Transcription
[Music]
Hello and welcome back to another episode of the Collective Clicks podcast. This is your host, Brandon Bateman, and today I'm joined by Garrett Cregan. We're going to talk about our top four AI tools used in digital marketing for real estate investors right now, starting with number four and going all the way to number one. I think all of these have massive implications for your marketing and business, and I'm excited to dive in. How are you doing today, Garrett?
Doing great. How are you?
Hey, fantastic! Thank you. Super excited for today's episode. We're going to be talking all about AI, which is something probably nobody's heard about, right? It's super new.
Low-key. Yeah, very underrated.
Recording this as of December 2022.
Right, exactly. Just way ahead of things. Um, now I'm—I'm excited to talk about this. This is not December 2022; this is August 2023, and it's been crazy seeing how the world of AI tools has changed over the past little while. So we're super excited. Garrett, a little context for everyone: he leads a lot of our internal marketing as well as our product, and he's been leading the charge of adopting new technologies and helping make them work for our clients' marketing and what we do ourselves. That's why I'm excited to have you here because I know you've been messing around with a lot of these different tools and have some pretty cool insights to share. Anything you'd add?
No, I just—what I don't want this podcast to be is just rote: "Here's how we use AI." This will be tailored to investors and this industry—what we're using day-to-day for our clients and our marketing. These aren't things we've seen on TikTok or YouTube; these are things we use daily that bring real value.
Awesome. So, without further ado, we have four specific tools we want to talk about. They're ordered from least cool to most cool, or most useful—whatever we want to call it. We also have an honorable mention. I think it's fair to start with our honorable mention: Google's ad platform AI. Any comments on that?
Yeah. When people talk about AI right now, the biggest focus is language models. That's what people see as AI right now because that's the most prominent thing in the ether—those text-based AI models. But what tends to go under the radar is the AI Google uses has come a long way from being a very manual platform to a very data-driven and data-fueled experience where the people with the best data win, not the people with the best keywords and bid adjustments. Who feeds the best data wins, and that's all fueled by AI.
Yeah, it's crazy. Just like you, I've seen this trend in language models becoming so popular, and I think it's because they've improved the most recently. They were so bad, and then, as of less than a year ago, they suddenly got really good in comparison. But AI's been better at analyzing numbers than people for a long time, right? No human looks at a dataset and makes a decision based on 10,000 different factors, but AI can. That's its strength—seeing correlations across everything at once. Those models are super popular, and sometimes they work really well, and sometimes they work pretty poorly. I know that one thing we talked about in other episodes is that AI models for bidding are really powerful right now, and if you feed them a ton of good-quality data, you can really improve your results. And it matters enough that if you've listened to our podcast for a long time, you're probably rolling your eyes right now because we've talked about that so much. If you're new to the podcast, check out some past episodes all about bidding strategies and bids. But Google's also taken some kind of sour moves towards AI, or at least sour for this industry. You want to talk about those kinds of things, Garrett?
Yeah, I'm thinking like Performance Max campaigns and other things like that. So they've—like you were saying—Google is in the business of making their platform as accessible as possible to the largest audience, and it's trying to do that by taking the levers out of the user's hands and putting them into the hands of machines. And that's great if it does what it claims it can do. So with Google, they've launched—I guess they've had this for what, two years now, probably?
Yeah, well, they've had iterations of this for a long time. Like once upon a time you had manual search campaigns, and then…
You have Express, and then did Smart campaigns come after Express?
Smart campaigns were a subset of Express.
Okay. Yeah, and then—and then you got PMax and stuff. So it's—I think they've been wanting to do this for a long time. The real trouble seems to be making it actually work.
Yeah. And so the way it works is we feed it some audiences and some images and some copy, and then it runs your ads on the entire suite of Google Ads placements—on YouTube, on display, on search, shopping if you're in e-commerce, for instance. Its whole goal is to get you the best cost per action that you're tracking. And so that works great if it's being given that target, but it's not always able to get the right target, and so it gets a lot of spam at times because, like we were saying earlier, it's as good as the data it gets. But that's definitely a drastically improved aspect of Google's AI; it's just not quite where they claim it is, I would say.
