Drive and Convert (Ep. 095): How to Leverage AI in UX Optimization

Listen to this episode:

About This Episode:

AI is a pretty buzzy term, and it can mean different things to different people. In this episode, Jon and Ryan engage in a lively discussion about the use of AI, with a specific focus on its application in UX and optimization. 

They start by differentiating actual AI tools from AI enabled applications. Then, they delve into the practical applications and potential benefits of ChatGPT, touching on its affordability, effectiveness in idea generation, and visual design capabilities.  

Listen to the full episode if you want to learn:

1. What you can accomplish with a free version of ChatGPT
2. How to write specific prompts for brainstorming and ideation
3. How to use ChatGPT to write codes and create prototypes for your website 
4. Why AI is important for rewriting and condensing data, information and website copy
5. How to apply AI in visual design and user research

Check out Tldraw in action!

If you have questions, ideas, or feedback to share, hit us up on Twitter. We’re @jonmacdonald and @ryangarrow.

• Apple Podcasts
• Google Podcasts
• Spotify
• Castbox
• Castro
• Overcast
• Pocket Casts
• Stitcher
• Subscribe via RSS

Episode Transcript:

Announcer:
You’re listening to Drive and Convert, a podcast about helping online brands to build a better e-commerce growth engine with Jon MacDonald and Ryan Garrow.

Ryan:
Jon, we have some exciting stuff today. Not only are we in the middle of holiday, which is chaotic regardless of where you fall in the e-comm ecosystem, but it’s fun and the data’s exciting. But you’ve put together some notes and things around AI in CRO and optimization of your conversion rates that I’m excited to get into, because I know probably more than most about what you do, and I probably wouldn’t have come up with an idea of how to bring in my ChatGPT system into CRO, but I bet people want to and you probably have to answer this question all the time.
I’m excited because I want to be able to use AI where I can, I guess, and where it actually makes me more efficient, rather than distracts me and sends me down rabbit holes, which maybe, I need to go down sometimes. But AI is a pretty buzzy term and it probably means a lot of different things to different people. So, when you hear AI, do you just cringe when people are talking to you about CRO? Or is it like, “Oh, this is going to be great and we can probably do a lot of good things”?

Jon:
I normally cringe because let’s be honest, when most people say AI, they’re probably talking about ChatGPT. That’s what the public knows about. And in fact, most of what we’re going to talk about today can be done with ChatGPT. So, I’m going to put that out there. Just know that’s probably the only tool you really need at this stage. And then, there’s a ton of SaaS apps out there, things like that, that say they’re an AI tool, when really, maybe they’re using the ChatGPT API, or they’re pulling from some other AI service, but they’re not natively AI, they’re not really AI. They maybe are AI enabled in some way. They’re thinking about user experience optimization, digital journey optimization. We do this with User Input, which is our user testing tool that we have.
You take your videos and we can provide sentiment analysis on those. We’re doing that through AI. That transcribes it, looks for sentiment, tells you, “There was a positive reaction at this time. There was a negative at this time.” But that doesn’t make it an AI tool. It is just using AI, it’s AI enabled to some degree, to make a function that would be much harder to do on our own. So, when I’m talking about AI, I really want people to understand how they could actually use AI to help them. It’s never going to replace them. Let me step that back. Maybe one day, but we’re not close. We’re nowhere close to replacing a human in understanding other human interactions and being able to optimize a digital purchase journey. It’s just unlikely to happen at this stage.

Ryan:
I feel like as a service-based business, if you don’t claim some level of AI, you’re automatically at a disadvantage because people aren’t going to trust that you’re advanced enough or efficient enough because you’re not using any type of AI.

