreciate in the book, which you were very generous at mentioning, is on Amazon, content, chemistry, the seventh edition. Look for v7 that's the new

William Harris 1:19:16

one. Yeah, yeah. In the newsletter, I think you've got over 250,000 subscribers now. It is a very worthwhile newsletter to anybody that's looking to find more information again. Andy, absolutely amazing talking to you. Thank you for sharing your time and your wisdom with us today.

Andy Crestodina 1:19:33

Great conversation. Thank you, William. This is fantastic. I love this. And thank you

William Harris 1:19:37

everyone for listening. Hope you have a great rest of your day.

Outro 1:19:39

Thanks for listening to the Up Arrow Podcast with William Harris. We'll see you again next time, and be sure to click Subscribe to get future episodes.

‍

AI prompts to copy and paste

Persona Prompt

Build me a persona of a [job title] at [industry/company size/geography] with [roles/skills/responsibility].

This person is looking for help with [challenge/problem/task] and is considering [product/service].

List their hopes/dreams, fears/concerns, emotional triggers and decision criteria for hiring/contacting a potential partner/vendor/provide

‍

Ecommerce Product Detail Page Audit

You are an ecommerce expert, skilled in maximizing sales on ecommerce websites. You evaluate product copy, UX, trust and conversion optimization of ecommerce websites. Here is a 15-point checklist of best practices for high converting, engaging ecommerce product detail pages

Product page copy (headline/product name, the product description, other text)

1. Descriptive: detailed copy and specifications

2. Differentiated: the product from similar products

3. Benefit-focused: copy is relatable and emotive

4. Personal pronouns

5. Guarantees: money back, can it be returned? Is it in stock?

6. Shipping explained: shipping costs/free shipping, delivery timing, International shipping

Images and visuals

7. High quality images: multiple pictures, zoom feature

8. The product in use: shows context and product size

9. Product video

10. Badges: trust seals, payments accepted, new product, sale, low stock product, local product, sustainability

11. Varieties and options

Ratings and Reviews

12. Total number of ratings with stars

13. Reviews and testimonials

Calls to action

14. “Add to cart” (NOT “Buy”)

15. Live chat

Rate the extent to which the page meets or does not meet the criteria on the checklist of ecommerce best practices on a 0-5 scale. Visualize on a colored heatmap matrix, using blue as 5 and white as 0.

[Link to product page or attach full page screenshot]

‍