Well, it's also really industry-specific. This is why it's so important that you speak to people who are in your industry actively doing this kind of marketing because what you hear about Google overall doesn't always apply to the real estate industry. From what I've seen, this real estate investment industry kind of lags a little bit behind the rest. I know several people—and I'm sure you know several people as well—doing fantastic with Performance Max campaigns in many different industries. It's actually known as Google's first actual successful venture into that, but still, in this industry, it's just not working. And I think it has something to do with how—this is just such a small niche. In the eyes of Google, this is tiny, and Google's AI just doesn't truly understand how you as a real estate investor are different from a real estate agent, for example, or any other real estate-related thing. They just kind of lump them all together, so it's not quite as good at distinguishing the type of targeting you're trying to go after and which platforms are working best. It's also a really data-sparse industry, which makes it really hard for those campaigns to perform well.
Yeah, and I think that having the added intelligence of AI is very powerful in the right hands and with the right inputs. So I wouldn't say that it's an unintelligent platform. I would say that, given the right nurture, it can drive much better results than a purely manual campaign ever could. It just needs a lot of attention and skill to help it break through that barrier, if that makes sense.
It does make sense. It's kind of interesting because it's counter to the general idea of AI, which is AI can do what you were doing 90% as well with 5% of the work, right? But especially with a lot of these things we're going to talk about today, you can actually exceed what you're able to do otherwise with AI. It's a point where AI, I think, is enabling in a lot of ways, and if you know how to use it properly, then it makes potentially an even better end product, granted that usually doesn't come with a lot less work, although sometimes it can be significant.
So, unless you have anything else to comment on with ad platforms, let's talk about our four biggest tools that we're using right now. Number one is Unbounce. You want to comment on that a little bit? Obviously, Unbounce isn't an AI platform, but they've been starting to integrate AI, and in some ways it's been pretty successful.
Yeah, so in Unbounce, they have two AI-driven features. The first is the ability to write copy for your landing page with AI and have it fit your brand and page structure. It's cool if you're building a brand-new landing page and just need 80% of the copy written; it does an awesome job and saves tons of time. But where we've seen the most benefit is with their smart traffic feature. How that works is you set a goal in Unbounce for what counts as a conversion—a link click, a form submission—those kinds of actions. It can also track post-click actions with a script, which is a more advanced tactic, but you can do it. So how this feature works is you can make five, six, or ten different variants of your page in Unbounce, and it dynamically learns which pages work best for which audience. It's not just A/B testing where it sees which variant does better in general; it looks at the source of traffic, the device, who the person is, and learns which page works better with each audience segment. This saves tons of time and helps you reach a winner a lot faster than doing several different tests in a row. It all happens at once and is based on the audience, not just the page. It's very powerful.
Just in case somebody's lost, if you're not familiar with Unbounce, it's basically landing-page-specific software. It makes pretty fast-loading, easy-to-design, easy-to-A/B-test landing pages. There are capabilities you wouldn't have natively on your site that are specifically for landing pages. That's why we use this, and I don't even know how many landing pages we have in Unbounce right now. It's probably more than 1500 active pages across the internet. We take our grandfathered subscription plan and push it to the max, so we use this tool a ton.