Jon:
Yeah, that’s a good point. I’ve intentionally not brought up AI on Drive and Convert because it just has felt like a buzzword. It’s felt like would be just going after listeners and to try to get some buzz, instead of actually delivering value. And I kind of still feel that way with The Good. We use AI internally, but we don’t really talk about it. It does make us more efficient in a lot of ways, but we’re not out there promoting saying, “We’re an AI optimization firm.” Are we? No. There’s humans behind everything we do. To say anything else that I feel would be disingenuous.
So, yes, there’s a lot out there. I feel like it’s confusing for the general population. I think that we’re at the cusp where it’s a tool and you need to use the tool appropriately. And if you’re not learning about the tools right now, then yeah, you’re going to be behind in pretty short order. So, it might be something you’re going to want to at least study up on. So, really, today I’m scratching the surface with this, right? I’m not getting too deep into it, but I’m thinking about ways that if you’re optimizing your site, what are ways you could use, let’s just say ChatGPT for the most part, you could use AI, ChatGPT to help you perform better, think about maybe different things that you wouldn’t have come up with yourself or take the hive mind, if you will, right? That entire knowledge base, that ChatGPT has behind it and offer up some suggestions?

Ryan:
Perfect. Okay, so step one is going to be getting a ChatGPT account, I assume. For most of what you’re talking about, is the free version going to be applicable or do you need to jump in and start paying for one [inaudible 00:04:57]?

Jon:
Yeah, you could use the free version, but the benefits of paying right now are that you can open an instance of ChatGPT and start training it on particular things, so you won’t have to repeat yourself every single time. So, you open a new ChatGPT session and you say, “I want to talk about this ideal customer profile. And I want to talk about this type of product line. And I’d love for you to write a product description that’s geared towards that ICP.” Then, you wouldn’t have to, every time, come back. You’d just say, “Oh, here’s another product. Write the description targeted towards that same ICP.” If you have the free version, every time you leave the chat window, you’re going to have to come back and start over, and that’s where it gets a little more time-consuming and just annoying, but it’s a great place to start. Could you do this? Yeah, most of this you could do in the free version, I would think.

Ryan:
Got it. But for 20 bucks a month, if you are optimizing your own site, 250 bucks a year is not something that should break the bank if you’re really leaning in to leverage these tools, because it also can do other things for you. It doesn’t have to just be for your CRO.

Jon:
Well, I mean, one of the things it can do now if you pay is it can generate images. So, that is an area I’ll talk about today, and that you do need a paid account. But again, $20 to play around with it. If you find it’s not useful for you yet, then just stop paying for it, right? 20 bucks is-

Ryan:
Yeah, you don’t have to prepay for a year, so easy.

Jon:
Yeah, exactly. If 20 bucks is going to hurt you, then you probably should just stop listening now and go back to some of our earlier episodes about more foundational things you can do.

Ryan:
We’re assuming you’re advanced at episode 95 by now, and not just jumping in. Okay, so you’ve got your ChatGPT ready, you’ve paid your 20 bucks because you’ve gotten to that point where that’s okay in the business. And it’s on the business card where you’re getting rewards, so that’s good.

Jon:
There you go.

Ryan:
Where are you going to start with this? What’s step one when you’ve got your prompt open?

Jon:
I would say idea generation, and I started here because this is the highest level.

Ryan:
It’s also my favorite. You got a brainstorm buddy. If you’re working at home that’s the part I hate about working at home, by the way. It’s not as easy to be like, “I need to come up with some ideas that aren’t mine to juice my brain up.”

Jon:
I would argue it’s more efficient too, because knowing you, I bet you miss the, let’s call it water cooler chat. Like, “Hey, how was your weekend?” That’s great, but I would say this is more efficient than in-person brainstorming even because you’re able to say, “These are the type of things I’m thinking about,” and you start circling it and then it starts helping you come up with those ideas. So, in this case for idea generation, I would really just prompt ChatGPT with constraints or themes to focus on. So, give it some barriers, and then ask it to find alternative suggestions or blind spots, those themes or constraints.

Ryan:
So, I’ve got my Shopify site and I’m like, “Am I going to prompt it with this URL? I want to improve conversion rates or I want to improve conversion rates on PDP pages for e-comm.”

Jon:
Right. So, the more specific you get, the better information you’re going to get. And you really can’t say, “Hey, I want to improve conversion.” You can, but what you’re going to get back is probably not amazing. But what you could say is, “Here’s my product detail page for this product and I want to tailor it more for this particular profile that we’re targeting in an ad campaign we’re running. And here’s the visuals for the ad. So, here’s the ad. Now, can you help me rewrite the copy on this page to be more tailored towards that audience?” And so, you give it some constraints. “I want you to write tailored copy. I want you to do it in a way that is in alignment with the entire digital journey.” So, starting with that ad that they saw and they click through, so you’re giving it constraints.
So, just start prompting it with more constraints and you’re going to get better stuff back. So, really, here what we’re looking for though is to find alternative suggestions. So, we’re looking for it to help you brainstorm and not necessarily do the work for you. So, in that sense, what you could say is, “Here’s my PDP. Here’s the ad campaign. How could I improve this copy?” It’s really good at copy, that’s why I keep going back to that. That’s probably it’s best is text. But we’ll talk later, there are ways to even use it for visuals quite a bit. So, one other thing you could do here is have it share examples. These are concepts maybe you’re interested in learning more about. So, you could say, “Hey, here’s my PDP, I’m tailoring. I want your help tailoring it for this audience. What are the competitors I should be thinking about?” Or, “What are other PDPs targeting this same audience?” So, then, you start kind of digging in and start asking it some more questions. You can get more examples that you can learn from.

Ryan:
So, using it in that sense, almost like a Google like, “Hey, who else seems to be targeting this product with the same persona?”

Jon:
Well, that right there, why do you think Microsoft owns a big portion of OpenAI? That’s exactly what they have it for, is to, eventually, try to overtake Google in search engine because they want to use ChatGPT or AI, essentially, to improve their search engine.

Ryan:
Got it. Okay. So, I’ve got some ideas of the PDP page and some of the copy. By the way, how many times do you think people have to rewrite that copy within ChatGPT to get something usable? Most often, I found the first one is like, sounds good, but that’s definitely not what-

Jon:
Again, the more constraints you give it, the better it’s going to do.

Ryan:
So, training your own, which is another reason for the $20 ChatGPT.

Jon:
Right.

Ryan:
It gets better and better knowing what you’re looking for and what your tone of voice-

Jon:
Yeah, exactly. Because then you could say, “Here’s all of my products.” Upload a CSV file of all your products and all the product descriptions, and it will learn how you write. Now, is it going to be perfect? Unlikely. Again, it’s still early. ChatGPT’s been out a year, I think. If you asked your toddler to write, probably not going to be great, even if you gave it a bunch of examples. You’re going to need some help. So, that’s really where it’s at, but it’s getting way, way better.

Ryan:
Okay. So, it’s got ideas. And then, you put a note in. The next one is what you refer to as prototyping. I mean, I understand what a prototype is, but what does that mean as you process through AI to get to the final product?

Jon:
Well, what I’m talking about here is let’s just say you have a new PDP and you’re trying to put all the content together that goes into it. So, you maybe want to run user testing on that before you launch it or build it into an actual template. So, the idea behind the prototyping stage here is what can you do before you have actual code written? Which ChatGPT can actually help you write some code. But the reality here is I’m saying, what’s the least amount of effort you could put in to get a high-impact feedback? And a prototype is really going to help you do that. So, you could have it write realistic copy, and so you feed in a prototype, you take a picture of your prototype or upload an image that has lorem ipsum, and let it write the copy there instead. It can replace it and do that.
I’ve mentioned visuals a few times. You could have it boost the realism of all your mock-up visuals. So, ChatGPT, if you pay for it, can now generate images, including ones that will match your brand tone and style. So, again, you got to provide enough constraints, maybe give it some examples, but it can certainly do that. What’s awesome is I saw this recently, you can actually upload a sketch of a wireframe. So, take a piece of paper, pen, draw it out, and it will turn it into a higher-fidelity wireframe or mock-up that you can then utilize for testing. And it can even do some visual design. So, you say, “Here’s my brand style guide and here’s a sketch of this page. Can you turn that into a wireframe or can you turn it into an actual page that I can use for user testing, a visual?” And it will do that.
And you can also take that even a step further and create actual functional prototypes. There’s a tool out there called tldraw, T-L-D-R-A-W. It’s a AI tool that allows you to turn your sketches into functional applications without any programming experience.