I want to contrast this clearly with normal A/B testing, which some people are familiar with. In normal A/B testing, you take A and you take B, and you put them side-by-side. You see how one does versus the other; you do a statistical test to say A works better than B or B works better than A after you get some traffic through the page. Then you know that one of them is better or not. Here's the downside of a statistical test: unless you're completely on it all the time, there's often a time period where you kind of know that one variant's better than the other, but you're still running the one that's not good. The other component is that it basically assumes that everything in A and everything in B are the same all the time, when, like Garrett mentioned, there are a lot of different things here—people coming on different devices; the page looks completely different on a desktop versus a mobile device; the type of traffic and the source they came from; if they're expecting one thing versus another thing. So if you use these predictive models, they can take all of that data into account and dynamically deliver not just the best page, but the best page for that specific person. It's a different level of optimization. It moves into a realm of analytics called predictive analytics rather than inferential analytics. In inferential analytics, the goal is to test A versus B and understand if A or B works better. In predictive, we don't really care which one works better; we just want to dynamically predict which one's going to work better in every individual scenario to have the best end result, which does match a lot of what we're trying to accomplish with marketing. It's kind of how ad testing is working within most ad platforms now, too. In Google Ads, if you're going to have multiple responsive ads in that group, or even multiple headlines in a responsive ad, or in Facebook if you're going to use a dynamic ad or have several ads within a single ad set, this is kind of like—A/B testing is pretty much dead when it comes to digital marketing. There's so little of it happening compared to this more dynamic, predictive type of testing.
Definitely. Well, that's awesome. So Unbounce—cool tool. Let's get to an even cooler one. Let's talk about the next one. Anyword. Okay.
Yeah, so Anyword is a copywriting AI platform, like the ones we know and love—Hemingway, Jasper—all those tools that help you write at scale. But what's cool about Anyword is it scores the copy you write based on how it expects it to perform to your target audience. You can give it an audience—age, interests, income—and then it can score your copy from zero to 100 based on how likely it thinks your audience is to engage with that copy. That's the first thing it does that's powerful. And then the second thing it does is it allows you to integrate it with your ad accounts—Google and Facebook—and then it can pull in that data to see which of your ads have the best performance, and then it uses that to help you make even better ad copy based on your target audience and what's worked well historically in your account. The more you test, the better it is. The audience piece of it is very powerful because most other AI writing tools don't store that kind of information; it has to be in your prompt every time: "Write this for homeowners who are wanting a cash offer," and then do it. But this will have that saved all the time, so it's always analyzing it with that in mind. And we've been using this for our clients and for our own marketing, and it writes better copy than I ever could on my own in seconds. Even though it's not perfect, it saves hours of planning, ideating, and edits, and gets me almost there in minutes.
Well, the crazy thing is it can—it can also just—it's not just saving you time, but it's considering more factors than you could ever consider in your brain, right? As humans, we're good at comparing two, three, or four things, but all the performance over the past two years of the 80-something ads that we've tested, we can't keep all those insights in our brain and use that to make good stuff. Which is so cool to me because you look at tools like ChatGPT, and they're so good at just using what's on the internet to optimize things, but they don't have any first-party data; they don't know what's actually worked for you and what hasn't. I think that feedback loop is so crazy powerful. Does this work with Facebook Ads and Google Ads, or just one of them?
Both. It also pulls in your copy on your website and helps you fix that copy based on what you're seeing work in your ads as well. So I suggest just one other area that it can impact is your Google Ads, your Facebook Ads, and your site performance—a feedback loop that's not very strong usually. I think most companies just aren't adjusting their website copy based on what they're seeing with ad performance. So that's awesome. Any other comments on Anyword, or should we move on?
Let's go. All right, Opus.
Okay, so—there's a big—so this—last week I went to a mastermind, and I heard a lot of investors talking about the benefits of posting on social media, in particular as a way to secure capital for purchases. And I know that it takes a ton of time to make that content, edit it, get it ready for social, and it's also an expensive skill in most cases. I hear a lot of investors are having this done by VAs, which, if you find when it works, it's fine. But Opus is cool because you can—so how it works is you upload an image or a video, and it listens to the whole transcript; it finds the clips it thinks have the best chance of going viral and gives you between 10 and 15 clips from that longer video that you made—be it a podcast or some kind of longer video. It has them organized by their odds of going viral and also says why it thinks it'll go viral and what could be changed to give it a better chance. It adds captions, adds emojis, and lets you edit the font, font size, and color to match your branding. It's free, and it works in about 15 minutes. It pulls all those clips and auto-focuses on faces, which is crazy. If you're walking around and moving, it'll crop to always keep your face in the middle. So it's crazy powerful, and it's free. I feel like it's a really underrated tool. I'm seeing it mentioned more and more online, but it's still free for now, so I would jump on before they roll out their paid model, which I think will be happening pretty soon. But even paid, it's still a huge value.