Ryan:
Holy smokes.

Jon:
So, that’s where I would go is you create a sketch of your application, it analyzes it, and then it uses AI to generate functional code for that. And you don’t have to know anything. So, we’ll put in the show notes. There’s a whole Instagram video that the company did, and we’ll include that in the show notes. But is it perfect? No. But is it way better? Say, I don’t know how to code at all and I want to start doing some user testing on a mock-up. This is amazing. This is a great way to do a clickable wireframe prototype that could be the entire customer journey of a website or a SaaS application I want do, or a mobile app. And now, I want to take it to market and do UX testing on it. And it would certify for that.

Ryan:
Dang. And I also like that their name starts with TLDR, which is appropriate. Like too long, didn’t read. That’s pretty much what I do a lot of. But coding would generally indicate you may not be on a Shopify platform. In a template, you’re somewhat limited.

Jon:
Well, it can write to Shopify template specs without a problem. I have clearly seen people who fed it a template file and said, “I need to make these changes. Can you make them?”

Ryan:
Perfect.

Jon:
It’s done that and it does it pretty well. Is it perfect? Not always, but it’s functional. It’ll pass. Maybe you have a developer look at it if needed, but you’re spending 15 minutes of their time looking over that template file versus two hours of them writing it. That’s a heck of a productivity boost.

Ryan:
Oh, yeah. Yeah. So, I would challenge a lot of Shopify template listeners out there that you can modify Shopify templates because they were modified from Shopify code to make it look like that. So, there is code in there. You got to be careful when you slow it down though. So, that’s where a developer is going to come into play. If you get this really cool, beautiful visual PDP, but it loads in 15 seconds, it doesn’t matter how good it could convert because it’s just not going to.

Jon:
Yeah, and look, I mean, is ChatGPT going to have your style of visual design that you want at this high level? Absolutely not. It’s not. But would it do a prototype that then you could hand to a developer and say, “Here’s what I need. Make sure the code’s cleaned up and ready to go”? Yeah, you could totally do that.

Ryan:
Huge cost savings there. You paid 20 bucks and some of your time to get it [inaudible 00:15:31].

Jon:
20 bucks a month. So, imagine how many times you could get a page out of it over a course of a month, right?

Announcer:
You’re listening to Drive and Convert, a podcast focused on e-commerce growth. Your hosts are Jon MacDonald, founder of The Good, a conversion rate optimization agency that works with e-commerce brands to help convert more of their visitors into buyers. And Ryan Garrow of Logical Position, the digital marketing agency offering pay-per-click management, search engine optimization, and website design services to brands of all sizes. If you find this podcast helpful, please help us out by leaving a review on Apple Podcasts and sharing it with a friend or colleague. Thank you.

Ryan:
Okay. And then, copywriting is what it’s known for. And so, outside of prototype or revamping a PDP page and giving it some ideas on there, are there pieces of the site that you find that ChatGPT is going to be better at versus… Truth be told, I know brand owners that have done it for their mission statements, and it’s not always for that, but-

Jon:
Just what you want a robot writing.

Ryan:
… where do you guide… Keep them first.

Jon:
Yeah. Well, I think in the simplest form, I like to talk about how brands often have these walls of text. They’re just paragraphs of text they put up on the site and nobody’s reading it. You do all the heat maps, eye tracking studies.

Ryan:
[inaudible 00:16:56].

Jon:
People read the first couple words and they’re like, “Oh yeah, I’m not reading all this,” and they just skip right through it. So, what we’ve actually done is use it to shorten the text while keeping the main points. So, you say, “Here’s a paragraph of text,” might be 100 words, “shorten this to 20 words, keep the main points.” And it usually does a pretty good job of that. Or, “Give me bullet points out of this.” That works really well too. So, you have a couple of good options there. You can also use it to, again, make copy persona specific. Mentioned this a few times. So, you describe those personas, ask it to rewrite the copy to be directed at that persona. That works really, really well.
In a slightly different approach, if you’re trying to gain support internally, this is a really cool trick. You can use it to rewrite your reports for clarity. So, you’re saying, “Hey, I have something that’s really technical. It has a lot of data about why we did something. I need you to rewrite this to target a VP level who doesn’t understand data science.” And it will do it. It will help you rewrite that copy. And I think this is a cheat code for service providers or consultants, but I think that it’s really great for gaining internal support. So, if you’re an e-comm manager, VP of e-comm, and you’re trying to present to a C-level who does not truly understand what’s being said to them and you really want them to get the results out of it, then I think this works extremely well for that.

Ryan:
Got it. So, it’s almost like condensing, here’s why we’re doing CRO. Here’s some of the results. It’s going to tell your story better than probably you would on the first pass through.

Jon:
Yeah. We’ve had clients who take our reports and run them through and say, “Give me the key points of this slide for somebody who does not understand optimization.” And I don’t know the exact prompt they used, but they told me they did this to present it to their executive team and that it worked extremely well. Because they were like, “I understand this information, but how do I present this to the executive team? And I don’t want to rewrite all these slides all the time.” So, they decided, “Oh, we’ll just paste it in and see what happens.” And it worked really well for them.

Ryan:
Dang. Yeah, telling that story ends up becoming so important from the marketing team to the exec team to justify and explain things. Like, yes, you do need CRO and you need it all the time, not just two months here, two months there. Anyway, I won’t do your sales for you.

Jon:
No, by all means, please continue.

Ryan:
Your next point here is my favorite part of, honestly, on AI, what people use AI tools for. But it’s visually, what can it do? The stuff it can do with images-

Jon:
It’s getting so much better.

Ryan:
… boggles my mind.

Jon:
Have you seen the latest trend where somebody took… He said, “Give me a picture of a happy kitten.” And then they said, “Make it happier.” And then, “Make it more happy.”

Ryan:
Oh, yeah. People do this all over LinkedIn. “If you like this, I’ll make it work harder,” or, “I’ll make the beaver-“

Jon:
Yeah. And so, there’s this whole meme of people now taking things and making them more. The first one that somebody did it was, “Make this kitten happier.” And they were just playing around with it, and it turned out that ChatGPT was coming back with these descriptions and these images that were hilarious. And the kitten kept getting happier to the point where the last one, it looks like something out of space. And it’s like this is the ultimate embodiment of happiness. There’s nothing in the world that is possibly happier than this. It is beyond the realm of physical, and it starts getting into this really theoretical stuff that was hilarious in the end, but it all started from a picture of a kitten that they were like, “Can you just make the kitten happier?” And then it’s like, “Sure, here’s a happier kitten.” And then, they’re like, “I wonder, could you do happier?” And it just keeps going.

Ryan:
That’s crazy.

Jon:
So, basically, the point here is there’s a lot of ways to use this for visual design. I think the first example I really like is persona illustrations. It really makes something that’s theoretical, concrete. What I mean by that is you prompt it with a description of your persona and then ask it to share a visual representation of it. Again, if you’re selling through to executives, this is great. You’re giving them a visual image of the persona you’re trying to reach and talk to, so that could be helpful.
And journey map. So, I really like the idea of asking it to feed it something, a site, et cetera, and then say, “Describe all the steps in a journey from research, to buying, to conversion.” And it will tell you what the steps are along the way. And you could even feed it that and say, “Create me a journey map, a visual representation of all these steps,” and it will do that for you.
So, other UX ways to use this, we’ve talked about drafting designs of pages. I think that’s a really valuable one. It works surprisingly well. Comes with risk as, again, not being as specific to your brand or your persona, but again, that’s all in the constraints you give it. It depends how much time you want to spend feeding the information and training it, and I’ll put that in air quotes. But you can ask it to provide a visual design for specific pages of your site and feed it information to occlude on that. “I need a product category page. Here’s all the images and descriptions and titles of the products. Go.” And it will create one. Does pretty good. It’ll get you part of the way there to at least something to start with.
And the last is kind of what we were talking about with the kitten, right? It’s AI-driven images and brand assets. So, I’ve seen a lot of smaller brands do this. I don’t see as many large brands do this because it’s not a resource constraint for them, but if you’re a brand that’s starting out on e-comm and you need some visual assets and don’t really have a design budget, maybe you want to use them in ad campaigns, et cetera. If you can get good at describing what you want, they work really, really well in a pinch. So, you can easily get some graphics together that way.