Yeah, that's crazy. So it makes it as easy for your average real estate investor as just going on a few podcasts, feeding that into the tool, and then posting the clips on your social media. You just need the origin of the content, but all that in-between part gets taken care of, which is insane. That's so much easier because we've spent at least a thousand—probably multiple thousands—per month just on that one piece, and it can literally be replaced by a completely free tool.
Yeah, it's crazy; it's insane. Okay, so to jump into the last one—this is the number-one, most interesting AI tool when it comes to digital marketing for real estate right now, with all that hype, Garrett, why don't you talk about ChatGPT's code analyzer?
Yeah, so this is a recent addition to the paid version of ChatGPT where it lets you upload sheets or CSVs of data, and it can analyze—so here's what it can do. In their own marketing, OpenAI—their parent company—has said that their goal is to have Code Interpreter be essentially a junior data analyst for your team for, what is it, 15 bucks a month? And so how it works is it can analyze a massive dataset; it can clean it for you; it can turn text fields into number fields; it can identify trends and correlations; and it's wild. As an example of how we used it, we have a huge dataset of all our clients—where they work, their market size, their number of agents, the exit tactics they use, all kinds of stuff—and then their lifetime value for us. We had it find which attributes of our clients have the strongest correlation to LTV to help us maximize that and build out our ideal client profile from that dataset. And it did it in seconds and gave me an R coefficient—basically how strong the ties are between that attribute and the target of maximum customer value. And we did that in an afternoon. So for an investor, if you can get a dataset of your properties—where they are, the age, the number of bedrooms, number of bathrooms, the bedroom-to-bathroom ratio, all kinds of attributes—and then their deal spread, you could run this type of analysis to find which home attributes are tied to your biggest spreads. I mean, that kind of data analysis is unheard of and so powerful. I don't think we quite understand yet how much that can impact your business strategy.
Yeah, I think the—I think the real biggest benefit is that to do that kind of stuff you used to need to be a data analyst, right? And I think you still need some of those skills, because tools are going to mislead you from time to time, and you have to understand the difference between correlation and causation and stuff like that. But just the sheer volume of insights you can drive and then know that you can dive deeper into those things is insane, right? It would have taken you six months to gather all this information and understand all the different correlations; you can get that done in an afternoon. And then from there, you can spend a lot of your time going deeper, more refined, understanding it on a better level. It's insane what you can do, and I don't know—I envision a world where spreadsheets—you almost won't have to use them at some point because this can be such a powerful method of making that data analyzed and getting insights from it. It's insane because the numbers—the statistical capabilities—have been there in AI for a long time, but the ability to bridge that gap between that and driving insights has never really been there in AI until just recently.
Yeah, and I think that it's just beginning to scratch the surface of what's possible, and when you can take that data insight that it finds and then tie that to truly actionable business tactical changes, that's when it really starts to be impactful. It's not just cool knowing that this attribute of a client, for us, means a better client, but how that can change how we market, how we sell, our messaging—that's where it gets really, really valuable. And I think that's where I think AI in general is going to change things. It's cool, right? And it's buzzy, but I think where there's real value is when its outputs are usable in your business day-to-day operations.
Yeah, well, that's what I've appreciated about some of the things you've shared. It's not just, "Look how cool this is," but every single one of these options has the potential to grow your business and change your strategy, and I think that's way cooler than just being cool, right? Because it's like—I think AI, for a little bit, for a lot of people, is just kind of a party trick almost—like, "Oh, that's really cool; I'm amazed that a computer can do that," but what are you going to do about it? You know, where are you going to save costs in your business because of this? Where are you going to shift your strategy because of this? Those things matter. It's on. And I've appreciated all of your insights. That's it for this week's episode of the Collective Clicks podcast. For everybody listening, I will see you next week. Thank you.