Ryan:
Have you ever used it to increase how large or the pixelation of an image? I see a lot of that where you get a really small image to start with. You’re like, “I like this, but it can’t be used in a vector.”

Jon:
Right. Yeah, I-

Ryan:
I assume it can do that.

Jon:
I haven’t tried it, but I imagine it could. I don’t see why it couldn’t. It can generate new images, so you could just say, “Hey, take this image and increase the resolution of it.” It would probably be very good at that. I don’t see why it wouldn’t. Now, it’s going to guess a little bit. Is it going to be the actual data that was there when it was a larger image? No. But I would imagine it’d get close enough. Somebody who hadn’t seen the image before wouldn’t know.

Ryan:
[inaudible 00:23:33] your vector and it’s too small. Use ChatGPT, “Give me a big image.”

Jon:
Yeah, there you go. Yeah, user research is really the last area that I would use this for in UX. I’ve mentioned it a few times already today, but I love to use it to draft interview questions. So, again, this is dependent on the prompt used. All of this is. I mean, that’s a huge preface for all of these. It’s like crap in, crap out. So, you really got to work with it a little bit. But you have to make it detailed enough to have it come up with great interview questions, but you can ask it to draft those for user testing. It’s not awesome, but it will do it, and I think it will only get better if you spend more time training it.
You can have a draft test tasks for user testing. So, not just the questions, but also, what you want people to do along the way. So. You’re saying, “Hey, I want you to add something to the cart. Take the next step along the user journey.” So, see how these are kind of building upon each other? You have it map out that digital journey. Then you say, “Hey, can you have some tasks that a user would do in user testing along that user journey?” Then, “What questions would you ask as they’re completing those tasks?” So, now you’re starting to get a whole user testing program together. You have that journey you want someone to take. You’re giving them tasks to get each of that step together, and then you’re having questions that you can ask them at each step as well to get good user feedback.
We talked a little bit about sentiment analysis. Then, you take all of the response videos and you feed them in and say, “Tell me what the overall sentiment is of this,” or, “What’s the high point and low point of the sentiment?” And then, you could go in and instead of having to watch 50 videos front to back, you can now just go to those time codes that it gives you and says, “Yeah, at minute 1:05, there was a positive sentiment, but at minute 3:05, it was negative.” Wow, you probably want to watch that two minutes. What the heck happened in that two minutes that somebody went from happy to being upset?
You could tell it to extract themes. So, user testing scripts and videos. You could say, “What’s the theme coming out of these?” And it will tell you, “Oh, well, a lot of people talked about the product detail page,” or, “A lot of people talked about the add to cart button.” Then, you kind of have some idea of where to go optimize first. And this is kind of along those lines, but categorizing issues from the interviews. So, use the transcripts and have it pool all the themes on ideas of what should be optimized. It’s really good at this type of work. Really good.
Give it lists of things and tell it to pull themes out and comb through that data. ChatGPT can do that in seconds and it saves you so much time. So, that’s a really good way to do it. My biggest concern with all of this is that folks will use it and aren’t really learning about user testing and user experience. Instead, they’re relying on ChatGPT to come up with all the answers. I think that’s the unfortunate part of doing this is you get really good at ChatGPT and not very good at user testing, because you just rely on it to give you the answers.

Ryan:
It’s user testing for a reason.

Jon:
Right, it’s not ChatGPT testing.

Ryan:
It’s actual users. Yeah, I mean it’s limited by the publicly-available information, and I’m guessing all of your user tests over the past 15 years are not publicly available to ChatGPT necessarily.

Jon:
No, but one thing that I’m not ready to fully announce yet, but I will say that we are coming up with… One of the things that OpenAI allows you to do now, they just announced, was create your own ChatGPT. And so, we are going to be doing one that feeds in our 15 years of content that we’ve put together. So, I’ve been on so many podcasts, we’ve written so many things, so many webinars, case studies. Every week we write at least 2,000 words. I have two books plus that I’ve been part of and written. All of that content, all the videos I’ve recorded, et cetera, can go into this. And then, some people can ask it questions, and it’s trained on all of that data already, so they don’t have to train it. They get the 15 years of The Good’s brain as a basis point.

Ryan:
And you’re just going to give it to everybody for free, right? It’s just going to be wonderful.

Jon:
We’ll see about that. I want to try it out first, but it’s in development.

Ryan:
You’ll be able access Jon’s brain through a ChatGPT. That’s pretty crazy.

Jon:
Yeah. And so, that’s really the idea. That was part of the basis for coming up with this. If I did that tool, how would people use it? Here are some ideas of how people might use that type of tool. And all the stuff I talked about today would be things they would use the JonGPT for or TheGoodGPT, whatever you want to call it.

Ryan:
Well, you can get some bracelets like WWJD, what would Jon do?

Jon:
There you go. Well, unfortunately that one’s probably taken.

Ryan:
That one might be taken WWJonD. What would Jon do? So, with all these things, I mean, you’ve told us how essentially to use ChatGPT. So, if I pay 20 bucks, I don’t need to work with any consultants, right? I mean, it’s done for 20 bucks.

Jon:
Yeah, I wish you luck with that. Here’s the thing, it has come a long way, but let’s keep in mind it is a year old. It is moving very quickly. Again, is it going to replace a human? Look, if I had to invest in some type of really long-term mutual fund around that idea, I would say yes. But I think probably not before I retire. I think that it’s very likely it will be able to do a lot of things, but I also think that there’s nuance in a lot of optimization that will be really hard for it to just completely take a human’s job. Now, segments of that for sure. Data analysis, all the stuff we talked about today, yeah, for sure. If that’s all you do, then you probably need to start broadening your skillset to be more valuable. That is how I would look at it.

Ryan:
The e-commerce industry as a whole has been using AI for a long time. It’s not new. I’ll say it’s like new direct-to-consumer is what AI is. It’s been B2B for a while. Google’s been using it inside their algorithms for a very long time. You’ve been using it on the backend within your systems for a long time. We use it at LP. It’s not new inside of our businesses.

Jon:
Right. What’s new is you’re giving the consumers of the work product a way to engage with that AI, so you’re moving it forward a little more, right? That’s what’s really happened. You’re right. Google Analytics, a lot of that was AI based. What do I mean by that? Well, in the sense that if you weren’t paying for Analytics 360, you weren’t getting 100% accurate data. You were getting data that was partly accurate, and then partly extrapolated from that by an algorithm, right? An algorithm can be artificial intelligence in that degree. So, now I can ask ChatGPT to pull in my analytics data that I feed it through a export and I want it to answer questions for me. That’s moving it forward another step towards the consumer, and that’s when it’s becoming more and more valuable.

Ryan:
Yeah, I think some of the minutia of extracting of data and analyzing that is what’s really becoming the biggest use case of AI that I’ve seen. Being able to look at large sets of data, and instead of doing that manually, it’s like, “Okay, I want you to condense this and focus where I need to spend my analysis. You see all these conversion rates year over year on all these landing pages. Where might my problems be?” Okay, well, that’s very simple for ChatGPT to look at all of it and be like, “I’d focus here.” Great. Now, we get the strategists and human brainpower to go look at those areas.

Jon:
Wonderful. Well-

Ryan:
I like it.

Jon:
… there’s hopefully enough in here for people to get started. Again, most of it, I tried to keep the ChatGPT because I think that’s the most accessible option. I’m sure there’s way more we could do here, but this is scratching the surface and hopefully we’ll start people down the right path.

Ryan:
Yeah. Well, congratulations on getting the busiest term in our podcast headline yet.

Jon:
Win.

Ryan:
So, hey, we may be off on a hockey stick growth. We’ll see.

Jon:
Thank you, Ry.

Ryan:
Thanks, Jon.

Announcer:
Thanks for listening to Drive and Convert with Jon MacDonald and Ryan Garrow. To keep up to date with new episodes, you can subscribe at driveandconvert.com.